The present invention relates to an image signal processing device and image signal processing method for eliminating image quality reduction due to shortage of gradation, in a display device for displaying a digital input image signal with multiple gradations.
When a display device where the number of displayable gradations is small performs multiple gradation display, smooth variation of brightness cannot be displayed. The brightness varies like a step, a contour-like pattern appears, and the image quality reduces.
For example, a display device such as a plasma display panel (PDP) or a digital mirror device (DMD) displays multiple gradations by dividing one field to a plurality of subfields having different luminance weight and combining light emission and non-emission in each subfield. It is known that specific noise called a motion picture pseudo-outline occurs in a motion picture part in this display device. For preventing the occurrence, the image display may be performed using not the gradations where the motion picture pseudo-outline is apt to occur but the gradations (hereinafter referred to as “available gradations”) where the motion picture pseudo-outline hardly occurs. Here, an error diffusing method can be used for artificially displaying all gradations using available gradations. For example, Japanese Patent Unexamined Publication No. 2000-276100 discloses a method having a gradation converting means for converting a gradation to be displayed to an available gradation. In this method, when the gradation to be displayed is not available, this gradation is converted to an available gradation closest to itself, the error occurring at this time is diffused to surrounding pixels and summed up. Thus, the gradation to be displayed is artificially displayed using only the available gradation.
It is advantageous from the aspect of the cost and mounting that a memory element such as a static random access memory (SRAM) is used as a gradation converting means. However, the access time of the SRAM is relatively slow, and an error cannot be diffused to the immediate right pixel when the access time exceeds one-data period. The SRAM cannot be therefore employed as the gradation converting means. Here, the one-data period is about 41 ns when one frame is 852×480 pixels, for example.
For addressing this problem, Japanese Patent Unexamined Publication No. 2000-276100 also discloses a method of using a lower bit of an image signal as a temporary error and diffusing the error to the immediate right pixel.
However, the error diffused to the immediate right pixel is not an accurate value in this method, so that the image quality disadvantageously reduces comparing with a case of correct error diffusion processing.
The present invention addresses the problems discussed above. The present invention provides an image signal processing device and image signal processing method that can perform error diffusion processing of accurately diffusing an error to the immediate right pixel even when the access time of a means for converting the gradation exceeds one-data period.
The present invention provides an image signal processing device that artificially displays M gradations by converting an input image signal of M gradations to an output image signal of N (N<M) gradations and by diffusing an error generated by the conversion to pixels around a corresponding pixel. The image signal processing device has the following elements:
An image signal processing device in accordance with an exemplary embodiment of the present invention will be described hereinafter with reference to the following drawings.
A concept of an image signal processing method of the present invention is firstly described.
About one-line period is allowed until respective display errors of pixels P1, P2 and P3 before target pixel P0 by one line are determined, multiplied by the weighting factors, and added to target pixel P0. The one-line period is 41 ns×852 pixels=about 3.5 μs, and the access can be performed in good time even when an SRAM is employed as the gradation converting means. While, until the display error of pixel P4 just before target pixel P0 is determined, multiplied by the weighting factor, and added to target pixel P0, only one-data period is allowed.
In the image signal processing method of the present embodiment, the display errors of pixels having sufficient lead time, namely pixels P1, P2 and P3 before target pixel P0 by one line, are multiplied by weighting factors k1, k2 and k3, respectively, and the sum of the products is added to the input image signal of target pixel P0. The gradation of the resultant image signal is converted to an available gradation. At this time, however, the display error of immediate left pixel P4 is not determined, so that the gradation of the resultant image signal is converted to a plurality of available gradation candidates that are convertible when the error of pixel P4 is added. Then, the error generated in immediate left pixel P4 is multiplied by weighting factor k4, and the product is added to the image signal of target pixel P0 including the errors of pixels P1 to P3. A gradation closest to the gradation of the image signal is selected from the plurality of gradation candidates, and is output as the image signal of target pixel P0. The display error of the image signal is calculated and diffused to surrounding pixels P5 to P8.
Even when a memory element such as an SRAM of which access time exceeds the one-data period is employed as the gradation candidate converting means, the error diffusion processing of accurately diffusing the error to the immediate right pixel can be performed.
First adding circuit 101 as the first adding means sums the input image signal, the output of multiplying circuit 114, the output of multiplying circuit 113, and the output of multiplying circuit 112. When the input image signal is assumed to be an image signal corresponding to target pixel P0, the outputs of multiplying circuits 114, 113 and 112 are error signals from pixels P1, P2 and P3, respectively.
Gradation candidate converting means 102 supplies three gradation candidates shown in
Gradation candidate converting means 102 is formed of a memory element of slow access time, or is formed of an SRAM where access time is five times longer than the one-data period in the present embodiment. Delaying circuit 103 as a delaying means delays the output of first adding circuit 101 by the access time of gradation candidate converting means 102. The delaying time of delaying circuit 103 is equal to 5-data period in the present embodiment.
Second adding circuit 104 as the second adding means sums the image signal delayed by delaying circuit 103 and an output of multiplying circuit 111. The output of multiplying circuit 111 corresponds to the error of pixel P4, so that the errors of four adjacent pixels P1, P2, P3 and P4 are diffused to the output of second adding circuit 104.
Gradation selecting means 105 selects a gradation closest to the output of second adding circuit 104 from three gradation candidates supplied from gradation candidate converting means 102, and outputs it as the output image signal. Differencing circuit 106 as an error calculating means determines a display error, namely difference between the gradation of the image signal of which error is diffused by second adding circuit 104 and the gradation selected as the output image signal by gradation selecting means 105, and outputs the difference as a display error of pixel P0 to delaying circuits 110, 109, 108 and 107.
The data period over which input image signal S0 corresponding to target pixel P0 is input is set to be T0. As discussed above, about one-line period is allowed until errors diffused from pixels P1, P2 and P3 before pixel P0 by one line are determined, so that delay amounts of delaying circuits 110, 109 and 108 can be adjusted to make the output times of errors Δε1, Δε2 and Δε3 diffused to pixel P0 from multiplying circuits 114, 113 and 112 coincide with period T0. Signal S′0 to which errors Δε1, Δε2 and Δε3 are added at period T1 is supplied from first adding circuit 101.
Since the access time of gradation candidate converting means 102 is 5-data period, three gradation candidates Σ0a, Σ0b and Σ0c are output at period T5. At this time, delaying circuit 103 delays signal S′0 to period T5. When error Δε4 diffused from immediate left pixel P4 exists at period T5, output signal Σ0 and display error ε0 corresponding to target pixel P0 can be determined. The image signal of immediate left pixel P4 is input before target pixel P0 by one-clock period, so that display error ε4 corresponding to pixel P4 is previously determined at period T4 before period T5 by one-clock period. Therefore, delaying circuit 107 delays the signal by one-data period and multiplying circuit 111 multiplies display error ε4 by the weighting factor, thereby providing error Δε4 to be diffused from pixel P4 to pixel P0 at period T5.
In the present embodiment, even when the access time of gradation candidate converting means 102 exceeds the one-data period, accurate error diffusion processing can be performed. An SRAM disposed inside or outside a large scale integration (LSI) separated from a logic unit for executing the image signal processing can be therefore used as gradation candidate converting means 102.
In the present embodiment, as shown in
The case of having three gradation candidates has been described, but the number of gradation candidates can be set at an arbitrary number.
In the present embodiment, it is assumed that the processing times of first adding circuit 101, second adding circuit 104, multiplying circuits 111 to 114, differencing circuit 106, and gradation selecting means 105 can be neglected. However, when the processing times of these circuits are long to a certain extent, the above-mentioned effect can be obtained by adjusting the delay times of delaying circuits 103, 107, 108, 109 and 110.
In the image signal processing device and the image signal processing method of the present invention, the error diffusion processing of correctly diffusing an error to the immediate right pixel can be executed even when the access time of a gradation converting means exceeds the one-data period.
In the image signal processing device and the image signal processing method of the present invention, the error diffusion processing of correctly diffusing an error to the immediate right pixel can be executed even when the access time of a gradation converting means exceeds the one-data period. Therefore, the device and method are useful as an image signal processing device and image signal processing method for eliminating image quality reduction due to shortage of gradation, in a display device for displaying a digital input image signal with multiple gradations.
Number | Date | Country | Kind |
---|---|---|---|
2003-433153 | Dec 2003 | JP | national |
Filing Document | Filing Date | Country | Kind | 371c Date |
---|---|---|---|---|
PCT/JP04/19696 | 12/22/2004 | WO | 7/11/2005 |