Patent Application
20090010492
The present application claims priority under 35 U.S.C. § 19 to Japanese Patent Application No. 2007-175021, filed Jul. 3, 2007. The contents of this application are incorporated herein by reference in their entirety.
1. Field of the Invention
The present invention relates to an image recognition device, a focus adjustment device, an imaging apparatus, an image recognition method, and a focus adjustment method.
2. Discussion of the Background
There is known an image tracking device for storing an image of a target to be focused as a reference image (template image), detecting an image that matches the reference image from repeatedly photographed images through template matching process, and tracking the target (see, for example, Japanese Laid-Open Patent Publication No. 2006-058431).
When the background of the target to be focused is similar in color information with the target, however, there is a possibility other portions such as the background different from the target might be falsely recognized.
According to one aspect of the present invention, an image recognition device includes an imaging unit, a holding unit, a detection unit, a recognition unit, and a calculation unit. The imaging unit is configured to acquire first image information and second image information. The holding unit is configured to hold reference information that functions as a reference image. The detection unit is connected to the holding unit and is configured to detect a first difference between partial information of at least a part of the first image information and the reference information and to detect a second difference between partial information of at least a part of the second image information and the reference information. The recognition unit is connected to the holding unit and is configured to recognize a first area corresponding to the reference image in the first image information. The calculation unit is connected to the detection unit and is configured to calculate a determination value based on a reference area in the second image information corresponding to the first area by weighting the second difference. The recognition unit is configured to recognize a second area corresponding to the reference image in the second image information based on at least one of the second difference and the determination value.
According to another aspect of the present invention, a focus adjustment device in the image recognition device in which the imaging unit is configured to acquire the first image information and the second image information through an optical system includes a focus detection unit, a selection unit, and a focus adjustment unit. The focus detection unit is configured to detect a focus state of the optical system at a plurality of detection areas provided in a screen by the optical system. The selection unit is connected to the focus detection unit and configured to select at least one detection area from the plurality of detection areas based on a position of a partial region recognized by the recognition unit and the focus state detected by the focus detection unit. The focus adjustment unit is configured to perform focus adjustment of the optical system based on the focus state detected by the focus detection unit with respect to the at least one detection area selected by the selection unit.
According to yet another aspect of the present invention, an imaging apparatus includes the focus adjustment device.
According to the other aspect of the present invention, an image recognition method includes acquiring first image information and second image information, holding reference information that functions as a reference image, detecting a first difference between partial information of at least a part of the first image information and the reference information, recognizing a first area corresponding to the reference image in the first image information, detecting a second difference between partial information of at least a part of the second image information and the reference information, weighting the second difference based on a reference area in the second image information corresponding to the first area to calculate a determination value, and recognizing a second area corresponding to the reference image in the second image information based on at least one of the second difference or the determination value.
According to further aspect of the present invention, an image recognition method includes acquiring first image information and second image information, holding reference information that functions as a reference image, detecting a first difference between partial information of at least a part of the first image information and the reference information, recognizing a first area corresponding to the reference image in the first image information, detecting a second difference between partial information of at least a part of the second image information and the reference information, weighting the second difference based on a reference area in the second image information corresponding to the first area to calculate a determination value, and recognizing a second area corresponding to the reference image in the second image information based on at least one of the second difference or the determination value.
A more complete appreciation of the invention and many of the attendant advantages thereof will be readily obtained as the same becomes better understood by reference to the following detailed description when considered in connection with the accompanying drawings, wherein:
The embodiments will now be described with reference to the accompanying drawings, wherein like reference numerals designate corresponding or identical elements throughout the various drawings.
One embodiment of an imaging apparatus (single-lens reflex digital still camera) having an automatic focus adjustment (AF) function of detecting a focus adjustment state (defocus amount in the embodiment) of a photographing lens at a plurality of focus detection areas set within a photographing screen and focus driving the photographing lens based on the defocus amount of one of the areas, and an image tracking function of storing an image of a subject to be tracked in the photographing images as a template image (reference image) and tracking the subject to be tracked while searching for a position of the image same as or similar to the template image of the repeatedly photographed images (template matching), and being equipped with an image recognition device for tracking the target while driving the photographing lens with the AF function and the image tracking function will be described.
A focus detection optical system 9 and a distance measuring element 10 for detecting the focus adjustment state of the photographing lens 8 are arranged at the bottom of the camera body 2. In this embodiment, an example of adopting a focus detection method by an eye division phase difference detection system is shown. The focus detection optical system 9 guides a pair of focus detection light fluxes passed through the photographing lens 8 to a light receiving surface of the distance measuring element 10, and images a pair of optical images. The distance measuring element 10 includes a pair of CCD line sensors, and the like, and outputs a focus detection signal corresponding to the pair of optical images. The quick return mirror 5 and the sub-mirror 6 are set at a position in the photographing optical path as shown with a broken line before photographing, where the pair of focus detection light fluxes from the photographing lens 8 are transmitted through a half mirror part of the quick return mirror 5, reflected by the sub-mirror 6, and guided to the focus detection optical system 9 and the distance measuring element 10.
A finder optical system is arranged at the upper part of the camera body 2. The quick return mirror 5 and the sub-mirror 6 are at a position shown with a broken line before photographing, where the subject light from the photographing lens 8 is reflected by the quick return mirror 5 and guided to a focusing plate 11 so that the subject image is imaged on the focusing plate 11. A liquid crystal display element 12 displays information such as focus detection area mark in a superimposed manner on the subject image imaged on the focusing plate 11, and also displays various photographing information such as exposure value at a position outside the subject image. The subject image on the focusing plate 11 is guided to an eyepiece window 15 through a pentagonal roof prism 13 and an eyepiece lens 14, so that the photographer can visually recognize the subject image.
A second imaging element 16 for imaging the subject image for subject tracking or photometry is arranged in the finder optical system at the upper part of the camera body 2. The subject image imaged on the focusing plate 11 is re-imaged on a light receiving surface of the second imaging element 16 through the pentagonal roof prism 13, a prism 17, and an imaging lens 18. The second imaging element 16 outputs an image signal corresponding to the subject image. The subject image imaged on the focusing plate 11 before photographing is guided to the second imaging element 16 through the pentagonal roof prism 13, the prism 17, and the imaging lens 18, and the subject image is re-imaged on the light receiving surface of the second imaging element 16. The details will be described later, where the tracking control and the exposure calculation are performed based on the subject image imaged by the second imaging element 16.
A body drive control device 19, an operation member 20, and the like are arranged in the camera body 2. The body drive control device 19 is configured by a microcomputer, peripheral components such as memory and A/D converter, and the like to be described later in detail, and performs various controls and calculations of the camera 1. The operation member 20 includes switches and selectors for operating the camera 1 such as shutter button, focus detection area selection switch, photographing mode selection switch, and the like.
A zooming lens 8a, a focusing lens 8b, an aperture 21, a lens drive control device 22, and the like are arranged in the interchangeable lens 3. In the embodiment, the photographing lens 8 is representatively indicated with the zooming lens 8a, the focusing lens 8b, and the aperture 21, but the configuration of the photographing lens 8 is not limited to the configuration shown in
The body drive control device 19 and the lens drive control device 22 communicate through a contact point 23 of a lens mount unit, where information such as lens driving amount and aperture value are transmitted from the body drive control device 19 to the lens drive control device 22, and lens information and aperture information are transmitted from the lens drive control device 22 to the body drive control device 19.
The tracking control unit 19e stores the image corresponding to the tracking target position manually specified by the photographer or the tracking target position automatically set by the camera 1 of the subject image imaged by the second imaging element 16 as a template image (reference image) in the memory 19d, and searches the image region that matches or is similar to the template image from the subsequently and repeatedly photographed images to recognize the position of the target. The exposure calculation unit 19f calculates an exposure value based on the image signal imaged by the second imaging element 16.
The focus detection calculation unit 19g detects the focus adjustment state, or defocus amount herein, of the photographing lens 8 based on the focus detection signal corresponding to the pair of optical images output from the distance measuring element 10. The details will be described later, but a plurality of focus detection areas is set within the photographing screen of the photographing lens 8, the distance measuring element 10 outputs the focus detection signal corresponding to the pair of optical images for every focus detection area, and the focus detection calculation unit 19g detects the defocus amount based on the focus detection signal corresponding to the pair of optical images for every focus detection area. The lens driving amount calculation unit 19h converts the detected defocus amount to the lens driving amount.
The subject tracking operation according to one embodiment will now be described.
A plurality of focus detection areas is set in the photographing lens 8, where an area mark is superimposed on the subject image on the focusing plate 11 by the liquid crystal display element 12, and the position of each focus detection area is displayed. In one embodiment, an example where the focus detection areas 45a to 45g are set at seven locations within the photographing screen is shown, as shown in
As shown in
In step 1, an initial tracking image (firstly acquired image after the image tracking process is started) is acquired by the second imaging element 16. The initial tracking image is represented in RGB value for every pixel.
R[x,y], G[x,y], B[x,y] (1)
The color information RG, BG of each pixel are calculated based on the RGB value of every pixel, and the luminance information L is calculated from the exposure time T, the gain Gain, and the color combining coefficients Kr, Kg, and Kb for a case where the image is acquired.
RG[x,y]=Log2(R[x,y])−Log2(G[x,y]),
BG[x,y]=Log2(B[x,y])−Log2(G[x,y]),
L[x,y]=Log2(Kr×R[x,y]+Kg×G[x,y]+Kb×B[x,y])−Log2(T)−Log2(Gain) (2)
In the following step 2, the initial process of tracking control shown in
An example of determining the tracking subject region 47 based on the subject color information is shown, where the size of the tracking subject region may be unified such as 3×3 to simplify the process, or the size of the subject region may be determined according to the distance information of the photographing lens 8.
In step 104, the image of the tracking subject region 47 in the initial tracking image is stored in the memory 19d as a template image 48 (see
RGref[rx,ry]=RG[x,y],
BGref[rx,ry]=BG[x,y],
Lref[rx,ry]=L[x,y](rx,ry=1 to 3,x=4 to 6,y=5 to 7) (3)
Next, a region enlarged by a predetermined pixel (two pixels herein) towards the front, the back, the left, and the right with the tracking subject region 47 as the center is set as a search region 49 in step 105. In the example shown in
After the initial process of tracking control is terminated, the process proceeds to step 3 of
In step 5, the tracking calculation process shown in
When the starting point position of the search region 49 is (scx, scy)=(2, 3) as shown in
Diff[dx,dy]=ΣΣ{ABS(RG[scx+dx−1+rx,scy+dy−1+ry]−RGref[rx,ry])+ABS(BG[scx+dx−1+rx,scy+dy−1+ry]−BGref[rx,ry])+ABS(L[scx+dx−1+rx, scy+dy−1+ry]−Lref[rx, ry])} (4)
In equation (4), dx, dy=1 to 5, rx, ry=1 to 3, scx=2, scy=3, and ΣΣ is the summation calculation of rx=1 to 3 and ry=1 to 3.
The difference Diff calculated in step 202 is then weighted. In the embodiment, a weighting table Wt[dc, dy] for the number (twenty-five in the example) of differences Diff[dx, dy](dx, dy=1 to 5) is set in advance, and a value obtained by multiplying a weighting coefficient Kw to the weighting table Wt is set as the weight.
Wt[dx, dy]={(2, 2, 2, 2, 2)
Diff[dx,dy]=(Kw×Wt[dx,dy])+Diff[dx,dy](dx,dy=1 to 5) (5)
The weighting table Wt shown in equation (5) is a weighting pattern in which the weight on the difference Diff of the region cut out from the center of the search region 49 is the minimum and the weight on the difference Diff of the cutout region is larger towards the periphery of the search region 49, where the position of the focus detection area used to perform focus adjustment of the photographing lens 8 in the previous focus adjustment is prioritized by using such weighting table Wt. Normally, in the repeatedly performed automatic focus adjustment, the possibility of capturing the tracking target is assumed to be high in the focus detection area or the vicinity thereof used as the area for performing the focus adjustment of the photographing lens 8 in the previous focus adjustment, and thus the minimum weight is given to the difference Diff on the region cut out from the area position used in the previous focus adjustment.
When the AF mode of performing photographing while capturing the subject in the selected focus detection area such as a single AF mode in which the focus detection is performed in the focus detection area selected by the user through the focus detection area selection switch of the operation member 20 and the focus adjustment of the photographing lens 8 is performed based on the focus detection result, and a dynamic AF mode in which the focus detection is performed backed up by the focus detection result of other focus detection areas even when the subject is outside the selected focus detection area while performing the focus detection and the focus adjustment in the focus detection area selected by the user through the focus detection area selection switch of the operation member 20 to continuously track the moving subject is set, the minimum weight is given to the difference Diff on the region cut out from the position of the selected focus detection area. The tracking target thus can be reliably captured.
As a result of setting the weighting table Wt as shown in equation (5), even when the tracking target does not have color information with clear difference with respect to other portions such as background, a different portion such as the background different from the target is prevented from being falsely recognized and tracked as the target, the position of the image tracking result is prevented from moving to various positions such as the background and other different portions when the position relationship between the camera and the tracking target is not changed, and the tracking operation is stabilized.
The weighting coefficient Kw in equation (5) is the coefficient for adjusting the effect by the weighting table Wt. The value of the weighting coefficient Kw is set larger the smaller the difference between the color information of the tracking target and the color information of other portions such as the background to increase the effect by the weighting table Wt, so that the tracking target is properly captured and the tracking operation is stabilized. When the difference between the color information of the tracking target and the color information of other portions such as the background is relatively large, the value of the weighting coefficient Kw is set small to suppress unnecessary effect by the weighting table Wt.
In step 203, the minimum difference of the difference Diff′[dx, dy] applied with weight is searched, and the cutout region in a case where the minimum difference is calculated in the search region 49 is determined as the new tracking subject region 47. Here, as shown in
A process of updating the image information of the template image 48 using the image information newly determined for the subject region 47 may be applied. In this case, the recent image information is gradually updated on the image information of the template image by adding 20% of the image information of the new subject region 47 with respect to 80% of the image information of the original template image 48, for example, so that change in the tracking subject can be easily followed. In this case, update is not performed every time the tracking calculation is performed, and update may be performed only when smaller than a certain threshold value based on the magnitude of the difference Diff′ of the position determined as the tracking subject region 47.
In step 204, a region enlarged by a predetermined pixel (two pixels herein) towards the front, the back, the left, and the right with the new tracking subject region 47 as the center is set as the search region 49. As shown in
In step 6 of
In step 8, the area for performing focus adjustment is determined based on the new tracking subject region 47 the image tracking result in step 5 and the focus detection area of the area search result in steps 6 and 7. In the embodiment, the focus detection area corresponding to the new tracking subject region 47 of the image tracking result and the focus detection area of the area search result are compared, and a focus detection area common to the image tracking result and the area search result is determined as the area for performing focus adjustment (hereinafter, referred to as focus adjustment area). In step 9, the defocus amount detected in the focus adjustment area is converted to the lens driving amount, and the focusing lens 8b is driven by the lens drive control device 22 to perform focus adjustment.
The full-pushing of the release button may be checked in step 3 after the focus adjustment. The processes of steps 4 to 9 are repeatedly executed while the shutter button is half-pushed, and the process proceeds to step 10 to execute the photographing process when the shutter button is fully pushed.
Thus, according to the embodiment, the second imaging element 16 for imaging the image imaged by the photographing lens 8 and outputting the color information, and the memory 19d for storing the reference color information (color information of template image) related to the image of the tracking target are arranged, where the difference Diff between the color information of the image and the reference color information is calculated while shifting the comparison position of the image and the image of the target, and weighting is performed on each difference Diff to determine the position at where the weighted difference Diff′ is a minimum as the new position of the tracking target. Thus, even when the tracking target does not have color information of clear difference with respect to other portions such as background, a different portion such as the background different from the target is prevented from being falsely recognized and tracked as the target, the position of the image tracking result is prevented from moving to various positions such as the background and other different portions when the position relationship between the camera and the tracking target is not changed, the tracking target is correctly captured, and the tracking operation can be stabilized.
According to one embodiment, the weight of the difference Diff corresponding to the position determined as the position of the tracking target in the previous image tracking is set the minimum and the weight on the difference Diff is set larger the farther away from the position determined as the position of the tracking target in the previous image tracking, and thus the possibility of capturing the tracking target is high.
In the embodiment described above, an example of a weighting pattern that prioritizes the position of the focus detection area used to perform the focus adjustment in the previous focus adjustment has been shown, but the weighting pattern of the weighting table Wt is not limited to the pattern according to the embodiment described above.
Furthermore, in the embodiment described above, an example of weighting the difference Diff of the color information calculated within the search region has been described, but the difference Diff in color information with respect to the image of the entire range of the photographing screen may be obtained, and the weighting may be performed on the difference Diff calculated in the entire range of the photographing screen to set the new tracking subject region. In searching for the tracking subject region from the entire range of the photographing screen, the weight of the difference Diff corresponding to the center of the photographing screen of the photographing lens 8 is set the minimum, and the weight on the difference Diff is set larger the farther away from the center of the photographing screen of the photographing lens 8. Normally, the possibility of capturing the main subject at the center of the photographing screen is high, and thus the possibility of capturing the tracking target in the repeatedly performed image tracking is high.
Furthermore, the maximum value and the minimum value of the difference Diff in the color information may be extracted, and the weighting coefficient Kw may be set according to the difference between the maximum value and the minimum value so that the effect of the weighting table Wt is optimized. When the difference between the maximum value and the minimum value is small and the image in the search region or the photographing screen has a color close to uniformity and does not include a stand-out color, the value of the weighting coefficient Kw is set large to increase the effect by the weighting table Wt, thereby preventing false recognition as the background etc. and stabilizing the tracking operation. When the difference between the maximum value and the minimum value is relatively large and the difference between the color information of the tracking target and the color information of other portions such as the background is relatively large, the value of the weighting coefficient Kw is set small and unnecessary effect by the weighting table Wt is suppressed.
Furthermore, the weighting coefficient Wk may be set according to the photographing mode so that the effect of the weighting table Wt is optimized. For instance, in the photographing mode of tracking the subject that is moving fast such as in the sport mode or the pet mode, the value of the weighting coefficient Wk is set small to suppress unnecessary effect by the weighting table Wt so that the movement of the subject can be easily followed. When photographing the subject that does not move much such as in the portrait mode, the possibility of the subject suddenly moving far is low, and thus the weighting coefficient Kw is set large to increase the effect of he weighting table Wt so that switch from the tracking target to other portions such as the background is less likely to occur.
Obviously, numerous 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 otherwise than as specifically described herein.
| Number | Date | Country | Kind |
|---|---|---|---|
| 2007-175021 | Jul 2007 | JP | national |
