JP4180007B2 - Image signal processing circuit - Google Patents

Description

  The present invention relates to an image signal processing circuit, and more particularly to an image signal processing circuit which is applied to, for example, a digital camera and generates a color difference signal and a luminance signal based on a color signal.

An example of a conventional image signal processing circuit of this type is disclosed in Patent Document 1. According to this prior art, the RGB signal generated by the color interpolation unit is converted into a luminance signal (Y) and a color difference signal (Cr, Cb) by the YC conversion unit. The converted luminance / color difference signal (YCrCb) is subjected to contour correction of the luminance signal by the contour emphasizing unit, and then subjected to noise reduction such as smoothing processing and median filter processing in the noise reduction processing unit.
JP 2003-274245 A [H04N 5/225, 5/232]

    However, in general, the median filter is superior to other filters in terms of resolution preservation characteristics, but is inferior to other filters in terms of noise reduction characteristics. In addition, with respect to luminance, preservation of resolution is more important than noise reduction, while for color, reduction of noise is more important than preservation of resolution.

    As a result, there is a problem in that good color reproducibility cannot be obtained with the conventional technique in which the median filter processing is performed on the luminance / color difference signals. If a filter that is superior in noise reduction characteristics rather than resolution preservation characteristics is used instead of the median filter, good color reproducibility can be obtained. However, luminance reproducibility is reduced.

    Therefore, a main object of the present invention is to provide an image signal processing circuit excellent in color reproducibility and luminance reproducibility.

The image signal processing circuit according to the first aspect of the present invention is a first reduction means for outputting a first reduced color signal of one line from which a signal component having a frequency equal to or higher than the first frequency is removed based on a plurality of continuous lines of unique color signals, The second reduction means for outputting the second reduced color signal for one line from which the signal components of the second frequency and higher than the first frequency are removed based on the unique color signals for a plurality of lines are output from the first reduction means. a second generating means for generating a luminance signal based on the first creating means, and a second reduced color signal output from the second reduction means for generating a color difference signal based on the first reduced color signal, the first reduction The means includes a plurality of cyclic filters that respectively input a plurality of lines of unique color signals, a plurality of first coefficient creation means that respectively create a plurality of first weighting coefficients based on outputs of the plurality of cyclic filters, and a plurality of first coefficients. 1 person Corresponding to the second coefficient creating means for creating one second weighting coefficient based on the plurality of weighting coefficients created by the creating means, and the original color signal of the first specific line forming the plurality of lines and the first specific line Addition means for adding the output of the recursive filter according to the second weighting coefficient created by the second coefficient creation means .

  The first reduction means outputs a first reduced color signal of one line from which signal components of the first frequency or higher are removed based on the continuous original color signals of a plurality of lines, and the second reduction means outputs the same plurality of lines. Based on the unique color signal, one line of the second reduced color signal from which the signal component of the second frequency or higher has been removed is output. Here, the second frequency is higher than the first frequency. The first creating means creates a color difference signal based on the first reduced color signal output from the first reducing means, and the second creating means is based on the second reduced color signal output from the second reducing means. Create a luminance signal.

The color difference signal and the luminance signal are respectively created based on the outputs of the first reduction unit and the second reduction unit having different characteristics. The second weighting coefficient is determined based on the unique color signal in the horizontal direction and the vertical direction. The unique color signal of the first specific line and the output of the cyclic filter corresponding to the first specific line are added according to the second weighting coefficient thus determined.
Therefore, both color reproducibility and luminance reproducibility can be improved , and good filter processing is realized.

The image processing circuit according to the present invention of claim 2 depends on claim 1, each of the plurality of recursive filter is a first multiplying the first coefficient is a decimal to reduce color signal of the preceding pixel preceding the pixel of interest Multiplying means, second multiplying means for multiplying the unique color signal of the pixel of interest by a second coefficient which is the complement of “1” of the first coefficient, and adding the output of the first multiplying means and the output of the second multiplying means. Adding means for generating a low-frequency color signal of the target pixel, and the first reduction means sets the first coefficient to “0” when the level difference between the original color signal of the target pixel and the reduced color signal of the target pixel exceeds a threshold value. It further includes a changing means for changing to "".

  By changing the first coefficient to “0” when the level difference between the unique color signal corresponding to the target pixel and the reduced color signal is large, the level variation of the reduced color signal after the edge is prevented. This makes it possible to accurately determine the first weighting coefficient and thus the second weighting coefficient.

An image signal processing circuit according to a third aspect of the present invention is dependent on the first or second aspect , wherein the second reduction means includes a median filter that inputs a plurality of lines of unique color signals, and a second specific line that forms a plurality of lines. Selection means for selecting either the unique color signal or the output of the median filter corresponding to the second specific line is included.

An image signal processing circuit according to a fourth aspect of the invention is dependent on the third aspect , and the selection means selects the unique color signal when the level of the edge enhancement signal exceeds a threshold value. As a result, good filter processing is realized.

A digital camera according to a fifth aspect of the invention includes the image signal processing circuit according to any one of the first to fourth aspects.

  According to the present invention, since the color difference signal and the luminance signal are respectively generated based on the outputs of the first reduction unit and the second reduction unit having different characteristics, both the color reproducibility and the luminance reproducibility are improved. be able to.

  The above object, other objects, features and advantages of the present invention will become more apparent from the following detailed description of embodiments with reference to the drawings.

  Referring to FIG. 1, a digital camera 10 of this embodiment includes an image sensor 12. The optical image of the object scene is irradiated on the light receiving surface of the image sensor 12 through an optical lens (not shown). The light receiving surface is covered with a color filter (not shown) of a primary color Bayer array, and the charge generated by photoelectric conversion in each light receiving element is any of R (Red), G (Green), and B (Blue). Or one color information.

  When the power is turned on, the CPU 18 instructs the TG 16 to repeat pre-exposure and thinning-out reading. The TG 16 performs pre-exposure on the image sensor 12 at a rate of once every 1/30 seconds, and reads out a part of the charges generated thereby in a raster scanning manner. When attention is paid to the vertical direction of the raw image signal formed by the read charges, the odd lines have color information of R, G, R, G,..., And the even lines have G, B, G, B,. Color information.

  The CDS / AGC / AD circuit 14 performs a series of processes of noise removal, gain adjustment, and A / D conversion on the raw image signal, and outputs raw image data that is a digital signal. The output raw image data is given to the color separation circuit 24 through the line memory control circuit 22. From the color separation circuit 24, continuous three-line color separation image data in which each pixel has all the R, G, and B color information is output.

  Note that the three output lines are updated at a rate of one line in one line period. That is, when the (N−1) th line, the Nth line, and the (N + 1) th line are output in a certain line period, the Nth line, the (N + 1) th line, and the (N + 2) th line are output in the next one line period.

  The three lines of color separation image data output from the color separation circuit 24 are supplied to NR (Noise Reduction) circuits 26, 28 and 30. That is, among the R data, G data, and B data forming the color separation image data of each line, R data is given to the NR circuit 26, G data is given to the NR circuit 28, and B data is given to the NR circuit 30. Given to. From the NR circuits 26 to 30, R data, G data, and B data corresponding to one central line among the three lines of interest and having reduced color noise are respectively output. The color difference matrix calculation circuit 40 performs matrix calculation on the outputs of the NR circuits 26 to 30 and outputs color difference data, that is, U data and V data.

  The three lines of color separation image data output from the color separation circuit 24 are also provided to the NR circuits 32, 34 and 36. As described above, of the R data, G data, and B data forming the color separation image data of each line, R data is supplied to the NR circuit 32, G data is supplied to the NR circuit 34, and B data is converted to NR. The circuit 36 is provided.

  The NRGEs 32 to 36 are also supplied with the EDGE data output from the edge detection circuit 38. The edge detection circuit 38 detects an edge of the object scene image based on the raw image data output from the CDS / AGC / AD circuit 14. Specifically, the brightness change direction and edge steepness with the edge as a boundary are detected, and EDGE data indicating the polarity and steepness corresponding to the change direction is generated. Such EDGE data is input to the NR circuits 32 to 36.

  Based on the given data, the NR circuits 32 to 36 respectively output R data, G data, and B data corresponding to one central line of the three lines of interest and with reduced color noise. The luminance matrix calculation circuit 42 performs matrix calculation on the outputs of the NR circuits 32 to 36 and outputs luminance data, that is, Y data.

  The signal processing circuit 44 converts the U data and V data output from the color difference matrix calculation circuit 40 and the Y data output from the luminance matrix calculation circuit 42 into an NTSC composite video signal. The converted composite video signal is applied to the LCD monitor 48. As a result, a real-time moving image (through image) of the object scene is displayed on the monitor screen.

  When the shutter button 20 is operated, the CPU 18 instructs the TG 16 to perform one main exposure and one full pixel readout. The TG 16 performs the main exposure on the image sensor 12 and reads out all the charges obtained thereby in a raster scanning manner. The raw image signal formed by the read charges has color information of R, G, R, G,... On the odd lines and G, B, G, B,. Has color information.

  The read raw image signal is subjected to the same processing as described above. As a result, U data and V data are output from the color difference matrix calculation circuit 40, and Y data are output from the luminance matrix calculation circuit 42. When the shutter button 20 is operated, the signal processing circuit 44 performs JPEG compression on the given U data, V data, and Y data, and records the compressed image data on the recording medium 46.

  The NR circuit 26 is configured as shown in FIG. Here, among the R data of the three lines of interest, the R data of the first line is defined as “R1 data”, the R data of the second line is defined as “R2 data”, and the R data of the third line Is defined as “R3 data”.

  An IIR (Infinite Impulse Response) filter 50a performs a filtering process on the R1 data and outputs R1 ir data. The subtractor 52a subtracts the R1 data from the R1irr data and outputs R1dif data indicating the absolute value of the difference between the two. The coefficient generation circuit 54a generates a coefficient K1 based on the R1dif data output from the subtractor 52a, and supplies the generated coefficient K1 to the LPF 58.

  The comparator 56a changes the filter coefficient A of the IIR filter 50a to “0” when the R1dif data value is equal to or greater than the threshold value TH1, and returns the filter coefficient A to the original when the R1dif data value is less than the threshold value TH1. By changing the filter coefficient A to “0”, the R1ir data matches the R1 data, and the R1dif data value becomes “0”. Therefore, the period during which the filter coefficient A is maintained at “0” is one pixel period.

  The IIR filter 50b performs filtering on the R2 data and outputs R2ir data. The subtractor 52b subtracts the R2 data from the R2ir data, and outputs R2dif data indicating the absolute difference between the two. The coefficient generation circuit 54b generates a coefficient K2 based on the R2dif data output from the subtractor 52b, and supplies the generated coefficient K2 to the LPF 58. The comparator 56b changes the filter coefficient A of the IIR filter 50b to “0” when the R2dif data value is greater than or equal to the threshold value TH1, and returns the filter coefficient A to the original when the R2dif data value is less than the threshold value TH1.

  The IIR filter 50c performs filtering on the R3 data and outputs R3ir data. The subtractor 52c subtracts the R3 data from the R3ir data, and outputs R3dif data indicating the absolute value of the difference between them. The coefficient generation circuit 54c generates a coefficient K3 based on the R3dif data output from the subtractor 52c, and provides the generated coefficient K3 to the LPF 58. The comparator 56c changes the filter coefficient A of the IIR filter 50c to “0” when the R3dif data value is equal to or greater than the threshold TH1, and returns the filter coefficient A to the original when the R3dif data value is less than the threshold TH1.

  By returning the filter coefficient A to “0” in this way, a situation in which the values of the R1 ir data, the R 2 ir data, and the R 3 ir data fluctuate unstable after passing the edge is avoided.

  FIG. 4 shows the relationship between the R1dif data to R3dif data and the coefficients K1 to K3. According to FIG. 4, the coefficients K1 to K3 decrease with the slope SL1 as the data value increases as long as the values of the R1dif data to R3dif data are below the threshold value TH1. When the values of the R1dif data to R3dif data become equal to or greater than the threshold value TH1, the coefficients K1 to K3 decrease with the slope SL2 as the data value increases.

  The LPF 58 is a FIR (Finite Impulse Response) filter, and performs a calculation according to the equation 1 to given coefficients K1 to K3. The coefficient K obtained in this way is most affected by the coefficient K2. The weighted addition circuit 60 performs weighted addition according to the coefficient K to the R2ir data and R2 data. Specifically, an operation according to Equation 2 is executed. The weighted addition circuit 46 outputs R2nr data with reduced color noise.

  Each of the IIR filters 50a to 50c is configured as shown in FIG. The R1 data, R2 data or R3 data is given to one input terminal of the adder 72 through the amplifier 70. The output of the adder 72, that is, R1 ir data, R 2 ir data, or R 3 ir data is delayed for one pixel period by the delay circuit 74, and then supplied to the other input terminal of the adder 72 through the amplifier 76. The gain of the amplifier 70 is “1-A”, and the gain of the amplifier 76 is “A”. Each of the IIR filters 50a to 50c has a characteristic indicated by a curve CV1 in FIG. According to the curve CV1, a component having a frequency f1 or higher is removed.

  Since the NR circuits 28 and 30 are configured in the same manner as the NR circuit 26, redundant description is omitted. Note that the LPF 58 provided in each of the NR circuits 28 and 30 executes the calculation according to the above-described equation 1. The weighted addition circuit 60 provided in the NR circuit 28 executes an operation according to the equation 3, and the weighted addition circuit 60 provided in the NR circuit 30 executes an operation according to the equation 4.

  The NR circuit 32 is configured as shown in FIG. Again, among the R data of the three lines of interest, the R data of the first line is defined as “R1 data”, the R data of the second line is defined as “R2 data”, and the R data of the third line Is defined as “R3 data”.

  The median filter 80 pays attention to a total of nine pixels of horizontal 3 pixels × vertical 3 pixels, and outputs data having a central data value (the fifth data value from the top) as R2md data. The median filter 80 has a characteristic indicated by a curve CV2 in FIG. According to the curve CV2, a component having a frequency f2 (f2> f1) or higher is removed.

  The selector 82 is supplied with the R2md data and the input R2 data. The absolute value detection circuit 84 detects the absolute value of the EDGE data, and supplies the detected absolute value to the comparator 86. The comparator 86 instructs the selector 82 to select R2 data when the given absolute value exceeds the threshold value TH2, and instructs the selector 82 to select R2md data when the given absolute value is less than or equal to the threshold value TH2. The selector 82 outputs the selected data as R2sl data.

  Since the NR circuits 34 and 36 are configured in the same manner as the NR circuit 32, redundant description is omitted.

  As can be seen from the above description, the NR circuits 26 to 30 output one line of RGB data from which components having a frequency of f1 or more have been removed based on three consecutive lines of RGB data. Based on the same three lines of RGB data, the NR circuits 32 to 36 output one line of RGB data from which components having a frequency of f2 or higher are removed. Here, the frequency f2 is higher than the frequency f1. The color difference matrix calculation circuit 40 creates UV data based on the RGB data output from the NR circuits 26-30, and the luminance matrix calculation circuit 42 generates Y data based on the RGB data output from the NR circuits 32-36. Create

  In this manner, UV data is created based on the outputs of the NR circuits 26 to 30 having frequency characteristics considering color reproducibility, and based on the outputs of the NR circuits 32 to 36 having frequency characteristics considering luminance reproducibility. Since Y data is created, both color reproducibility and luminance reproducibility can be improved.

It is a block diagram which shows the structure of one Example of this invention. It is a block diagram which shows an example of a structure of the NR circuit applied to the FIG. 1 Example. It is a block diagram which shows one Example of a structure of the IIR filter compared with FIG. 2 Example. FIG. 5 is an illustrative view showing one portion of an operation of a coefficient generation circuit applied to the embodiment in FIG. 2; It is a block diagram which shows an example of a structure of the other NR circuit applied to the FIG. 1 Example. 6 is a graph showing characteristics of an IIR filter applied to the embodiment in FIG. 2 and a median filter applied to the embodiment in FIG. 5.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 ... Digital camera 12 ... Image sensor 24 ... Color separation circuit 26-36 ... NR circuit 40 ... Color difference matrix arithmetic circuit 42 ... Luminance matrix arithmetic circuit 50a-50c ... IIR filter 80 ... Median filter

Claims (5)

A first reduction means for outputting a first reduced color signal of one line from which signal components of a first frequency or higher are removed based on a plurality of continuous lines of unique color signals;
Second reduction means for outputting a second reduced color signal for one line from which signal components of a second frequency and higher than the first frequency are removed based on the unique color signals of the plurality of lines;
First creating means for creating a color difference signal based on the first reduced color signal output from the first reducing means, and creating a luminance signal based on the second reduced color signal output from the second reducing means Comprising a second creation means ;
The first reduction means includes a plurality of cyclic filters that respectively input the plurality of lines of unique color signals, and a plurality of first coefficients that respectively generate a plurality of first weighting coefficients based on outputs of the plurality of cyclic filters. Creating means, second coefficient creating means for creating one second weighting coefficient based on a plurality of weighting coefficients created by the plurality of first coefficient creating means, and a first specific line forming the plurality of lines An image signal processing circuit comprising an adding means for adding the unique color signal and the output of the cyclic filter corresponding to the first specific line in accordance with the second weighting coefficient created by the second coefficient creating means. . Each of the plurality of cyclic filters includes a first multiplying unit that multiplies the reduced color signal of the preceding pixel preceding the target pixel by a first coefficient that is a decimal, and the original color signal of the target pixel is “ Second multiplication means for multiplying a second coefficient that is a complement of 1 ″, and addition for generating a low-frequency signal of the pixel of interest by adding the output of the first multiplication means and the output of the second multiplication means Including means,
The first reduction means further includes changing means for changing the first coefficient to "0" when a level difference between the original color signal of the target pixel and the reduced color signal of the target pixel exceeds a threshold value. image signal processing circuit 1 described. The second reduction means includes: a median filter that inputs the plurality of lines of unique color signals; a unique color signal of the second specific line that forms the plurality of lines; and an output of the median filter that corresponds to the second specific line; The image signal processing circuit according to claim 1 , further comprising selection means for selecting any one of the above. The image signal processing circuit according to claim 3 , wherein the selection unit selects the unique color signal when the level of the edge enhancement signal exceeds a threshold value. A digital camera comprising the image signal processing circuit according to any one of claims 1 to 4 .

Images