Camera having an image shift detecting function

BACKGROUND OF THE INVENTION

1. Field of the Invention

The present invention relates to a video camera and an electronic still camera, and more particularly, to a camera having a function to detect a relative shift between the camera body and the subject image caused due to camera shake in photographing.

2. Description of the Prior Art

Typically, a solid-state image sensing device such as a charge coupled device (CCD) is used in a camera of the above-mentioned type. In the present specification, hereinafter, a relative movement between a camera body and a subject is referred to “image shift”. The precision of image shift detection by the solid-state image sensing device depends on the luminance distribution of the subject. Conversely, the precision of image shift detection can be estimated to some degree by analyzing the luminance distribution of the subject image formed on the light receiving surface of the solid-state image sensing device. Therefore, by detecting an image shift by using only a partial area of the solid-state image sensing device which is estimated to have a sufficiently high image shift detection precision, a high detection precision is obtained with a small number of data and a small amount of calculation. What is important in this case is to correctly estimate the detection precision.

As conventional technical means for compensating for camera shake in photographing with a camera of the above-mentioned type, for example, the following means are known: a means as disclosed in Japanese Laid-open Patent Application S63-166369 in which the image plane is divided into a plurality of blocks and the image shift amount is obtained for each block to determine the vector of the entire image plane (first prior art); a means as disclosed in Japanese Published Patent Application H3-65509 in which a plurality of moving subject detecting devices are provided to detect a moving subject by calculating the image shift amount for each moving subject detecting device and adding the calculated shift amounts to average them (second prior art); a means as disclosed in U.S. Pat. No. 5,386,264 in which, in an apparatus to detect an image shift based on a video signal, the image shift detection is performed only when a high-frequency component of the video signal or a sharpness exceeds a predetermined value and is not performed when the high-frequency component or the sharpness is the predetermined value or below (third prior art); and a means as disclosed in U.S. Pat. No. 5,218,442 in which the image plane is divided into a plurality of blocks and a plurality of blocks are selected from thereamong in the order of height of the contrast so that the image shift detection is performed only with respect to the selected blocks (fourth prior art).

However, in the first prior art, since the calculation of the image shift amount is performed for all the blocks and not for only a part of the blocks, the amount of calculation inevitably increases, so that a large capacity memory is required. In the second prior art, since the calculation is performed for all the moving subject detecting devices, the same problem as that of the first prior art arises. In the third prior art, the detection area is selected only by using a high-frequency component of a video signal or a sharpness and the direction of luminance variation of the subject is not used as a factor for the image shift detection. Likewise, in the fourth prior art, the detection area is selected by using the magnitude of the contrast and the direction of luminance variation of the subject is not used.

Subsequently, a specific operation to select a part of the solid-state image sensing device as the detection area will be described in more detail. Typically, the image shift detection is made with respect to two directions: the horizontal and vertical directions of the photographic image plane. It is therefore considered to select as the detection area a part where the sharpness or the contrast of the video signal in the horizontal and vertical directions is high. In the above-mentioned fourth prior art (U.S. Pat. No. 5,218,442), contrast is detected with respect to the two directions for each block and an area is selected based on the magnitude of the contrast.

Assume now that a certain part of a solid-state image sensing device receives a subject image as shown in FIG.

1

. The subject image has a sufficient sharpness or contrast in the horizontal and vertical directions. Therefore, according to the above-described selection method, the part receiving this subject image may be selected as the detection area.

In actuality, however, in this subject image, since the output of the solid-state image sensing device does not vary when the subject is moved in the direction L in the figure, it is impossible to detect any image shift. Therefore, such a subject image is inappropriate for image shift detection and the part receiving it should not be selected as the detection area.

Assume that there are two subject images of different shapes as shown in

FIGS. 2A and 2B

. For the same reason as that of the above case, in the subject image of

FIG. 2A

, although the image shift detection precision is high when the subject is moved in the direction M in the figure, the image shift detection precision is inferior when the subject is moved in the direction N. This is because the luminance variation (contrast) of the subject image is great in the direction M but small in the direction N. On the contrary, in the subject image of

FIG. 2B

, since the direction of luminance variation is not biased, image shift detection can be made with a high precision no matter in which direction the subject is moved. Therefore, comparing these two subject images, the subject image of

FIG. 2B

is more suitable for image shift detection than that of FIG.

2

A.

Thus, even if the sharpness or contrast of a subject image is sufficient, some subject images are suitable but others are not suitable for image shift detection according to the configuration of the subject image, in particular, the direction of luminance variation (contrast). On the contrary, in the above-mentioned prior arts, since the detection area is selected merely based on the magnitude of the sharpness or contrast, a part receiving a subject image not suitable for image shift detection may be selected.

SUMMARY OF THE INVENTION

An object of the present invention is to provide a camera having an image shift detecting function which detects an image shift by estimating the image shift detection precision with the direction of luminance variation of a subject image formed on a light receiving surface of a solid-state image sensing device for detecting an image shift and by using only a partial area of the solid-state image sensing device which is estimated to have a sufficiently high detection precision.

To achieve the above-mentioned object, according to the present invention, a camera having an image shift detecting function is provided with an optical system which forms a subject image; an image sensing device which receives the subject image formed by the optical system to convert the subject image into an electric signal; image shift detecting means for detecting a relative shift between a camera body and a subject based on an output of the image sensing device; luminance variation direction detecting means for detecting a direction of luminance variation of the subject image based on the output of the image sensing device; selecting means for selecting a partial area of the image sensing device based on an output of the luminance variation direction detecting means; and controlling means for controlling the image shift detecting means to detect an image shift by using only an output corresponding to the partial area of the image sensing device selected by the selecting means.

According to such features, since in addition to the conventionally-used contrast, the direction of luminance variation of a subject image formed on the light receiving surface of an image sensing device is used to estimate the detection precision of a solid-state image sensing device which detects an image shift, the detection precision is more correctly estimated compared to the above-described prior arts. In addition, by detecting an image shift by using only a partial area of the solid-state image sensing device where the detection precision is sufficiently high, the image shift detection is more precisely made with a smaller amount of calculation.

BRIEF DESCRIPTION OF THE DRAWINGS

This and other objects and features of this invention will become clear from the following description, taken in conjunction with the preferred embodiments with reference to the accompanied drawings in which:

FIG. 1

shows a subject image to explain the selection of a detection area;

FIGS. 2A and 2B

show subject images to explain the selection of a detection area;

FIG. 3

is a block diagram schematically showing the general arrangement of a camera shake compensating system in an embodiment of a camera having an image shift detecting function according to the present invention;

FIG. 4

is a block diagram schematically showing the arrangement of an image shift detecting system;

FIG. 5

is a schematic view showing a light receiving surface of an image shift detecting sensor;

FIGS. 6A

is a schematic view showing image data of standard data;

FIG. 6B

is a schematic view showing image data of reference data;

FIG. 7

is a schematic view showing the distribution of image shift amount correlation values represented by contour lines;

FIG. 8

is a schematic view showing coordinates obtained through a one-dimensional interpolation calculation;

FIGS. 9A

to

9

D are schematic views of assistance in explaining a two-dimensional interpolation calculation;

FIGS. 10A and 10B

are schematic views of assistance in explaining a calculation to calculate a single data from a plurality of pixels;

FIG. 11

is a schematic view of assistance in explaining a calculation to simulatively calculate coordinates of two lines through a self-correlation calculation using only the standard data;

FIG. 12

is a schematic view of assistance in explaining a calculation performed in extracting the reference data;

FIG. 13

is a diagram showing an example of a flow of image shift detection and compensation;

FIG. 14

is a diagram showing an example of a relationship between an image shift amount and time;

FIG. 15

is a flowchart showing the general sequence of the camera shake compensating system;

FIG. 16

is a longitudinal cross-sectional view schematically showing the arrangement of a driving unit which is the basic unit of a driver;

FIG. 17

is a front view showing a compensating lens holding mechanism;

FIG. 18

is a longitudinal cross-sectional view of the mechanism;

FIG. 19

is a perspective view showing a mode of an engagement between a lens frame and a pinion;

FIG. 20

is a perspective view showing another mode of the engagement between the lens frame and the pinion;

FIG. 21

is a front view showing a lens driver of the embodiment;

FIG. 22

is a cross-sectional view taken on line X—X of

FIG. 21

;

FIG. 23

is a cross-sectional view taken on line Y—Y of

FIG. 21

;

FIG. 24

is a perspective view showing a modification of the lens driver;

FIGS. 25A and 25B

are front views showing conditions of the front and rear compensating lenses at their basic positions;

FIGS. 26A and 26B

are front views showing conditions of the front and rear compensating lenses after they have rotated counterclockwise;

FIG. 27

is a circuit diagram showing a drive control block;

FIG. 28

is a front view showing the arrangement of a position detecting sensor;

FIG. 29

is a schematic view of assistance in explaining a position detection principle of a position sensitive device (PSD) for one-dimensional detection;

FIG. 30

is a block diagram showing a drive control intergrated circuit (IC);

FIG. 31

is a front view showing the arrangement of the compensating lens;

FIG. 32

is a schematic view of assistance in explaining a calculation performed in extracting the reference data;

FIG. 33

is a flowchart showing a cycle of camera shake compensation;

FIG. 34

is a flowchart showing a subroutine of a target position signal output;

FIG. 35

is a schematic view showing how a change-with-time compensation coefficient is obtained;

FIG. 36

is a flowchart showing how the change-with-time compensation coefficient is obtained;

FIG. 37

is a flowchart showing a subroutine to obtain a change-with-time compensation coefficient of the front compensating lens; and

FIG. 38

is a flowchart showing a subroutine to obtain a change-with-time compensation coefficient of the rear compensating lens.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

Hereinafter, an embodiment of the present invention will be described with reference to the drawings. Referring to

FIG. 3

, there is shown the general arrangement of a camera shake compensating system according to the embodiment of the present invention. As shown in the figure, the camera system of this embodiment includes in a camera body

1

an optical system

2

for forming a subject image, driving circuits

3

a

and

3

b

which drive compensating lenses G

1

and G

2

provided in the optical system

2

, an image shift detecting system which detects the amount of a relative shift between the camera body

1

and the subject image, and a microcomputer

4

which controls the driving circuits

3

a

and

3

b

to compensate for the relative shift between the camera body

1

and the subject image based on the amount of longitudinal and lateral image shifts supplied from the image shift detecting system.

The details of the image shift detecting system will be described with reference to

FIGS. 4

to

14

. Referring to

FIG. 4

, an image shift detecting sensor

5

is an area sensor including an image sensor

6

which accumulates optical image information on its light receiving surface to convert it into an electric signal, an output amplifier

7

which amplifies the electric signal outputted from the image sensor

6

, an illuminance monitor

8

for controlling the accumulation time of the image sensor

6

and the gain of the output amplifier

7

, and a driving circuit

9

.

The light receiving surface of the image sensor

6

is constituted by I×J pixels, and K×L detection blocks constituted by (M+4)×(N+4) adjoining pixels are set thereon. The driving circuit

9

controls the accumulation time of the image sensor

6

according to an output of the illuminance monitor

8

or a control signal from an image shift detecting sensor controller

10

, controls the gain of the output amplifier

7

according to an output of the illuminance monitor

8

, and supplies a clock to the image sensor

6

and an analog-to-digital (AID) converter

11

according to a control signal from the image shift detecting sensor controller

10

.

A detecting system lens

20

is a fixed focal length lens with a focal length fd and forms a subject image on the light receiving surface of the image shift detecting sensor

5

. An electric signal (subject image information) outputted from the image shift detecting sensor

5

is inputted to the A/D converter

11

to be converted into a digital two-dimensional image data and then inputted to an image shift amount detector

13

. The image shift amount detector

13

includes an image data memory

14

, a standard data memory

15

, a reference data memory

16

, an image shift amount memory

17

and a calculator

18

. The inputted image data are stored in the image data memory

14

.

The calculator

18

extracts a standard data or a reference data from the image data stored in the image data memory

14

to store it in the standard data memory

15

or the reference data memory

16

and performs predetermined comparison calculation and estimation calculation to calculate the shift amount (image shift amount) with time of the subject image formed on the light receiving surface of the image shift detecting sensor

5

. The detected image shift amount (including longitudinal and lateral image shift amounts) is stored in the image shift amount memory

17

and transferred to a coefficient converter

19

shown in FIG.

3

.

Subsequently, an image shift detecting method will be described. In this embodiment, I=68, J=52, M=4, N=4, K=6 and L=4. Referring to

FIG. 5

, there is schematically shown the light receiving surface of the image shift detecting sensor

5

. The smaller squares represent pixels P and the squares enclosed by thick lines represent detection blocks B. In the figure, the block at the lower left end is represented as B

1.1

; the block at the upper right end, as B

6.4

; the pixel at the lower left end, as P

1.1

; and the pixel at the upper right end, as P

68.52

. The block B

k.e

(1≦k≦6, 1≦e≦4) includes

8×8 adjoining pixels P

i.j

(8×(k−1)+11≦i 8×(k−1)+18, 8×(e−1)+11≦j≦8×(e−1)+18).

The summary of the image shift detection is as follows: (i) The accumulation of the subject image (first time) is performed by the image shift detection sensor

5

. Then, the output is A/D converted and a digital image data obtained by the A/D conversion is stored in the memory

14

. (ii) Calculation is performed to determine the suitability as the standard data for each block by using the image data stored in the image data memory

14

. When there is a suitable block, the data is extracted as the standard data and stored in the standard data memory

15

. When there is no suitable block, the process returns to the above (i).

(iii) The accumulation of the subject image (second and succeeding times) is performed by the image shift detecting sensor

5

. Then, the output is A/D converted and the digital image data obtained by the A/D conversion is stored in the image data memory

14

. (iv) A reference data is extracted from the image data stored in the image data memory

14

and stored in the reference data memory

16

. Then, the image shift amount of the reference data in view of the standard data is calculated and stored in the image shift amount memory

17

.

(v) By using a plurality of image shift amount data stored in the image shift amount memory

17

, image shift amount calculation and (image shift amount) estimation calculation to compensate for an image shift compensation time lag generated by the driving of the compensating lenses G

1

and G

2

are performed. (vi) The calculated (estimation) image shift amount is stored in the image shift amount memory

17

and transferred to the coefficient converter

19

. (vii) The process returns to the above (iii).

The calculation processing performed by the calculator

18

includes the following: (i) image shift amount calculation; (ii) image data processing; (iii) selection of a detection block (extraction of a standard data); (iv) extraction of a reference data; and (v) (image shift amount) estimation calculation. These will subsequently be described.

(1) Image Shift Amount Calculation

Here, an example using a standard data constituted by an image data S

m.n

(m and n are integers, 1≦m≦M, 1≦n≦N) including M×N pixels and a reference data constituted by an image data R

m.n

(m and n are integers, −H

1

≦m≦M+H

2

+1, −V

1

≦n≦N+V

2

+1) including (M+H1+H

2

+2)×(N+V1+V

2

+2) pixels will be described with reference to

FIGS. 6A and 6B

. A standard data S

1.1

shown in

FIG. 6A and a

reference data R

−H1.−V1

shown in

FIG. 6B

are image data corresponding to the pixel at the lower left end. A standard data S

M.N

and a reference data R

M+H2+1,N+V2+1

are image data corresponding to the pixel at the upper left end.

The image shift amount on the light receiving surface of the image shift detecting sensor

5

is represented as (x, y) using a lateral image shift amount x (in the figure, leftward direction is positive) and a longitudinal image shift amount y (upward direction is positive) with the length of one side of a pixel as a unit. When the standard data S

m.n

(1≦m≦M, 1≦n≦N) and a part of the reference data R

m.n

(x+1≦m≦x+M, y+1≦n≦N, x and y are integers, −H1-e≦x≦H

2

+1, −V

1

−e≦y≦V

2

+1) coincide with each other, the image shift amount of the reference data in view of the standard data is (x, y).

\begin{matrix} H (x, y) = \sum_{m = 1}^{M} \sum_{n - 1}^{N} &LeftBracketingBar; S_{m, n} - R_{m + x, n + y} &RightBracketingBar; & Expression 1 \end{matrix}

where x and y are integers, −H

1

−e≦x≦H

2

+1, and −V

1

−e≦y≦V

2

+1.

Further, as a value representative of the degree of coincidence between the standard data and the reference data, a correlation value is defined by the above expression 1. Specifically, when H(x

0

, y

0

)=0, the image shift amount is (x

0

, y

0

). The farther x is away from x

0

and the farther y is away from y

0

, the greater the correlation value H(x, y) is.

Referring to

FIG. 7

, there are shown contour lines representing the distribution of the correlation value H(x, y) when the image shift amount is (x

0

, y

0

). While in actuality, correlation values are obtained only at the lattice points in the figure since x and y are integers, assuming that correlation values are present between the lattices, the points where the correlation values equal each other are connected to form the contour lines.

In

FIG. 7

, while the correlation value H(x

0

, y

0

) which does not really exist is assumed to be 0, a correlation value H(x

min

, y

min

) at a lattice point (x

min

, y

min

) closest to the coordinate (x

0, y

0

) is considered to be the minimum of existing correlation values. Specifically, by calculating the correlation value H(x, y) with respect to the integers x and y of −H

1

−e≦x≦H

2

+1, −V

1

−e≦y≦V

2

+1 to find the minimum value H(x

min

, y

min

), an approximate value (x

min

, y

min

) of the image shift amount (x

0

, y

0

) is obtained. This calculation is called a correlation calculation.

Subsequently, an interpolation calculation will be described. The interpolation calculation is a calculation to find a coordinate (x

0

, y

0

) where the correlation value is estimated to be 0, for example, by using only the correlation values at the lattice points of FIG.

7

. Only by the correlation calculation, only the approximate value (x

min

, y

min

) of the image shift amount (x

0

, y

0

) can be calculated and it is impossible to detect and compensate for an image shift with a high precision. The interpolation calculation is indispensable for such high-precision image shift detection and compensation.

Assume that R

min

(x

min

+1≦m≦M+x

min

, y

min

+1≦n≦N+y

min

) is obtained through the above-described correlation calculation as the approximate value (x

min

, y

min

) of the image shift amount, in other words, as a portion of the reference data which more accurately coincides with the standard data. For the interpolation calculation, a reference data R

min

(x

min

≦m≦M+x

min

+1, y

min≦n≦N+y

min

+1) of a slightly wider range is necessary. For this reason, −H

1

≦x

min

≦H

2

, −V

1

≦y

min

≦V

2

, so that in the correlation calculation, the correlation value H(x, y) is calculated with respect to the integers x and y of −H

1

≦x≦H

2

and −V

1

≦y≦V

2

.

By using these image data, first, a coordinate (x

v

, y

min

+v) is calculated where the correlation value is estimated to be minimum on a straight line y=y

min

+v (v=−1, 0, 1). To do so, an expression 2 shown below is defined as a new correlation value. Then, a correlation value Hx

0

(x, y

min

+v) is calculated with respect to x=x

min

−1, x

min

, x

min

+1 to obtain the minimum value Hx

0

(x

minv

, y

min

+v). Then, a total of three data are calculated by an expression 3 shown below. Then, an interpolation calculation on the straight line y=y

min

+v is performed by an expression 4 shown below to calculate the value of x

v

, thereby calculating a coordinate (x

v

, y

min

+v) where the correlation value is estimated to be minimum on the straight line y=y

min

+v.

\begin{matrix} {Hx}_{0} (x, y) = \sum_{n = 1}^{N} (\frac{&LeftBracketingBar; S_{1, n} - R_{1 + x, n + y} &RightBracketingBar;}{2} + \sum_{m = 2}^{M - 1} &LeftBracketingBar; S_{m, n} - R_{m + x, n + y} &RightBracketingBar; + \frac{&LeftBracketingBar; S_{M, n} - R_{M + x, n + y} &RightBracketingBar;}{2}) {Hx}_{- 1} (x, y) = \sum_{n = 1}^{N} \sum_{m = 2}^{M} &LeftBracketingBar; S_{m, n} - R_{m + x - 1, n + y} &RightBracketingBar; {Hx}_{1} (x, y) = \sum_{n = 1}^{N} \sum_{m = 2}^{M - 1} &LeftBracketingBar; S_{m, n} - R_{m + x + 1, n + y} &RightBracketingBar; where x_{\min} - 1 \leq x \leq x_{\min} + 1, y_{\min} - 1 \leq y \leq y_{\min} + 1. & Expression 2 \end{matrix}

\begin{matrix} H_{- 1} = {Hx}_{- 1} (x_{minv}, y_{\min} + v) H_{0} = {Hx}_{0} (x_{minv}, y_{\min} + v) H_{1} = {Hx}_{1} (x_{minv}, y_{\min} + v) & Expression 3 \end{matrix}

\begin{matrix} When H_{- 1} \leq H_{1}, x_{v} = x_{minv} + \frac{1}{2} \times \frac{H_{- 1} - H_{1}}{H_{1} - H_{0}} When H_{- 1} > H_{1}, x_{v} = x_{minv} + \frac{1}{2} \times \frac{H_{- 1} - H_{1}}{H_{- 1} - H_{0}} & Expression 4 \end{matrix}

Then, a coordinate (x

min

+h, y

h

) where the correlation value is estimated to be minimum on a straight line x=x

min

+h (h=−1, 0, 1) is calculated. To do so, Hy is defined by an expression 5 shown below in a similar manner to Hx. Then, a correlation value Hy

0

(x

min

+h, y) is calculated with respect to y=y

min

−1, y=y

min

, y=y

min

+1 to obtain the minimum value Hy

0

(x

min

+h, y

minh

).

\begin{matrix} {Hy}_{0} (x, y) = \sum_{m = 1}^{M} (\frac{&LeftBracketingBar; S_{m, 1} - R_{m + x, 1 + y} &RightBracketingBar;}{2} + &AutoLeftMatch; \sum_{n = 2}^{N - 1} &LeftBracketingBar; S_{m, n} - R_{m + x, n + y} &RightBracketingBar; + \frac{&LeftBracketingBar; S_{m, n} - R_{m + x, N + y} &RightBracketingBar;}{2}) {Hy}_{- 1} (x, y) = \sum_{m = 1}^{M} \sum_{n = 2}^{N} &LeftBracketingBar; S_{m, n} - R_{m + x, n + y - 1} &RightBracketingBar; {Hy}_{1} (x, y) = \sum_{m = 1}^{M} \sum_{n = 2}^{N - 1} &LeftBracketingBar; S_{m, n} - R_{m + x, n + y + 1} &RightBracketingBar; where x_{\min} - 1 \leq x \leq x_{\min} + 1, y_{\min} - 1 \leq y \leq y_{\min} + 1. & Expression 5 \end{matrix}

\begin{matrix} H_{- 1} = {Hy}_{- 1} (x_{\min} + h, y_{minh}) H_{0} = {Hy}_{0} (x_{\min} + h, y_{minh}) H_{1} = {Hy}_{1} (x_{\min} + h, y_{minh}) & Expression 6 \end{matrix}

\begin{matrix} When H_{- 1} \leq H_{1}, y_{h} = y_{minh} + \frac{1}{2} \times \frac{H_{- 1} - H_{1}}{H_{1} - H_{0}} When H_{- 1} > H_{1}, y_{h} = y_{minh} + \frac{1}{2} \times \frac{H_{- 1} - H_{1}}{H_{- 1} - H_{0}} & Expression 7 \end{matrix}

Then, a total of six data are calculated by the above expression 6. Then, by using the above expression 7, an interpolation calculation on the straight line x=x

min

+h is performed to calculate the value of y

h

, thereby calculating a coordinate (x

min

+h, y

h

) where the correlation value is estimated to be minimum on the straight line x=x

min

+h. By the above-described correlation calculation, as shown in

FIG. 8

, the coordinates (x

−1

, y

min

−1), (x

0

, y

min

), (x

1

, y

min

+1), (x

min

−1, y

−1

), (x

min

, y

0

) and (x

min

+1, y

1

) where the correlation values are estimated to be minimum on six straight lines are obtained.

Subsequently, a two-dimensional interpolation calculation will be described with reference to

FIGS. 9A

to

9

D. The coordinate (x

0

, y

0

) where the correlation value is estimated to be zero is present in an area surrounded by four straight lines x=x

min

, x=x

min

+h, y=y

min

, y=y

min

+v (h=±1, v=±1). The coordinates where the correlation value is minimum on the straight lines are (x

min

, y

0

), (x

min

+h, y

h

), (x

0

, y

min

) and (x

v

, y

min

+v), respectively.

It is assumed here that a straight line

91

passing the coordinates (x

min

, y

0

) and (x

min

+h, y

h

) and a straight line

92

passing the coordinates (x

0

, y

min

) and (x

v

, y

min

+v) pass the vicinity of the coordinate (x

0

, y

0

), and approximately, the point of intersection of the two straight lines

91

and

92

coincides with the coordinate (x

0

, y

0

). The approximate values of x

0

and y

0

are obtained by an expression 9 shown below by using the coordinate of the point of intersection of the straight lines

91

and

92

by an expression 8 shown below:

\begin{matrix} Straight line 91 : y = \frac{y_{h} - y_{0}}{h \times (x - x_{\min}) + y_{0}} Straight line 92 : y = \frac{v}{(x_{v} - x_{0}) \times (x - x_{0}) + y_{\min}} & Expression 8 \end{matrix}

\begin{matrix} x_{0} \approx x_{\min} + \frac{x_{0} - x_{\min} + (y_{0} - y_{\min}) \times \frac{x_{v} - x_{0}}{v}}{1 - \frac{x_{v} - x_{0}}{v} \times \frac{y_{h} - y_{0}}{h}} y_{0} \approx y_{\min} + \frac{y_{0} - y_{\min} + (x_{0} - x_{\min}) \times \frac{y_{h} - y_{0}}{h}}{1 - \frac{x_{v} - x_{0}}{v} \times \frac{y_{h} - y_{0}}{h}} & Expression 9 \end{matrix}

Since the coordinate (x

0

, y

0

) thus calculated should exist in an area surrounded by the four straight lines x=x

min

, x=x

min

+h, y=y

min

, y=y

min

+v (h=±1, v=±1):

when h=1, x

min

≦x

0

≦x

min

+1 should be fulfilled;

when h=−1, x

min

−1≦x

0

≦x

min

should be fulfilled;

when v=1, y

min

≦y

0

≦y

min

+1 should be fulfilled; and

when v=−1, y

min

−1≦y

0

≦y

min

should be fulfilled.

Therefore, it is assumed that the calculation of the coordinate (x

0

, y

0

) is impossible when these are not fulfilled.

Returning to

FIG. 8

, when the six coordinates (x

−1

, y

min

−1), (x

0

, y

min

), (x

1

, y

min

+1), (x

min

−1, y

−1

), (x

min

, y

0

) and (x

min

+1, y

1

) are present, the following four combinations are obtained as shown in

FIGS. 9A

to

9

D:

(i) h=v=1;

(ii) h=1, v=−1;

(iii) h=−1, v=1; and

(iv) h=v=−1.

In this case, the coordinate (x

0

, y

0

) is calculated for each of (i) to (iv). When the coordinate (x

0

, y

0

) cannot be calculated for all of the four, the calculation of the coordinate (x

0

, y

0

) is impossible. When the coordinate (x

0

, y

0

) can be calculated for one of the four, the calculated coordinate is the coordinate (x

0

, y

0

). When the coordinate (x

0

, y

0

) can be calculated for some of the four, the coordinate (x

0

, y

0

) is the average of the calculated coordinates. The image shift amount (x

0

, y

0

) of the reference data in view of the standard data is thus calculated with a high precision.

(2) Image Data Processing

The two-dimensional image data obtained by A/D converting the output of the image shift detecting sensor

5

includes low-frequency components and noise components which degrade the precision of the image shift amount calculation. Therefore, to enable a highly precise image shift detection, it is necessary to perform a specific calculation processing of the image data stored in the image data memory

14

to reduce the components which can degrade the calculation precision.

In this embodiment, as shown in

FIG. 10A

, one processing data G

i.j

is calculated by using image data D

i.j−1

, D

i−1.j

, D

i.j

, D

i+1.j

and D

i.j+1

corresponding to four adjoining pixels P

i.j−1

, P

i−1.j

, P

i.j

, P

i+1.j

and P

i.j+1

.

For example, as shown in

FIG. 10B

, by performing a specific calculation processing (image data processing) of the image data D

i.j

(8×(k−1)+11≦i≦8×(k−1)+18, 8×(e−1)+11≦j≦8×(e−1)+18) of the block B

k.1

constituted by 8×8 adjoining pixels P

i.j

, 6×6 processing data G

i.j

(8×(k−1)+12≦i≦i8×(k−1)+17, 8×(e−1)+12≦j≦8×(e−1)+17) are obtained. In this case, the image data corresponding to the pixels at the four corners of the block B

k.1

are not used.

As the operation expression, an expression such as an expression 10 shown below is used. By such an image data processing, the components which can degrade the calculation precision are reduced, so that a highly precise image shift detection is enabled.

\begin{matrix} G_{i, j} = \frac{D_{i . j - 1} + D_{i - 1. j} - D_{i + 1. j} - D_{i . j + 1}}{D_{i . j - 1} + D_{i - 1, j} + D_{i . j} + D_{i + 1. j} + D_{i . j + 1}} & Expression 10 \end{matrix}

(3) Selection of Detection Block (Extraction of Standard Data)

In this embodiment, 6×4 detection blocks are set on the light receiving surface of the image shift detecting sensor

5

. A calculation to determine the suitability as the standard data is performed for each block to select a suitable block as the standard data. Thereby, only a block superior in image shift amount calculation precision is selected, so that a high calculation precision is obtained with a small amount of calculation. By providing a plurality of detection blocks, in other words, by providing a wide image shift detection area, the probability of the presence of a block superior in image shift amount calculation precision increases, so that a more highly precise image shift detection is enabled.

The suitability as the standard data, i.e. the superiority or inferiority in the image shift amount calculation precision will be described by using as an example a standard data constituted by an image data S

m.n

(m and n are integers, 1≦m≦M, 1≦n≦N). As a factor to determine the superiority or inferiority in the calculation accuracy, contrast is considered. Therefore, a lateral contrast HC and a longitudinal contrast VC of the standard data are defined by an expression 11 shown below. In a standard data having high contrasts HC and VC, the correlation values in the image shift calculation are also high. For this reason, such a standard data is not readily affected by the noise components included in the image data, so that the calculation precision is high.

\begin{matrix} HC = \sum_{m = 1}^{M - 1} \sum_{n = 1}^{N} &LeftBracketingBar; S_{m, n} - S_{m + 1, n} &RightBracketingBar; VC = \sum_{m = 1}^{M} \sum_{n = 1}^{N - 1} &LeftBracketingBar; S_{m, n} - S_{m, n + 1} &RightBracketingBar; & Expression 11 \end{matrix}

Further, as a factor to determine the superiority or inferiority in the two-dimensional interpolation calculation precision of the image shift amount calculation, an angle is considered at which the straight line

91

passing the coordinates (x

min

, y

0

) and (x

min

+h, y

h

) and the straight line

92

passing (x

0

, y

min

) and (x

v

, y

min

+v) intersect each other. It is considered that the calculation precision is the highest when the two straight lines intersect at 90°. In that case,

(Inclination of straight line

91

)×(inclination of straight line

92

)=−1

This is modified into

\frac{y_{h} - y_{0}}{h} + \frac{x_{v} - x_{0}}{v} = 0

The left member of this expression is expressed by A (h, v). The smaller |A(h, v)| is, the closer to 90° the angle at which the two straight lines intersect and the higher the calculation precision is. Further, as a factor to determine the superiority and inferiority in the calculation precision, the sum total of |A(h, v)|:

A=|A(1, 1)|+|A(1, −1)|+|A(−1,1)|+|A(−1, −1)|

is defined.

To simulatively calculate the factor A by using only the standard data, a self-correlation calculation based on the standard data is performed. For this calculation, a data S

m.n

(m and n are integers, 0≦m≦M+1, o≦n≦N+1) of a slightly wider range is necessary. In this case, the image data of a detection block constituted by (M+4)×(N+4) adjoining pixels is converted into (M+2)×(N+2) processing data by the image data processing. By using the processing data, the calculation to determine the suitability as the standard data is performed. When the processing data are determined to be suitable as the standard data, M×N data of the processing data are extracted as the standard data.

The self-correlation calculation based on the standard data will be described. First, a coordinate (x

v

, v) where the correlation value is estimated to be minimum on the straight line y=v (v=±1) is calculated. To do so, as a new correlation value, Sx is defined from an expression 12 shown below. Then, a correlation value Sx

0

(x, y) is calculated with respect to x=−1, 0, 1 to obtain the minimum value Sx

0

(x

minv

, v).

\begin{matrix} \begin{matrix} {Sx}_{0} (x, y) = \sum_{n = 1}^{N} (\frac{&LeftBracketingBar; S_{1, n} - S_{1 + x, n + y} &RightBracketingBar;}{2} + \\ \sum_{m = 2}^{M - 1} &LeftBracketingBar; S_{m, n} - S_{m + x, n + y} &RightBracketingBar; + \frac{&LeftBracketingBar; S_{M, n} - S_{M + x, n + y} &RightBracketingBar;}{2}) \\ {Sx}_{- 1} (x, y) = \sum_{n = 1}^{N} \sum_{m = 2}^{M} &LeftBracketingBar; S_{m, n} - S_{m + x - 1, n + y} &RightBracketingBar; \\ {Sx}_{1} (x, y) = \sum_{n = 1}^{N} \sum_{m = 1}^{M - 1} &LeftBracketingBar; S_{m, n} - S_{m + x + 1, n + y} &RightBracketingBar; \\ w here - 1 \leq x \leq 1, y = \pm 1 \end{matrix} & Expression 12 \end{matrix}

Then, three data are calculated by an expression 13. Then, an interpolation calculation on the straight line y=v is performed by an expression 14 to calculate x

v

. When the correlation value becomes minimum on the straight line y=v, an estimated coordinate (x

v

, v) is calculated.

\begin{matrix} S_{- 1} = {Sx}_{- 1} (x_{minv}, v) S_{0} = {Sx}_{0} (x_{minv}, v) S_{1} = {Sx}_{1} (x_{minv}, v) & Expression 13 \end{matrix}

\begin{matrix} When S_{- 1} \leq S_{1}, x_{v} = x_{minv} + \frac{1}{2} \times \frac{s_{- 1} - S_{1}}{S_{1} - S_{0}} When S_{- 1} > S_{1}, x_{v} = x_{minv} + \frac{1}{2} \times \frac{S_{- 1} - S_{1}}{S_{- 1} - S_{0}} & Expression 14 \end{matrix}

Then, a coordinate (h, y

h

) where the correlation value is estimated to be minimum on the straight line x=h (h=±1) is calculated. To do so, like the case of Sx, Sy is defined by an expression 15 shown below. Then, a correlation value Sy

0

(h, y) is calculated with respect to y=−1, 0, 1 to obtain the minimum value Sy

0

(h, y

minh

)

Then, three data are calculated by an expression 16 shown below. Then, an interpolation calculation on the straight line x=h is performed by using an expression 17 shown below to calculate a coordinate (h, y

h

) where the correlation value is estimated to be minimum on the straight line x=h.

The coordinate where the correlation value on straight lines x=0 and y=0 is (0, 0), and x

0

=y

0

=0. Thus, by the self-correlation calculation using only the standard data, x

−1

, x

0

, x

1

, y

−1

, y

0

and y

1

are simulatively calculated as shown in FIG.

11

. By using these, the factor A to determine the superiority and inferiority in the two-dimensional interpolation calculation precision is expressed as

A=|y

1

+x

1

|+|y

1

−x

1

|+|y

−1

−x

1

|+|y

−1

+x

−1

|

\begin{matrix} \begin{matrix} {Sy}_{0} (x, y) = \sum_{m = 1}^{M} (\frac{&LeftBracketingBar; S_{m, 1} - S_{m + x, 1 + y} &RightBracketingBar;}{2} + \\ \sum_{n = 2}^{N - 1} &LeftBracketingBar; S_{m, n} - S_{m + x, n + y} &RightBracketingBar; + \frac{&LeftBracketingBar; S_{m, N} - S_{m + x, N + y} &RightBracketingBar;}{2}) \\ {Sy}_{- 1} (x, y) = \sum_{m = 1}^{M} \sum_{n = 2}^{N} &LeftBracketingBar; S_{m, n} - S_{m + x, n + y - 1} &RightBracketingBar; \\ {Sy}_{1} (x, y) = \sum_{m = 1}^{M} \sum_{n = 1}^{N - 1} &LeftBracketingBar; S_{m, n} - S_{m + x, n + y + 1} &RightBracketingBar; \\ where x = \pm 1, - 1 \leq y \leq 1. \end{matrix} & Expression 15 \end{matrix}

\begin{matrix} S_{- 1} = {Sy}_{- 1} (h, y_{minh}) S_{0} = {Sy}_{0} (h, y_{minh}) S_{1} = {Sy}_{1} (h, y_{minh}) & Expression 16 \end{matrix}

\begin{matrix} When S_{- 1} \leq S_{1}, y_{h} = y_{minh} + \frac{1}{2} \times \frac{s_{- 1} - S_{1}}{S_{1} - S_{0}} When S_{- 1} > S_{1}, y_{h} = y_{minh} + \frac{1}{2} \times \frac{S_{- 1} - S_{1}}{S_{- 1} - S_{0}} & Expression 17 \end{matrix}

The lateral contrast HC and the longitudinal contrast VC of the standard data and the factor A to determine the superiority and inferiority in the two-dimensional interpolation precision of the image shift amount calculation are thus calculated for each block, and a block where the contrasts HC and VC are high and the factor A is small is selected and extracted as the standard data. By such a selection of the detection block, a highly precise image shift detection can be performed with a small amount of calculation.

(4) Extraction of Reference Data

As shown in the section (1) of the image shift amount calculation, when the image shift amount calculation is performed with a standard data constituted by image data corresponding to M×N pixels and a reference data constituted by (M+

2

S+2)×(N+

2

S+2) image data, an image shift amount (x, y) is obtained in a range of approximately −S−e≦x≦S+1, −S−e≦y≦S+1. That is, the larger the number of reference data is, the larger image shift amount can be handled. However, the larger the number of reference data is, the larger the calculation amount is, so that the time lag of the image shift compensation increases. As a result, the image shift compensation cannot be made with a high precision. Therefore, it is necessary to extract a reference data with which the amount of image shift calculation is minimized and a large amount of image shift can be handled.

The extraction of the reference data will be described. For example, assume that a block

k.1

constituted by 8×8 adjoining pixels is selected as the detection block and that 4×4 processing data G

i.j

(8×(k−1)+13≦i≦8×(k−1)+16, 8×(e−1)+13≦j≦8×(e−1)+16) are stored in the standard data memory

15

. The image shift amount of the reference data obtained by the nth accumulation by the image shift detection sensor

5

based on the standard data is assumed to be (x

n

, y

n

), and the image shift amount of the reference data obtained by the next n+1th accumulation is assumed to be (x

n+1

, y

n+1

). When the shift amount of the subject image on the light receiving surface of the image shift detecting sensor

5

between the nth and n+1th accumulations is (x

s

, y

s

); x

n+1

=x

n

+x

s

and y

n+1

=y

n

+y

s

.

Assuming that the range of (x

s

, y

s

) is −s≦x

s

≦s, −s≦y

s

≦s, the range of (x

n+1

, y

n+1

) can be expressed by x

n

−s≦x

n+1

≦x

n

+s, y

n

−s≦y

n+1

≦y

n

+s. This is modified into

INT(x

n

−s)≦x

n+1

<INT(x

n

+s)+1

INT(y

n

−s)≦y

n+1

<INT(y

n

+s)+1

(see FIG.

12

). INT(x) is a function to omit the figures below the decimal point of x to make it integral.

To calculate an image shift amount in this range, a reference data constituted by processing data G

i.j

(8×(k−1)+INT(x

n

−s)+13≦i≦8×(k−1)+INT(x

n

+s)+17, 8×(e−1)+INT(y

n

−s)+13≦j≦8×(e−1)+INT(y

n

+s)+17) is extracted.

Since INT(a+b)−INT(a−b)=INT(2×b) or INT(2×b)+1, the necessary number of reference data is (INT(2×s)+6)×(INT(2×s)+

6

) at maximum. Then, the standard data is replaced by S

m.n

(1≦m≦4, 1≦n≦4) assuming that it can be expressed by an expression 18 shown below, and the reference data is replaced by R

m.n

, (−INT(x

n

+0.5)+INT(x

n

−s)+1≦m≦INT(x

n

+s)−INT(x

n

+0.5)+5, −INT(y

n

+0.5)+INT(y

n

−s)+1≦n≦INT(y

n

+s)−INT(y

n

+0.5)+5) assuming that it can be expressed by an expression 19.

S

m.n

=G

8×(k−1)+m+12, 8×(e−1)+n+12

Expression 18

R

m.n

=G

8×(k−1)+m+INT(xn+0.5)+12, 8×(e−1)+n+INT(yn+0.5)+12

Expression 19

By using these data, the image shift calculation shown in the above section (1) is performed to calculate the image shift amount (x

0

, y

0

), and (INT(x

n

+0.5)+x

0

, INT(y

n

+0.5)+y

0

) is obtained as the image shift amount of a reference data obtained by the (n+1)th accumulation. The extraction of the reference data thus performed minimizes the amount of image shift calculation and enables the apparatus to handle a large image shift amount.

(5) Estimation Calculation (of Image Shift Amount)

Referring first to

FIG. 13

, there is shown a brief flow of the image shift detection and compensation (only in the lateral direction). x

n

is an image shift amount of a reference data obtained by the nth accumulation by the image shift detecting sensor. Since the reference data is considered to be subject information obtained at an accumulation central time t

n

of the nth accumulation, the image shift amount x

n

is an image shift amount at the time t

n

. At this time, the image shift amount x

n

is calculated at a time t

n

+t

e1

due to the time lag of the reading out of the image data and the time lag of the image shift amount calculation, and transferred to the coefficient converter

19

.

Further, by the time lag of driving of the compensating lenses, the compensation of the image shift amount x

n

is completed at a time t

n

+t

e1

+t

e2

. For this reason, the difference between the image shift amount at the time t

n

+t

e1

+t

e2

and the image shift amount x

n

is left as an uncompensated image shift, so that it is impossible to highly precisely compensate for the image shift. To avoid this, an estimation calculation to eliminate the time lag of the image shift detection and compensation is necessary.

The details of the image shift amount estimation calculation will be described with reference to FIG.

14

. In the figure, x

n

, x

n−1

, x

n−2

, . . . , x

n−9

are image shift amounts of the reference data obtained by the nth, (n−1)th, (n−2)th, . . . , (n−9)th accumulations by the image shift detecting sensor

5

, respectively, and are image shift amounts at the accumulation central times t

n

, t

n−1

, t

n−2

, . . . , t

n−9

, respectively. By using these data, the image shift amount at the time t

n

+t

e1

+t

e2

is estimated.

\begin{matrix} v_{1} = \frac{x_{n} - x_{n - 2}}{t_{n} - t_{n - 2}} & Expression 20 \end{matrix}

First, a present image shift speed v

1

is calculated by the expression 20. The image shift speed is not calculated between the times t

n

and t

n−1

in order to increase the denominator (time) to improve the image shift speed detection precision. By using this, an estimated image shift amount px is calculated as

px=x

n

+v

1

×(t

e1

+t

e2

)

This operation expression is premised that the image shift speed is constant. When the variation in image shift speed is great and when the time lag (t

e1

+t

e2

) is great, in view of the variation in speed (acceleration), a present image shift acceleration a is calculated by an expression 22 shown below by using a past image shift speed v

2

obtained by an expression 21 shown below.

\begin{matrix} v_{2} = \frac{x_{n - 1} - x_{n - 3}}{t_{n - 1} - t_{n - 3}} & Expression 21 \end{matrix}

\begin{matrix} a = \frac{2 \times (v_{1} - v_{2})}{t_{n} - t_{n - 1} + t_{n - 2} - t_{n - 3}} & Expression 22 \end{matrix}

By using these, the estimated image shift amount px is calculated as

px = x_{n} + v_{1} \times (t_{e1} + t_{e2}) + a \times \frac{{(t_{e1} + t_{e2})}^{2}}{2}

By using this operation expression, a highly precise image shift amount estimation calculation is enabled even when the variation in image shift speed is great and when the time lag (t

e1

+t

e2

) is great.

When the accumulation time of the image shift detecting sensor

5

is short because of a high-luminance subject to decrease the intervals among the accumulation central times t

n

, t

n−1

, t

n−2

, . . . , and t

n−9

and when more precise detection of the image shift speed and image shift acceleration is desired, the present image shift speed v

1

is calculated by an expression 23 shown below, the past image shift speed v

2

is calculated by an expression 24 shown below, and the present image shift acceleration a is calculated by an expression 25 shown below.

\begin{matrix} v_{1} = \frac{x_{n} - x_{n - 4}}{t_{n} - t_{n - 4}} & Expression 23 \end{matrix}

\begin{matrix} v_{2} = \frac{x_{n - 5} - x_{n - 9}}{t_{n - 5} - t_{n - 9}} & Expression 24 \end{matrix}

\begin{matrix} a = \frac{2 \times (v_{1} - v_{2})}{t_{n} + t_{n - 4} - t_{n - 5} - t_{n - 9}} & Expression 25 \end{matrix}

Thus, even if the luminance of a subject varies to vary the intervals (image shift detection intervals) among the accumulation central times and even if the focal length of the taking lens varies due to zooming to vary a required level of the image shift detection precision, these cases can be handled by employing an image shift amount estimation operation expression which is determined to be optimum.

By the image shift amount estimation calculation shown above, the time lag of the image shift detection and compensation is eliminated to enable a highly precise image shift compensation.

Subsequently, the general arrangement of the camera shake compensating system of this embodiment provided with the above-described function of the image shift detecting system will be described. Returning to

FIG. 3

, the detecting system forms light from a subject into an image on the image shift detecting sensor

5

through the detecting system lens

20

. The image shift detecting sensor

5

converts the formed subject information into an electric signal and supplies the electric signal representative of the subject information to the image shift amount detector

13

through the A/D converter

11

.

At the image shift amount detector

13

, subject information obtained just before the image shift compensation driving, is stored as the standard data. By using as the reference data the subject information inputted through the A/D converter

11

during the image shift compensation driving, a predetermined comparison calculation of the standard and reference data is performed to generate an image shift amount data to drive the compensating lenses G

1

and G

2

. The image shift amount data includes a longitudinal image shift amount showing an image shift amount when the camera body

1

)shakes upward and downward (in a direction perpendicular to the plane of the figure) and a lateral image shift amount showing an image shift amount when the camera body

1

shakes to the left and right.

The coefficient converter

19

has the following working: The longitudinal image shift data and the lateral image shift data from the image shift amount detector

13

are inputted to the coefficient converter

19

together with an output data of a temperature sensor T. The coefficient converter

19

converts the data into data associated with target movement positions of the compensating lenses G

1

and G

2

necessary for the compensating lenses G

1

and G

2

to actually compensating for the longitudinal and lateral image shift amounts appropriately.

The coefficient converter

19

also absorbs the following error factors associated with camera shake compensation to increase the precision of the camera shake compensation: (i) an error caused by the fact that proportional coefficients of the target lens positions relative to the image shift amount differ among each camera; (ii) an error caused by the fact that the proportional coefficient varies due to a change of camera with time; (iii) an error caused by the fact that the proportional coefficient varies with a variation in characteristics of each lens and a variation in electric characteristics due to a variation in temperature; and (iv) an error caused by a difference between the image shift directions of the longitudinal and lateral image shift amounts and the movement directions of the compensating lenses G

1

and G

2

. The error (iv) is caused by a difference in direction generated in assembling the camera or by the arrangement of a subsequently-described driving mechanism of the compensating lenses G

1

and G

2

.

The data associated with the coefficient converter

19

are stored in an electrically erasable programmable read only memory (EEPROM) because of the non-uniformity and the change with time of the camera.

The driving system is provided with the driving circuits

3

a

and

3

b

which drive the compensating lenses G

1

and G

2

. Based on data of a target image shift compensation position data portion

21

a

of the compensating lens G

1

and a target image shift compensation position data

21

b

of the compensating lens G

2

from the coefficient converter

19

, the G

1

and G

2

driving circuits

3

a

and

3

b

move the compensating lenses G

1

and G

2

to the target positions.

The G

1

driving circuit

3

a

reads out the position of the compensating lens G

1

based on a signal of the position detecting sensor P and drives an actuator Aa through feedback to move the compensating lens G

1

to the target image shift compensation position. The G

2

driving circuit

3

b

moves the compensating lens G

2

in a like manner. The outputs of the position detecting sensors Pa and Pb are also inputted to the microcomputer

4

through the A/D converter

22

to generate data to confirm the present positions of the compensating lenses G

1

and G

2

at a G

1

and G

2

present position data portion

23

.

Describing the optical system, by moving the compensating lens G

1

substantially in a direction perpendicular to the plane of the figure and moving the compensating lens G

2

substantially in a direction horizontal to the plane of the figure, the optical path of the light from the subject can be bent in a direction perpendicular to the optical axis according to the movement amounts of the lenses G

1

and G

2

. During exposure, the lenses G

1

and G

2

are driven by the above-described driving system and the shift of the subject image due to camera shake is compensated for so that the subject image always remains stationary on the film surface, thereby obtaining a picture without any image blur even if the exposure time is long. Reference numeral

24

represents a glass cover.

Subsequently, the general operation sequence of the camera shake compensating system will be described with reference to FIG.

15

. When the main switch is turned on at step #

5

, the coefficient converter

19

performs a coefficient correcting processing at step #

10

. This processing is performed to correct the conversion coefficient used to convert the image shift amount from the detecting system into the target movement positions of the compensating lenses to remove the conversion errors caused by the change of the camera with time.

When a release button is depressed at step #

15

, the calculation of the image shift amount data by the detecting system is started at step #

20

. First, the subject information is read out from the image shift detecting sensor

5

and stored as the standard data. Then, based on a signal from the image shift detecting sensor controller

10

, the reading of the subject information is continued until subsequently-performed exposure is completed and the image shift amount data is calculated as the reference data.

While the image shift amount data is being calculated, a lens system stopper is deactivated at step #

25

, and a signal is generated to drive the compensating lenses to initial positions located substantially in the center of the lens movable ranges at step #

30

. Then, in order to drive the compensating lenses G

1

and G

2

according to the image shift amount data from the detecting system, the target lens movement data obtained through the coefficient converter based on the image shift amount data from the detecting system are transmitted to the G

1

and G

2

driving circuits

3

a

and

3

b.

At the microcomputer

4

, since the data on the present positions of the compensating lenses G

1

and G

2

are obtained, the absolute value of the difference between the present position and the target position (i.e. the present uncompensated amount) is determined. When the compensating lenses are started to be driven from a stationary condition at step #

35

, at first, the present position does not follow the target position (i.e. the present uncompensated amount is great) for some period of time. If exposure is started during this period, the camera shake compensating function does not work correctly. To avoid this, a driving system following completion routine is executed at step #

40

so that the process wait until the present uncompensated amount is stabilized at a small value.

When the present uncompensated amount is stabilized at a small value, exposure is started at step #

45

, and exposure is performed for a predetermined period of time by a non-illustrated photometry circuit. During exposure, the compensating lenses are moved by the driving system based on the image shift amount data from the detecting system to compensate for the camera shake and the absolute value of the difference between the present and target positions of the compensating lenses is monitored. The maximum value or the average value of the absolute value during exposure is stored as the uncompensated amount. After exposure is completed at step #

50

, the processings by the detecting system and the driving system are completed at step #

60

. Then, at step #

65

, to stop the compensating lenses G

1

and G

2

, the lens system stopper is activated.

At step #

70

, the uncompensated amount during exposure can be great when compensation is insufficient due to a drastic change of camera shake during exposure. In this case, there are some blurs left in the exposed picture. The user is notified of this through a warning display.

Subsequently, the mechanical structure and operation of the compensating lens driver will be described with reference to

FIGS. 16

to

24

. For simplicity,

FIGS. 16

to

18

shows a driving unit for one lens G as a basic structure of the compensating lens driver of this embodiment. A lens frame

25

which holds the compensating lens G is pivotally supported by a rotational shaft

26

and formed to be rotatable about the compensating lens G along its periphery. The rotating axis

26

is fixed to a cylindrical holder

27

. Reference numeral

28

represents a curbing member which curbs unnecessary movements of the lens frame

25

along the optical axis by nipping a driven portion

25

a

of the lens frame

25

.

Reference numeral

29

represents an actuator supported by the cylindrical holder

27

. The actuator

29

is arranged outside the photographic optical path and drives the lens frame

25

around the rotational shaft

26

. A part of the actuator

29

is located closer to the optical axis than a periphery of the compensating lens G. The cylindrical holder

27

has portions

27

a

and

27

b

which restrict the rotation range of the lens frame

25

. L represents a master lens.

When photographing is not performed, the lens frame

25

is pressed against a rotation range restricting portion

27

a

by a pushing spring

30

, and in photographing, it moved to a predetermined position by the actuator

29

. The driving force of the actuator

29

is transmitted to the lens frame

25

by an engagement between a pinion

31

and the driven portion

25

a

of the lens frame

25

. The engagement between an inner diameter portion

25

b

of the driven portion of the lens frame

25

and the pinion

31

may be made by a gear connection as shown in

FIG. 19

or by a frictional force between the inner diameter portion

25

b

of the driven portion of the lens frame

25

and the pinion

31

as shown in FIG.

20

.

FIGS. 21

to

23

show the structure of the driver of this embodiment using two lens driving units of the above-described structure. As shown in these figures, the two driving units

32

a

and

32

b

are arranged so that their movement directions are substantially orthogonal to each other and that their lens centers are located one in front and the other in the rear so as to coincide with the optical axis when the lenses are placed in the center of the compensating lens movable range.

Moreover, the driver of this embodiment is assembled so that the movement direction of one of the compensating lenses G

1

and G

2

is vertical and the movement of the other is horizontal when the camera is laterally held. In the mechanism, the movement direction of a weighty lens unit, i.e. a lens unit having a great moment of inertia is set to be vertical in order to reduce power consumption.

As a structure slightly different from the above-described structure, for example as shown in

FIG. 24

, a structure is practicable which employs an engagement between an actuator

29

′ and a driven portion of a lens frame driven thereby.

A calculation for controlling the positions of the compensating lenses will be described.

FIG. 25A

shows the front compensating lens G

1

located at its basic position and

FIG. 25B

shows the rear compensating lens G

2

located at its basic position. Under this condition, a straight line between a central axis

33

a

of the front compensating lens G

1

and its rotational shaft

26

a

and a straight line connecting between a central axis

33

b

of the rear compensating lens G

2

and its rotational shaft

26

b

coincide with the horizonal axis and the vertical axis of the camera, respectively. While the compensating lenses G

1

and G

2

are separately shown in

FIGS. 25A and 25B

, the central axes

33

a

and

33

b

of the compensating lenses G

1

and G

2

coincide with each other.

The positions of the central axis

33

a

of the front compensating lens G

1

and the central axis

33

b

of the rear compensating lens G

2

under this condition are defined as (0, 0) and (0, 0), respectively, by using coordinates on the horizonal and vertical axes of the camera. The rotational shafts

26

a

and

26

b

and the central axes

33

a

and

33

b

of the compensating lenses G

1

and G

2

are parallel to each other, and the compensating lenses G

1

and G

2

move within surfaces parallel to each other, i.e. vertical to the central axes

33

a

and

33

b.

To the lens frames

25

a

and

25

b

of the lenses G

1

and G

2

, light emitting diodes (LEDs)

34

a

and

34

b

are attached. PSD

35

a

and

35

b

for monitoring the present positions of the LEDs

34

a

and

34

b

are fixed to the cylindrical holder

27

so that their light receiving surfaces are parallel to the surface within which the compensating lenses G

1

and G

2

move.

When the compensating lenses G

1

and G

2

are located at their basic positions, straight lines between the rotational shafts

26

a

and

26

b

of the lenses G

1

and G

2

and the centers of the LEDs

34

a

and

34

b

, and detection directions of the PSDs

35

a

and

35

b

are vertical to each other. The irradiation light of the LEDs

34

a

and

34

b

are projected to the centers of the PSDs

35

a

and

35

b

. The light projection positions of the LEDs

34

a

and

34

b

to the PSDs

35

a

and

35

b

at this time are defined as 0 by using a coordinate parallel to the detection directions of the PSDs

35

a

and

35

b.

The distances from the rotational shafts

26

a

and

26

b

of the compensating lenses G

1

and G

2

to the central axes

33

a

and

33

b

thereof are r

1

and r

2

, respectively. The distances from the rotational shafts

26

a

and

26

b

of the compensating lenses G

1

and G

2

to the centers of the PSDs

35

a

and

35

b

are e

1

and e

2

, respectively.

FIGS. 26A and 26B

show a condition where the front compensating lens G

1

has rotated counterclockwise by θ and a condition where the rear compensating lens G

2

has rotated counterclockwise by δ, respectively. A position (x

1

, y

1

) of the central axis

33

a

of the front compensating lens G

1

and a position (x

2

, y

2

) of the central axis

33

b

of the rear compensating lens G

2

under this condition are respectively

x

1

=r

1

×(cos θ−1),

y

1

=r×

1

sin θ

x

2

=−r

2

×sin δ,

y

2

=r

2

×(cos δ−1)

where −45°<θ<45°, −45°<δ<45°.

A light projection position p

1

on the PSD

35

a

of the LED

34

a

of the compensating lens G

1

and a light projection position p

2

on the PSD

35

b

of the LED

34

b

of the compensating lens G

2

are respectively

p_{1} = e_{1} \times \sin θ = \frac{e_{1}}{r_{1}} \times y_{1}

p_{2} = e_{2} \times \sin δ = - \frac{e_{2}}{r_{2}} \times x_{2}

where the positive directions of θ and p

1

are the same and the positive directions of δ and p

2

are the same.

The compensation of the subject image by the compensating lenses G

1

and G

2

is made by bending the optical axis by θ

x

around the vertical axis and by θ

y

around the horizontal axis. The compensation angles θ

x

and θ

y

are expressed by the following expression by using the position (x

1

, y

1

) of the central axis

33

a

of the compensating lens G

1

and the position (x

2

, y

2

) of the central axis

33

b

of the compensating lens G

2

:

\begin{matrix} θ_{x} = k_{1} \times x_{1} + k_{2} \times x_{2} \\ = k_{1} \times r_{1} \times (\cos θ - 1) - k_{2} \times r_{2} \times \sin δ \end{matrix}

\begin{matrix} θ_{y} = k_{1} \times y_{1} + k_{2} \times y_{2} \\ = k_{1} \times r_{1} \times \sin θ + k_{2} \times r_{2} \times (\cos δ - 1) \end{matrix}

By modifying the above expressions, an expression 26 shown below is obtained. Thus, from the compensation angles θ

x

and θ

y

, the position (x

1

, y

1

) of the central axis

33

a

of the compensating lens G

1

and the position (x

2

, y

2

) of the central axis

33

b

of the compensating lens G

2

are calculated

\begin{matrix} k_{1} \times x_{1} = \frac{θ_{x} - K_{1}}{2} + \frac{θ_{x} + K_{1}}{2} \times B - (θ_{y} + K_{2}) \times A k_{1} \times y_{1} = \frac{θ_{y} + K_{2}}{2} + \frac{θ_{y} + K_{2}}{2} \times B + (θ_{x} + K_{1}) \times A k_{2} \times x_{2} = \frac{θ_{x} + K_{1}}{2} - \frac{θ_{x} + K_{1}}{2} \times B + (θ_{y} + K_{2}) \times A k_{2} \times y_{2} = \frac{θ_{y} - K_{2}}{2} - \frac{θ_{y} + K_{2}}{2} \times B - (θ_{y} + K_{1}) \times A where K_{1} = k_{1} \times r_{1}, K_{2} = k_{2} \times r_{2} B = \frac{K_{1}^{2} - K_{2}^{2}}{{(θ_{x} + K_{1})}^{2} + {(θ_{y} + K_{2})}^{2}} A = \frac{1}{2} \times \sqrt{- 1 + \frac{2 \times (K_{1}^{2} + K_{2}^{2})}{{(θ_{x} + K_{1})}^{2} + {(θ_{y} + K_{2})}^{2}} - B^{2}} & Expression 26 \end{matrix}

Now, an approximation for simplifying these expressions will be considered. The higher the order of the approximate expression is, the higher the calculation precision is, but the calculation amount increases. In this case, a fourth-order approximate expression shown by an expression 27 shown below is used to truncate the terms below the fourth order, and by an expression 28 shown below, the values of x

1

, y

1

, x

2

and y

2

are obtained. The projection positions p

1

and p

2

of the LEDs

34

a

and

34

b

on the PSDs

35

a

and

35

b

corresponding to the approximation values take values shown by an expression 29 shown below.

\begin{matrix} \frac{1}{1 + a} \approx 1 - a + a^{2} - a^{3} \sqrt{1 + a} \approx 1 + \frac{a}{2} - \frac{a^{2}}{8} + \frac{a^{3}}{16} & Expression 27 \end{matrix}

\begin{matrix} x_{1} \approx - \frac{θ_{y}}{2 \times k_{1} \times K_{1}} \times (θ_{y} + \frac{θ_{x}^{2}}{K_{2}}) y_{1} \approx \frac{θ_{y}}{k_{1}} + \frac{θ_{x}}{2 \times k_{1} \times K_{2}} \times (θ_{x} + \frac{θ_{y}^{2}}{K_{1}}) x_{2} \approx \frac{θ_{x}}{k_{2}} + \frac{θ_{y}}{2 \times k_{2} \times K_{1}} \times (θ_{y} + \frac{θ_{x}^{2}}{K_{2}}) y_{2} \approx - \frac{θ_{x}}{2 \times k_{2} \times K_{2}} \times (θ_{x} + \frac{θ_{y}^{2}}{K_{1}}) & Expression 28 \end{matrix}

\begin{matrix} p_{1} \approx \frac{e_{1} \times θ_{y}}{K_{1}} + \frac{e_{1} \times θ_{x}}{2 \times K_{1} \times K_{2}} \times (θ_{x} + \frac{θ_{y}^{2}}{K_{1}}) p_{2} \approx - \frac{e_{2} \times θ_{x}}{K_{2}} - \frac{e_{2} \times θ_{y}}{2 \times K_{1} \times K_{2}} \times (θ_{y} + \frac{θ_{x}^{2}}{K_{2}}) & Expression 29 \end{matrix}

That is, by controlling the lens positions so that the light projection positions of the LEDs

34

a

and

34

b

of the compensating lenses G

1

and G

2

on the PSDs

35

a

and

35

b

are the positions p

1

and p

2

, the optical axis can be bent substantially by θ

x

around the vertical axis and by θ

y

around the horizontal axis.

For example, when r

1

=r

2

=20 mm, k

1

=−k

2

=k, θ

x

=±2×k and θ

y

=±2×k, the positions of the central axes of the lenses G

1

and G

2

are as shown in Table 1. On the contrary, when the light projection positions p

1

and p

2

of the LEDs on the PSDs are controlled by using the above approximate expression (expression 27), the positions are as shown in Table 2. Moreover, when r

1

=r

2

=20 mm, k

1

=−k

2

/2=k, θ

x

=±4×k and θ

y

=±2×k, the positions of the central axes of the lenses G

1

and G

2

are as shown in Table 3. On the contrary, when the light projection positions p

1

and p

2

of the LEDs on the PSDs are controlled by using the above approximate expression (expression 27), the positions are as shown in Table 4.

That is, when r

1

=r

2

=20 mm, if approximately x

2

=±2 mm and y

1

+2 mm, x

2

and y

1

are controlled with a precision of approximately 10 μm by a third-order approximate expression and the error is approximately ± several millimeters. When a higher precision is necessary, a higher-order approximate expression is used. When the precision is not very important, a first- or second-order approximate expression is used. Thus, by using an approximate expression of an order in accordance with the required control precision of the compensating lenses, the positions of the compensating lenses can be controlled with a comparatively small amount of calculation.

Subsequently, the control of the positions of the compensating lenses will be described. Referring to

FIG. 27

, there is shown the circuit arrangement of a control block serving as the compensating lens driving circuit (block diagram only in one direction is shown). The drive control block drives the compensating lenses G

1

and G

2

in correspondence with the image shift amount calculated in the microcomputer

4

. The target position data of the compensating lenses G

1

and G

2

calculated in the microcomputer

4

and converted by the coefficient converter

19

are inputted to a drive control servo circuit through a D/A converter

36

(x

t

). In actuality, the position information of the compensating lenses G

1

and G

2

is outputted within the circuit by a subsequently-described PSD.

In

FIG. 27

, the position information of x

t

and x

n

is differentiated by differentiators

37

and

38

to output a target speed v

t

and an actual drive speed v

n

. Gains A

p1

and A

v1

optimum for a position deviation δx=x

t

−x

n

and a speed deviation δv=v

t

−v

n

are set by gain generators

39

and

40

to output a motor control voltage E=Ep+Ev. The optimum gains A

p1

and A

v1

are, for example, gains at which responsibility is excellent and oscillation barely occurs.

By voltage-to-current converting the outputs of the gain generators

39

and

40

by a motor driver

41

to conduct the motor, a motor

42

is driven, so that a driving force T is generated to an inertia J

43

. By the driving force T, the motor

42

is angularly accelerated at dw/dt, so that the lenses receive the acceleration a through a transmission system

44

to move. By taking out the motion information as a position signal by a PSD

35

, a feedback control is performed so that the deviations of x

t

and x

n

are eliminated.

Referring to

FIG. 28

, there is shown the structure of a position detecting sensor P. The position detecting sensor P includes an LED

34

, a slit

45

and the PSD

35

. For simplicity, the structure shown in

FIG. 28

slightly differs the actual structure of this embodiment. First, the drive principle of the driving system will be described. The actuator

29

is of a moving coil type and generates a driving force of F=ηIBL (efficiency coefficient of η≈0.9) based on a current I to be flowed through a coil

46

, a magnetic flux density B of a magnetic circuit

47

and an active coil length L of the coil.

By the driving force, the compensating lens G moves upward and downward along guiding shafts

48

and

49

. The LED

34

and the slit

45

are integrally provided above the compensating lens G to move in correspondence with the movement of the compensating lens G. That is, light discharged from the slit

45

is the same as the movement of the compensating lens G. The movement of the light is detected by the fixed PSD.

Referring now to

FIG. 29

, the principle of position detection of a PSD for one-dimensional detection will be described. When the center of gravity of light is located at a position which is a distance x away from the center of the PSD to the right, the relationship among an output current obtained from each pole and x and L are as shown by an expression 30 shown below. The relationship of an expression 31 shown below can be obtained by using voltages V

x1

and V

x2

obtained by current-to-voltage converting the current.

\begin{matrix} \frac{X_{2} - X_{1}}{X_{1} + X_{2}} = \frac{2 x}{L} & Expression 30 \end{matrix}

where L is the area of the light receiving surface of the PSD, x is the position of the center of gravity of light irradiated thereto, and X

1

and X

2

are output signals (photocurrents) obtained from the poles.

\begin{matrix} \frac{V_{x2} - V_{x1}}{V_{x1} + V_{x2}} = \frac{2 x}{L} V_{x1} = {RX}_{1} V_{x2} = {RX}_{2} & Expression 31 \end{matrix}

where R is a current-to-voltage converting resistor.

\begin{matrix} When V_{x1} + V_{x2} is fixed, V_{x2} - V_{x1} \propto x & Expression 32 \end{matrix}

Subsequently, a drive control integrated circuit (IC) shown in

FIG. 30

will be described. The target specifications provided to the drive control system are shown in Table 5. These are target specifications provided for the drive control system in order to curb the total image shift on the film surface within a range of 60 μm as an image shift compensating camera.

In Table 5, the total movement amount ±2 mm (on the optical axis) is a movement amount necessary to compensate for an image shift of ±1.5°. That is, when the lens is moved by 2 mm, the light passing through the compensating lens is bent at 1.5°. The reason why the axial total movement amount ±2 mm (a range of 4 mm) is different from the total movement amount ±1 (a range of 2 mm) of the position detector is that this embodiment employs the structure of the compensating lens portion shown in FIG.

31

.

Referring to

FIG. 31

, the advantage will be described. First, the drive principle of the compensating lens G will be described. The compensating lens G is a lens for compensating for image shifts in the left and right direction. Its movements are shown by a straight line figure and a dash and dotted line figure. To give the priority to moving the lens smoothly, the compensating lens is moved about the center of rotation of the lens. The center of optical axis is a distance

2

R away from the center of rotation of the lens. The position detector is a distance R away from the center of rotation of the lens. In this arrangement, when the compensating lens G is rotated at an angle

6

, the relationship between the movement amount x

1

about the center of optical axis and the movement amount x

2

of the position detector is x

1

=2Rtan θ and x

2

=Rtan θ. Therefore, x

2

=(½)x

1

.

Therefore, even when an image shift of ±1.5° on the optical axis (±2 mm in movement amount) is compensated for, the movement amount at the position detector P is ±1 mm. To detect the position of ±1 mm (a range of 2 mm), even if the attachment precision of the position detector P is considered, a sufficient detection range is obtained by using a PSD of approximately 1×3.5 mm size.

Returning to

FIG. 30

, the role of each block of the drive control IC will be described. As the drive control IC, a control IC for G

1

lens compensation and a control IC for G

2

lens compensation are provided on the same chip. A motor drive IC

50

is arranged as a device which supplies a voltage corresponding to a gain calculated by the microcomputer

4

as a side which supplies an input signal. The motor drive IC

50

rotates driving motors Ma and Mb to drive the compensating lenses G

1

and G

2

.

The upper part of the drive control IC is for G

1

lens compensation and the lower part thereof is for G

2

lens compensation. Since they have substantially the same arrangement, the lower part for G

2

lens compensation will mainly be described. In

FIG. 30

, the elements of the upper part for G

1

lens compensation the same as those of the lower part are denoted by the same reference numerals with the last letters changed from b to a.

First, the position detector will be described. Reference designation

51

b

represents an LED for a position sensor which emits light to a PSD

52

b

through a slit. The PSD

52

b

receives the light and has a reverse bias voltage applied thereto by a PSD reverse bias generator

53

b.

I

b1

and I

b2

are currents corresponding to the position of the lens G

2

and converted into voltages V

b1

and V

b2

by I/V converters

54

b

and

55

b

. Then, to make V

b1

+V

b2

constant, a signal of V

b1

+V

b2

is obtained at an adder

56

b

. Reference designation

57

b

represents an LED current control block which, when V

bt

=V

b1

+V

b2

changes from a predetermined value, e.g. −5V, a current corresponding to the difference is added to or subtracted from an LED drive current to maintain constant the light quantity of the LED

51

b

, i.e. v

bt

.

G

b

{v

bt

−(−5)}+I

e0

=I

e

where I

e0

is a reference current for driving the LED so that the sum total of the photocurrents of the PSD is −5V, I

e

is an actual drive current of the LED, and G

b

is a gain for a voltage difference (photocurrent difference).

At a subtracter

58

b

, since V

bt

is constant, V

b1

−V

b2

is obtained by using the principle of the above expression 31 to obtain position information of the lens G

2

. Reference designation

59

b

represents a voltage (V/V) converter to obtain the correspondence between a target position voltage signal V

bxt

and an actual lens position voltage signal V

bxn

. The signal becomes (the lens G

2

position information−Vbxn) after having passed through the V/V converter

59

b.

Subsequently, the output of the target position signal will be described. The target position signal from the microcomputer

4

is transmitted onto the circuit as a target position signal of V

axt

and V

bxt

through a D/A converter

60

, sample and hold (S/H) circuits

61

a

and

61

b

, and low pass filters (LPF)

62

a

and

62

b.

Confirming now the general flow of the image shift compensation of this embodiment, as shown in the flowchart of

FIG. 32

, the cycle of one image shift compensation is such that after the image shift compensation operation is started at step #

105

, the image shift amount is detected at step #

110

and an estimation drive calculation is performed at step #

115

. Then, target compensation positions axt and bxt of the compensating lenses G

1

and G

2

are obtained at step #

120

and the result is outputted to the drive control circuit at step #

125

.

In this sequence, in order to realize an image shift compensation, a method is easily considered to provide two D/A converters in the image shift drive control circuit to hold its output until the next detection result is calculated. However, in order to detect with a precision of {fraction (1/1000)} within a total movement amount of 4 mm (2 mm in the position detector) to realize a position control precision of {fraction (1/250)}, an output precision of {fraction (1/250)} is insufficient.

Consequently, a precision of at least {fraction (1/1000)} is required as the precision of the output D/A converter. The D/A converter satisfying this precision is at least a type having 10 bit precision and increasing monotonously or a type having at least 12 bit precision. In this case, a D/A converter of a precision of at least 10 bit is high in cost and great in area when manufactured in integrated circuit form. Such a problem is solved by the arrangement of this embodiment which uses S/H circuits.

The sequence is shown in the flowchart of FIG.

33

. This describes the details of the target position signal output of FIG.

32

. When this subroutine is called at step #

205

, the S/H circuit

61

a

is set in sampling state to output data to the compensating lens G

1

.

Then, at step #

215

, a digital signal of axt representative of a target position of the compensating lens G

1

is outputted to the D/A converter. Consequently, after passing through the

10

bit D/A converter

60

, the signal is transmitted to the LPF

62

a

by way of the S/H circuit

61

a

and has its high-frequency component cut. Then, the signal is outputted to G

1

motor gain setting portions

63

a

to

67

a

as a voltage signal representative of a target position of Vaxt.

At step #

220

, the process waits until the output of the S/H circuit

61

a

to correctly set the axt signal is stabilized. Then, at step #

225

, the S/H circuit

61

a

is set in hold state. Thereby, the target position signal is maintained Vaxt until the S/H circuit

61

a

is set in sampling state next time. Then, a target position data Vbxt is outputted to the compensating lens G

2

.

At steps #

230

to #

245

, the operations of steps #

210

to #

225

are executed for the S/H circuit

61

b

and LPF

62

b

for the lens G

2

. Consequently, the target position data Vbxt is outputted to G

2

motor gain setting portions

63

b

to

67

b

to maintain Vbxt until the next target position is calculated.

Subsequently, the gain setting circuit will be described. For the target position signal voltage Vbxt and the present position signal voltage Vbxn, the position deviation is calculated by an adding circuit

65

b

to set an optimum gain Gbx by a V/V circuit

66

b

. The target position signal passing through a differentiating circuit

63

b

is converted into a target speed signal Vvtb, and at a differentiating circuit

64

b

, −Vvnb is outputted. The optimum gain Gbv is set to a speed difference signal by the V/V circuit

67

b

to be transmitted to the drive IC

50

as a voltage signal. The optimum gains Gbx and Gbv as referred to here are gains which responds at a high speed with a response delay within the target specification and does not oscillate even if disturbance intrudes.

Compensation coefficients K

M

and L for compensating for the change with time will be described. The compensation coefficients are conversion coefficients to control the lens to move to a correct position by automatically compensating for the target position signal even when there is a difference between the target position signal and the actual position of the compensating lens because of a mechanical change with time or a change of a circuit signal with time.

Referring to

FIG. 34

, there is shown a relationship between the compensation coefficients of the compensating lens G

1

. e

Pa

and e

Ma

are distances from the center of the compensation range to a mechanical stop position. In this case, e

Pa

=+2 mm, e

Ma

=−2 mm. a

xt1

is outputted from the D/A converter when the lens reaches the mechanical position e

Pa

on the + side. A target position signal a

xt2

is a target position signal when the lens reaches the mechanical position e

Ma

on a − side. From the relationship between e

Pa

, e

Ma

, a

xt1

and a

xt2

, an actual position a

xn

to which the lens is desirably moved and a signal a

xt

outputted from the D/A converter as the target position can be expressed by the following expression 33:

\begin{matrix} L_{a} = \frac{a_{xt1} + a_{xt2}}{2} K_{Ma} = \frac{a_{xt1} - a_{xt2}}{e_{Pa} - e_{Ma}} a_{xt} = K_{Ma} \times a_{xn} + L_{a} & Expression 33 \end{matrix}

How the change-with-time compensation coefficients are obtained will specifically be described.

FIGS. 35 and 36

show the general flows. In this embodiment, the change-with-time compensation coefficients are obtained (1) when the power of the camera is activated (step #

305

) and (2) when the power is activated after the camera has been in sleep state (a state where the camera sleeps due to low power consumption) for a predetermined period of time (step #

310

).

After the power activation of (1) or (2), the microcomputer

4

is activated, and the process waits until the output of the microcomputer

4

is stabilized (steps #

315

and #

320

). Then, with the lens maintained stopped, the process wait until the output voltage of the PSD (actually, the output of the A/D converter) is substantially stabilized (step #

325

). Thereafter, the change-with-time compensation coefficients of the compensating lenses G

1

and G

2

are measured (steps #

330

and #

335

).

Referring to

FIG. 37

, there is specifically shown how the change-with-time coefficient of the compensating lens G

1

is obtained. When this subroutine is called, first, a sequence is executed to obtain the end in the + direction. a

xt1

is set to 1 mm and the value is outputted from the D/A converter (steps #

405

and #

410

). After the process waits 10 ms until the actually-driven lens reaches the target position, an actual position signal is inputted from the A/D converter (steps #

415

and #

420

).

This value is inputted to a variable m and compared with an A/D value n obtained when the previous target position (in this case, a value smaller by {fraction (1/1000)} mm) is outputted (steps #

425

and #

430

). When the present value m equals the previous value n, the position is determined to be end and the process proceeds to the detection of the end on the − side. When the values m and n differ from each other, since the lens has not reached the end, m is substituted for n and a sequence is repeated where a signal obtained by adding {fraction (1/1000)} mm to a

xt1

is used as the output of the D/A circuit (steps #

435

and #

440

).

Subsequently, a processing at the − side end performed after the lens reaches the + side end will be described. Basically, this process is the same as that of the case of the + side. −1 mm is substituted for a

xt2

, and a

xt2

is outputted from the D/A converter (steps #

445

and #

450

). After the process waits 10 ms, the a

xn2

signal is inputted from the A/D converter. This signal is substituted for m, and when it equals the value n of the previous target position signal, it is determined that the lens has reaches the end. When they are not equal to each other, since the lens has not reached the end yet, a loop where (a

xt2

−{fraction (1/1000)}) is substituted for the a

xt2

signal and outputted from the D/A converter (steps #

455

to #

480

). After the lens has reached the − side end, calculations of steps #

485

and #

490

are performed to obtain L

a

and K

Ma

shown by the above expression 33. Finishing this processing, the process returns at step #

495

.

FIG. 38

shows in detail how the change-with-time compensation coefficient of the compensating lens G

2

is obtained. This processing will not be described since its flow is substantially the same as that of the compensating lens G

1

shown in FIG.

37

. Through this subroutine, L

b

and K

Mb

are calculated as shown by the following expression 34:

\begin{matrix} L_{b} = \frac{b_{xt1} + b_{xt2}}{2} K_{Mb} = \frac{b_{xt1} - b_{xt2}}{e_{Pb} - e_{Mb}} b_{xt} = K_{Mb} \times b_{xn} + L_{b} & Expression 34 \end{matrix}

As described above, according to the present invention, by using the direction of luminance variation of a subject image formed on the light receiving surface of an image sensing device in addition to the conventionally-used contrast thereof in order to estimate the detection precision of the solid-state image sensing device which detects an image shift, more correct estimation of detection precision is enabled. By detecting an image shift by using only a partial area of the solid-state image sensing device where the detection precision is estimated to be sufficiently high, more precise image shift detection is enabled with a smaller amount of calculation.

Obviously, many modifications and variations of the present invention are possible in light of the above teachings. It is therefore to be understood that within the scope of the appended claims, the invention may be practiced other than as specifically described.

TABLE 1

(θ

x

, θ

y

)

(x

1

, y

1

)

(x

2

, y

2

)

(2xk, 2xk)

(−0.0896, 1.8905)

(−2.0896, −0.1095)

(2xk, −2xk)

(−0.1118, −2.1118)

(−2.1118, −0.1118)

(−2xk, 2xk)

(−0.0913, 1.9087)

(1.9087, −0.0913)

(−2xk, −2xk)

(−0.1095, −2.0896)

(1.8905 −0.0896)

Unit: mm

TABLE 4

Input

Output

(θ

x

, θ

y

)

(x

1

, y

1

)

(x

2

, y

2

)

(θ

x

, θ

y

)

(4xk, 2xk)

(−0.0803, 1.79)

(−2.04, −0.1043)

(3.9997xk,

1.9986xk)

(4xk, −2xk)

(−0.1225, −2.21)

(−2.06, −0.1064)

(3.9975xk,

−1.9973xk)

(−4xk, 2xk)

(−0.0821, 1.81)

(1.96, −0.0963)

(−4.0021xk,

2.0025xk)

(−4xk, −2xk)

(−0.1203, −2.19)

(1.94, −0.0943)

(−4.0003xk,

−2.0014xk)

Unit: mm

Number	Name	Date	Kind
5012270	Sekine et al.	Apr 1991	A
5053876	Blissett et al.	Oct 1991	A
5107293	Sekine et al.	Apr 1992	A
5130735	Kusaka et al.	Jul 1992	A
5157732	Ishii et al.	Oct 1992	A
5184216	Kobayashi	Feb 1993	A
5204741	Sakaguchi	Apr 1993	A
5218442	Hamada et al.	Jun 1993	A
5291300	Ueda	Mar 1994	A
5313296	Ohki	May 1994	A
5317685	Morimura et al.	May 1994	A
5365304	Hamada et al.	Nov 1994	A
5371539	Okino et al.	Dec 1994	A
5386264	Sekine et al.	Jan 1995	A
5563652	Toba et al.	Oct 1996	A
5581309	Okino et al.	Dec 1996	A
5627586	Yamasaki	May 1997	A

Number	Date	Country
59-28606	Feb 1984	JP
63-166369	Jul 1988	JP

Camera having an image shift detecting function

Information

Patent Number

Date Filed

Date Issued

Inventors

Original Assignees

Examiners

Agents

CPC

US Classifications

Field of Search

US

International Classifications

Abstract

Description

Claims

Priority Claims (1)

US Referenced Citations (17)

Foreign Referenced Citations (2)