JP2004151907A - Image signal processor - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To achieve a good filtering process for an image signal processor. <P>SOLUTION: Noise components contained in Y data output from a YUV conversion circuit 46 are eliminated by an adaptive median filter 52. A chromaticness/hue computing circuit 56 calculates chromaticness C and hue H based on U data and V data output from the YUV conversion circuit 46. An area identifying/coefficient determining circuit 58 varies the characteristic of the adaptive median filter 52 according to the calculated chromaticness C and hue so that it is in inverse proportion to human relative visibility. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to an image signal processing device, and more particularly, to an image signal processing device applied to, for example, an electronic camera and removing a noise component from an image signal.
[0002]
[Prior art]
A conventional image signal processing apparatus of this type is disclosed in Japanese Patent Application Laid-Open No. H11-163,837. This conventional technique increases the tap coefficient of the median filter in accordance with the enlargement ratio of the displayed image, thereby both reducing shaggy caused by the image enlargement process and preventing image quality deterioration caused by the filter process. It was made possible.
[0003]
[Patent Document 1]
JP-A-2002-44425
[0004]
[Problems to be solved by the invention]
However, in the related art, the characteristics of the display image are not reflected in the adjustment of the tap coefficient, and a satisfactory filtering process is not always performed.
[0005]
SUMMARY OF THE INVENTION Therefore, a main object of the present invention is to provide an image signal processing device capable of performing an excellent filtering process on an image signal.
[0006]
[Means for Solving the Problems]
An image signal processing device according to a first aspect of the present invention includes a filter unit that removes a noise component from an image signal, a color evaluation unit that evaluates a color component of an image signal, and a parameter value related to the effectiveness of the filter unit. And a changing unit for changing the relative luminosity factor of the human based on the evaluation result so as to be inversely proportional to the human luminous efficiency.
[0007]
An image signal processing apparatus according to a second aspect of the present invention includes a filter unit for removing a noise component from an image signal, a brightness evaluation unit for evaluating a brightness component of the image signal, and a filter unit based on the evaluation result of the brightness evaluation unit. There is provided changing means for changing the degree of effectiveness.
[0008]
[Action]
According to the first aspect, the noise component included in the image signal is removed by the filter unit. The color evaluation means evaluates the color component of the image signal, and the change means changes a parameter value related to the effectiveness of the filter means so as to be inversely proportional to human relative luminous efficiency based on the evaluation result of the color evaluation means. I do.
[0009]
Human vision is sensitive to yellow and green light, but insensitive to red and blue light. This means that if the filter characteristics for the yellow and green signal components are strengthened, the resolution will decrease, and that even if the filter characteristics for the red and blue signal components are strengthened, the resolution will not decrease. Therefore, in the first invention, the parameter value related to the effectiveness of the filter means is changed so as to be inversely proportional to the relative luminosity factor of a human. As a result, both maintaining the resolution and reducing noise are compatible.
[0010]
Preferably, the color evaluation means includes a saturation detection means for detecting the saturation of the image signal, and a hue detection means for detecting the hue of the image signal. By using the parameters of the saturation and the hue, the separation between the chromatic color and the achromatic color and the distinction of the color are simultaneously realized.
[0011]
Preferably, the brightness evaluation unit evaluates the brightness component of the image signal, and the determination unit determines the effectiveness of the filter unit based on the evaluation result of the brightness evaluation unit. In this case, the above parameter values are reference values used for evaluation by the brightness evaluation means. Thereby, the brightness component of the image signal is reflected in the change of the characteristic of the filter means.
[0012]
Preferably, the determining means increases the effectiveness of the filtering means as the brightness component of the image signal decreases.
[0013]
More preferably, the image signal is subjected to γ correction by γ correction means. In the γ correction, the amplification degree of the low luminance signal component is larger than the amplification degree of the high luminance signal component. This means that noise superimposed on the low-luminance signal component is conspicuous. Therefore, the noise superimposed on the low luminance portion can be made less noticeable by increasing the effectiveness of the filter means as the brightness component decreases.
[0014]
Preferably, the image signal is a signal output from a photographing means for photographing a subject. As a result, good filtering can be performed.
[0015]
According to the second aspect, the noise component included in the image signal is removed by the filter unit. The brightness evaluation means evaluates the brightness component of the image signal, and the change means changes the effectiveness of the filter means based on the evaluation result of the brightness evaluation means. Thereby, the brightness component is reflected in the change of the filter characteristic.
[0016]
Preferably, the image signal is a signal in which each pixel has only color information of one of a plurality of colors. The color separation unit performs color separation on the image signal from which the noise component has been removed by the filter unit, and assigns color information of a plurality of colors to each pixel. By performing a filtering process on an image signal in which each pixel has only one color information, color noise is suppressed.
[0017]
【The invention's effect】
According to the first aspect, by changing the parameter value related to the effectiveness of the filter means so as to be in inverse proportion to the relative luminosity factor of a human, it is possible to maintain both the sense of resolution and the reduction of noise. Good filtering is realized.
[0018]
According to the second aspect, since the brightness component of the image signal is reflected in the change in the effectiveness of the filter means, good filtering can be performed.
[0019]
The above and 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.
[0020]
【Example】
Referring to FIG. 1, a digital camera 10 of this embodiment includes an optical lens 12 and an infrared cut filter 13. The optical image of the subject enters the light receiving surface of the image sensor 14 via these members. The light receiving surface is covered with a color filter 14a having a primary color Bayer array as shown in FIG. 2, and electric charges (pixel signals) generated by respective light receiving elements (pixels) formed on the light receiving surface are R, G, and B signals. It has only one color information.
[0021]
When the power is turned on, a through image display process is executed. First, the CPU 34 instructs a TG (Timing generator) 18 to repeat thinning-out reading. The TG 18 performs thinning-out reading on the image sensor 14 at a frame rate of 30 fps. As a result, a raw image signal having a reduced vertical resolution is output from the image sensor 14 in a raster scanning manner. However, even lines of the output low-resolution raw image signal include pixel signals having color information of R, G, R, G,..., And odd lines of G, B, G, B,. A pixel signal having color information is included.
[0022]
The CDS / AGC / AD circuit 16 performs a series of processes of noise removal, gain adjustment, and A / D conversion on the raw image signal of each frame output from the image sensor 14, and outputs raw image data as a digital signal. . The output raw image data is converted to YUV data by the signal processing circuit 20, and the converted YUV data is written to the SDRAM 24 by the memory controller 22.
[0023]
The video encoder 26 reads the YUV data of each frame from the SDRAM 24 through the memory controller 22, performs a reduced zoom on the read YUV data, and converts the zoomed YUV data into an NTSC composite video signal. The converted composite video signal is supplied to the LCD 28, and as a result, a real-time moving image of the subject, that is, a through image is displayed on the entire LCD 28.
[0024]
When the shutter button 32 is operated, a recording process is executed. First, the CPU 34 instructs the TG 18 to read all pixels. The TG 18 reads all pixels for one frame to the image sensor 14, and as a result, a high-resolution raw image signal is output from the image sensor 14 in a raster scanning manner. The even lines of the output raw image signal include pixel signals having color information of R, G, R, G,..., And the odd lines have pixels having color information of G, B, G, B,. Signal is included.
[0025]
The raw image signal read in response to the operation of the shutter button 32 is subjected to the same processing as described above by the CDS / AGC / AD circuit 16 and the signal processing circuit 20. High resolution YUV data is output from the signal processing circuit 20, and the YUV data is written to the SDRAM 24 via the memory controller 22. The high-resolution YUV data stored in the SDRAM 24 is then supplied to a video encoder 26 via a memory controller 22 and converted into a composite video signal in NTSC format through a reduced zoom. As a result, a still image of the subject at the time when the shutter button 32 is pressed, that is, a frozen image, is displayed on the LCD 28 in its entirety.
[0026]
When the shutter button 32 is operated, the CPU 34 gives a compression command to the JPEG codec 30. The JPEG codec 30 reads high-resolution YUV data from the SDRAM through the memory controller 22, performs JPEG compression on the read YUV data, and writes the compressed YUV data, that is, JPEG data, to the SDRAM 24 via the memory controller 22. The CPU 34 reads the JPEG data stored in the SDRAM 24 through the memory controller 22 and records the read JPEG data on the recording medium 38 through the I / F 36.
[0027]
Note that the recording medium 38 is detachable, and is connected to the I / F 36 at the time of mounting. JPEG data is recorded on the recording medium 38 in a file format.
[0028]
The signal processing circuit 20 is configured as shown in FIG. The raw image data is subjected to white balance adjustment by a white balance adjustment circuit 40, γ correction by a γ correction circuit 42, and color separation by a color separation circuit 44. Interpolated image data generated by color separation is data in which each pixel has all the R, G, and B color information. Thereafter, the interpolated image data output from the color separation circuit 44 is converted into YUV data by the YUV conversion circuit 46. The converted YUV data has a ratio of Y: U: V = 4: 2: 2.
[0029]
The Y data forming the YUV data is supplied to the blocking circuit 50 every three lines. That is, of the three lines of interest, the Y data of the first line is directly provided to the blocking circuit 50 as data L0, and the Y data of the second line is provided to the blocking circuit 50 as data L1 via the 1H delay line 48a. The Y data on the third line is supplied to the blocking circuit 50 as data L2 via the 1H delay lines 48a and 48b. Therefore, two lines overlap between the three lines simultaneously input to the blocking circuit 50 during one certain line period and the three lines simultaneously input to the blocking circuit 50 during the next one line period.
[0030]
The blocking circuit 50 extracts Y data of 3 horizontal pixels × 3 vertical pixels from the input Y data of 3 lines, and outputs the extracted 9 pixels of Y data at the same time as data D0 to D8. An overlap of 2 horizontal pixels × 3 vertical pixels occurs between the nine pixels output in one certain pixel period and the nine pixels output in the next one pixel period.
[0031]
Referring to FIG. 4, blocking circuit 50 includes six registers 50r-1 to 50r-6. Data L0 is delayed by one pixel period by registers 50r-1 and 50r-2, data L1 is delayed by one pixel period by registers 50r-3 and 50r-4, and data L3 is registered by registers 50r-5 and 50r-6. Is delayed by one pixel period. Therefore, data L0 of three consecutive pixels is output as data D0 to D2, data L1 of three consecutive pixels is output as data D3 to D5, and data L2 of three consecutive pixels is output as data D6 to D8. You.
[0032]
Of the data D0 to D9, the data D4 is the data of the pixel existing at the center of the extracted nine pixels. Therefore, the data D4 is defined as “target pixel data”, and the data D0 to D3 and D5 to D9 are defined as “peripheral pixel data”.
[0033]
The adaptive median filter 52 fetches the data D0 to D8 from the blocking circuit 50 and executes a filter process using an amount of data according to the tap coefficient T. More specifically, with reference to FIG. 7, if the tap coefficient T is “9”, data D0, D1, D3, D5, D6, D7, D8, and D4 for nine pixels are sorted in the order according to the data values. Then, the data arranged at the center is selected. If the tap coefficient T is "11", the data D0, D1, D3, D5, D6, D7, D8, D4, D4 and D4 for 11 pixels are sorted in the order according to the data values, and the data arranged at the center. Is selected. If the tap coefficient T is "13", the data D0, D1, D3, D5, D6, D7, D8, D4, D4, D4, D4 and D4 for 13 pixels are sorted in the order according to the data values, and The arranged data is selected. If the tap coefficient T is "15", the data D0, D1, D3, D5, D6, D7, D8, D4, D4, D4, D4, D4, D4, and D4 for 15 pixels are sorted in the order according to the data values. Then, the data arranged at the center is selected.
[0034]
As can be seen from FIG. 7, as the tap coefficient T increases, the ratio of the data D4 to all data increases, and the data D4 is more easily selected. In other words, of the numerical values “9”, “11”, “13”, and “15” that the tap coefficient T can take, the smaller the numerical value, the stronger the filter effect, and the larger the numerical value, the more the filter effect. become weak.
[0035]
If the tap coefficient T is “0”, no filtering is performed, and the data D4 is selected unconditionally.
[0036]
Referring to FIG. 5, adaptive median filter 52 includes one encoder 52e, seven selectors 52s-1 to 52s-7, and 79 comparators 52c-1 to 52c-79. .
[0037]
Of the data D0 to D8 output from the blocking circuit 50, the data D0 and D1 are provided to the upper input terminal and the lower input terminal of the comparator 52c-1, respectively, and the data D2 and D3 are provided to the comparator 52c-2. It is provided to the upper input terminal and the lower input terminal, respectively. Data D5 and D6 are provided to an upper input terminal and a lower input terminal of comparator 52c-3, respectively, and data D7 and D8 are provided to an upper input terminal and a lower input terminal of comparator 52c-4, respectively. Further, the data D4 is provided to the lower input terminal of the comparator 52c-17.
[0038]
Each of the comparators 52c-1 to 52c-79 compares the data input from the upper input terminal and the data input from the lower input terminal with each other, and outputs data having a larger numerical value from the upper output terminal. Data with a small numerical value is output from the lower output terminal. Since the connection relation of the comparators 52c-1 to 52c-79 is clear from FIG. 5, detailed description of the connection relation is omitted, but from the upper output terminal of the comparator 52c-24, the data D0 to D8 The data having the largest numerical value is output as data D_MAX.
[0039]
The encoder 52e outputs control data s0 to s6 having numerical values shown in FIG. 6 according to the input tap coefficient T. If the tap coefficient T is “9”, “1111110” is output as the control data s0 to s6, and if the tap coefficient T is “11”, “1111001” is output as the control data s0 to s6. If the tap coefficient T is “13”, “1100002” is output as the control data s0 to s6, and if the tap coefficient T is “15”, “000000003” is output as the control data s0 to s6. Further, if the tap coefficient T is “0”, “0000004” is output as the control data s0 to s6.
[0040]
The control data s0 to s5 are provided to selectors 52s-1 to 52s-6, respectively. Each of selectors 52s-1 to 52s-6 selects data D4 when the applied control data value indicates "0", and selects "0" when the applied control data value indicates "1". .
[0041]
The control data s6 is provided to the selector 52s-7. The selector 52s-7 selects the data D_M9 output from the lower output terminal of the comparator 52c-39 when the control data value indicates “0”, and selects the data D_M9 when the control data value indicates “1”. The data D_M11 output from the upper output terminal of 52c-55 is selected. The selector 52s-7 selects the data D_M13 output from the lower output terminal of the comparator 52c-64 when the control data value indicates “2”, and selects the data D_M13 when the control data value indicates “3”. The data D_M15 output from the upper output terminal of the detector 52c-79 is selected. The selector 52s-7 further selects the data D4 when the control data value indicates "4".
[0042]
Therefore, when the tap coefficient T indicates "9", the comparators 52c-1 to 52c-39 are enabled, and when the tap coefficient T indicates "11", the comparators 52c-1 to 52c-55 are enabled. Is done. When the tap coefficient T indicates "13", the comparators 52c-1 to 52c-64 are enabled, and when the tap coefficient T indicates "15", the comparators 52c-1 to 52c-79 are enabled. Is done. When the tap coefficient T indicates “0”, all of the comparators 52c-1 to 52c-79 are invalidated.
[0043]
The data selected by the selector 52s-7 is output from the adaptive median filter 52 as data D_MED. The data D_MAX output from the adaptive median filter 52 is output from the signal processing circuit 20 as Y data.
[0044]
Returning to FIG. 3, the U data and V data generated by the YUV conversion circuit 46 are output from the signal processing circuit 20 as they are, and sent to the saturation / hue calculation circuit 56 via the 1H delay lines 54a and 54b, respectively. Given. The saturation / hue calculation circuit 56 calculates the saturation C according to Equation 1, and calculates the hue H according to Equation 2.
[0045]
(Equation 1)
C = (U²+ V²)^1/2
[0046]
(Equation 2)
H = arctan (U / V)
The area determination / coefficient determination circuit 58 determines which of the zones Zuv0 to Zuv5 shown in FIG. 9 the calculated chroma C and hue H belong to. Specifically, the saturation C is compared with a threshold value THC, and the hue H is compared with each of the threshold values THH1 to THH4 to determine a zone to which the saturation C and the hue H belong.
[0047]
Achromatic colors are distributed in the zone Zuv0, and blue is distributed in the zones Zuv1 and Zuv5. Red is distributed in zone Zuv2, yellow is distributed in zone Zuv3, and green is distributed in zone Zuv4. The reason for forming the zone Zuv3 corresponding to yellow in addition to the three primary colors red, green and blue is that this zone Zuv3 contains a flesh color and performs special processing on the flesh color.
[0048]
For example, when zone Zuv0 is determined, region determination / coefficient determination circuit 58 determines numerical values TH0Z0 to TH3Z0 as thresholds THL0 to THL3, as shown in FIG. The area discrimination / coefficient determination circuit 58 then compares the data D_MAX output from the adaptive median filter 52 with each of the threshold values THL0 to THL3, and sets the tap coefficient T to “9”, “11”, “ 13 "," 15 "and" 0 ". The adaptive median filter 52 is provided with the tap coefficient T thus determined.
[0049]
The area determination / coefficient determination circuit 58 is configured as shown in FIG. Hue H is compared with thresholds THH1 to THH4 in comparators 58c-1 to 58c-4, and saturation C is compared with threshold THC in comparator 58c-5. The comparison data cph1 output from the comparator 58c-1 indicates “1” when H> THH1, and indicates “0” when H ≦ THH1. The comparison data cph2 output from the comparator 58c-2 indicates “1” when H> THH2, and indicates “0” when H ≦ THH2. The comparison data cph3 output from the comparator 58c-3 indicates “1” when H> THH3, and indicates “0” when H ≦ THH3. The comparison data cph4 output from the comparator 58c-4 indicates “1” when H> THH4, and indicates “0” when H ≦ THH4. The comparison data cpc output from the comparator 58c-5 indicates “1” when C> THC, and indicates “0” when C ≦ THC.
[0050]
The first decoder 58d-1 generates the zone discrimination data ZUV based on the comparison data cph1 to cph4 and cpc. At this time, the relationship shown in FIG. 11 is established between the comparison data cph1 to cph4 and cpc and the zone determination data. In FIG. 11, "X" indicates that the data value may be either "0" or "1". Numerical values “0” to “5” that can be taken by the zone determination data ZUV respectively correspond to the zones Zuv0 to Zuv5 shown in FIG.
[0051]
According to FIG. 11, if the comparison data cpc is “0”, the zone determination data ZUV indicates “0”. If each of the comparison data cph1 to cph4 is “0” and the comparison data cpc is “1”, the zone determination data ZUV indicates “1”. If each of the comparison data cph1 and cpc is “1” and each of the comparison data cph2 to cph4 is “0”, the zone determination data ZUV indicates “2”. If each of the comparison data cph1, cph2 and cpc is “1” and the comparison data cph3 and cph4 are “0”, the zone discrimination data ZUV indicates “3”. If each of the comparison data cph1 to cph3 and cpc is “1” and the comparison data cph4 is “0”, the zone discrimination data ZUV indicates “4”. If all of the comparison data cph1 to cph4 and cpcl are “1”, the zone determination data ZUV indicates “5”.
[0052]
The selectors 58s-1, 58s-2, 58s-3 and 58s-4 select the threshold values TH0Z0, TH1Z0, TH2Z0 and TH3Z0 when ZUV = 0, and select the threshold values TH0Z1, TH1Z1, TH2Z1 and TH3Z1 when ZUV = 1. select. When ZUV = 2, thresholds TH0Z2, TH1Z2, TH2Z2 and TH3Z2 are selected, and when ZUV = 3, thresholds TH0Z3, TH1Z3, TH2Z3 and TH3Z3 are selected. Further, when ZUV = 4, the threshold values TH0Z4, TH1Z4, TH2Z4 and TH3Z4 are selected, and when ZUV = 5, the threshold values TH0Z5, TH1Z5, TH2Z5 and TH3Z5 are selected.
[0053]
Here, the threshold values TH * Z1 to TH * Z5 (*: 0 to 3) respectively corresponding to the zones Zuv1 to Zuv5 have a correlation with the curve A indicating the standard relative visibility shown in FIG. More specifically, the curves B0 to B3 are curves that are inversely proportional to the curve A. Thresholds TH0Z1 to TH0Z5 are distributed along curve B0, thresholds TH1Z1 to TH1Z5 are distributed along curve B1, thresholds TH2Z1 to TH2Z5 are distributed along curve B2, and thresholds TH3Z1 to TH3Z5 are along curve B3. Distribute. Note that the standard relative luminosity curve is a curve indicating that human vision is sensitive to yellow or green light, but insensitive to red or blue light.
[0054]
Returning to FIG. 8, the threshold value selected by the selector 58s-1 is provided to the comparator 58c-6 as the threshold value THL0, and the threshold value selected by the selector 58s-2 is provided to the comparator 58c-7 as the threshold value THL1. The threshold value selected by the selector 58s-3 is provided to the comparator 58c-8 as the threshold value THL2, and the threshold value selected by the selector 58s-4 is provided to the comparator 58c-9 as the threshold value THL3.
[0055]
The comparators 58c-6 to 58c-9 compare the threshold values THL0 to THL3 provided from the selectors 58s-1 to 58s-4 with the data D_MAX output from the adaptive median filter 70, respectively. The comparison data cpl0 output from the comparator 58c-6 indicates “1” when D_MAX> THL0, and indicates “0” when D_MAX ≦ THL0. The comparison data cpl1 output from the comparator 58c-7 indicates “1” when D_MAX> THL1, and indicates “0” when D_MAX ≦ THL1. The comparison data cpl2 output from the comparator 58c-8 indicates “1” when D_MAX> THL2, and indicates “0” when D_MAX ≦ THL2. The comparison data cpl3 output from the comparator 58c-9 indicates “1” when D_MAX> THL3, and indicates “0” when D_MAX ≦ THL3.
[0056]
The second decoder 58d-2 generates a tap coefficient T based on the comparison data cpl0 to cpl3. At this time, the relationship shown in FIG. 12 is established between the comparison data cpl0 to cpl3 and the tap coefficient T.
[0057]
According to FIG. 12, if all of the comparison data cpl0 to cpl3 are “0”, the tap coefficient T indicates “9”. If the comparison data cpl0 is “1” and the comparison data cpl1 to cpl3 are “0”, the tap coefficient T indicates “11”. If the comparison data cpl0 and cpl1 are “1” and the comparison data cpl2 and cpl3 are “0”, the tap coefficient T indicates “13”. If the comparison data cpl0 to cpl2 are “1” and the comparison data cpl3 is “0”, the tap coefficient T indicates “15”. If all of the comparison data cpl0 to cpl3 are “1”, the tap coefficient T indicates “0”.
[0058]
Therefore, as the data D_MAX is larger, the tap coefficient T indicates a larger value among “9”, “11”, “13”, and “15”, and as the data D_MAX is smaller, the tap coefficient T is “9”, “”. It shows a small value among "11", "13", and "15". Therefore, the larger the data D_MAX is, the weaker the filter is, and the easier it is to select the data D4. Conversely, the smaller the data D_MAX is, the stronger the effect of the filter is, and the more difficult it is to select the data D4.
[0059]
As can be understood from the above description, the noise component of the Y data output from the YUV conversion circuit 46 shown in FIG. 3 is removed by the adaptive median filter 52. The saturation / hue operation circuit 56 calculates the saturation C and the hue H based on the U data and the V data output from the YUV conversion circuit 46. The calculated saturation C is compared with the threshold value THC by the comparator 58c-5 shown in FIG. 8, and the hue H is compared with the threshold values THH1 to THH4 by the comparators 58c-1 to 58c-4 shown in FIG. The first decoder 58d-1 outputs the zone determination data ZUV based on the comparison data cpc output from the comparator 58c-5 and the comparison data cph1 to cph4 output from the comparators 58c-1 to 58c-4. . Thereby, the color evaluation result of the target pixel is obtained.
[0060]
The selectors 58s-1 to 58s-4 output thresholds THL0 to THL3 having numerical values corresponding to the zone determination data ZUV. Data D_MAX, which is Y data related to the target pixel, is compared with thresholds THL0 to THL3 by comparators 58c-6 to 58c-9. As a result, a brightness evaluation result related to the pixel of interest is obtained. The second decoder 58d-2 determines the tap coefficient T based on the comparison data cpl0 to cpl3 output from the comparators 58c-6 to 58c-9.
[0061]
The threshold values THL0 to THL3 indicate numerical values that are inversely proportional to human relative luminous efficiency characteristics, and the effectiveness of the filter is changed so as to be inversely proportional to the relative luminous efficiency characteristics. Further, the tap coefficient T becomes smaller as the data D_MAX becomes lower, and thus the effect of the filter becomes stronger as the image has lower luminance.
[0062]
Human vision is sensitive to yellow and green light, but insensitive to red and blue light. This means that if the filter characteristics for the yellow and green signal components are strengthened, the resolution will decrease, and that even if the filter characteristics for the red and blue signal components are strengthened, the resolution will not decrease. Therefore, in this embodiment, the characteristics of the adaptive median filter 52 are changed so as to be inversely proportional to the relative luminosity factor of a human. As a result, the filter characteristics are weakened in yellow or green, where human vision is sensitive, and a reduction in resolution is prevented. Further, in red and blue colors in which human vision is insensitive, filter characteristics are strengthened and noise is removed. In this way, the maintenance of a sense of resolution and the reduction of noise are compatible with each other, and good filter processing can be performed.
[0063]
Further, the γ correction circuit 42 shown in FIG. 3 performs γ correction on the raw image data before the YUV conversion of the YUV conversion circuit 46. In the γ correction, the amplification degree of the low luminance component is larger than the amplification degree of the high luminance component. This means that noise superimposed on the low-luminance signal component is conspicuous. Therefore, in this embodiment, the lower the luminance is, the stronger the effect of the filter is. This makes it possible to make noise superimposed on the low-luminance portion inconspicuous. Since maintaining the sense of resolution is not important in the low luminance portion, both noise reduction and maintaining the sense of resolution are compatible here.
[0064]
The digital camera 10 of the other embodiment is the same as the embodiment of FIG. 1 except that the signal processing circuit 20 is configured as shown in FIG. For this reason, different points will be mainly described, and redundant description of similar portions will be omitted.
[0065]
Referring to FIG. 14, raw image data is provided to RB blocking circuit 64 and G blocking circuit 66 every five lines. That is, of the five lines of interest, the raw image data of the first line is directly supplied to the RB blocking circuit 64 and the G blocking circuit 66, and the raw image data of the second line is supplied to the RB block via the 1H delay line 62a. The raw image data of the third line is supplied to the RB blocking circuit 64 and the G blocking circuit 66 via the 1H delay lines 62a and 62b. The raw image data of the fourth line is provided to the RB blocking circuit 64 and the G blocking circuit 66 via the 1H delay lines 62a to 62c, and the raw image data of the fifth line is provided via the 1H delay lines 62a to 62d. RB blocking circuit 64 and G blocking circuit 66.
[0066]
Therefore, five lines of raw image data provided to the RB blocking circuit 64 and the G blocking circuit 66 during a certain one line period, and to the RB blocking circuit 64 and the G blocking circuit 66 during the next one line period. Four lines overlap with the raw image data for five lines.
[0067]
When the matrix of 5 horizontal pixels × 5 vertical pixels forms the pattern shown in FIG. 15A, the RB blocking circuit 64 extracts nine pixel data having R color information as data D0 to D8. I do. When the matrix of 5 horizontal pixels × 5 vertical pixels of interest forms the pattern shown in FIG. 15B, the RB blocking circuit 64 converts pixel data of 9 pixels having B color information as data D0 to D8. Extract. On the other hand, when the matrix of 5 horizontal pixels × 5 vertical pixels of interest forms the pattern shown in FIG. 15C, the G blocking circuit 66 converts pixel data of 9 pixels having G color information into data D0 to D8. Extract as
[0068]
Also in this embodiment, the data D4 is data of a pixel existing at the center of the extracted nine pixels. Therefore, the data D4 is defined as “target pixel data”, and the data D0 to D3 and D5 to D9 are defined as “peripheral pixel data”.
[0069]
The selector 68 selects the RB blocking circuit 64 when the pattern shown in FIG. 15A or 15B is formed, and selects the G blocking circuit 66 when the pattern shown in FIG. Select Therefore, pixel data of 9 pixels having only R color information, 9 pixel data having only G color information, or 9 pixel data having only B color information are input to the adaptive median filter 70. You.
[0070]
The adaptive median filter 70 has the same configuration as the adaptive median filter 52 shown in FIG. 3, and data having the largest numerical value among the input data D0 to D8 is input to the brightness determination circuit 72 as data D_MAX. . The brightness determination circuit 72 compares the value of the input data D_MAX with each of the thresholds TH1 to TH4 (TH1 <TH2 <TH3 <TH4), and inputs a tap coefficient T according to the comparison result to the adaptive median filter 70. I do.
[0071]
Referring to FIG. 16, if D_MAX <TH1, tap coefficient T input to adaptive median filter 70 indicates “9”. If TH1 ≦ D_MAX <TH2, the tap coefficient T input to the adaptive median filter 70 indicates “11”. If TH2 ≦ D_MAX <TH3, the tap coefficient T input to the adaptive median filter 70 indicates “13”. If TH3 ≦ D_MAX <TH4, the tap coefficient T input to the adaptive median filter 70 indicates “15”. If TH4 ≦ D_MAX, the tap coefficient T input to the adaptive median filter 70 indicates “0”.
[0072]
Therefore, as the data D_MAX is larger, the tap coefficient T indicates a larger value among “9”, “11”, “13”, and “15”, and as the data D_MAX is smaller, the tap coefficient T is “9”, “”. It shows a small value among "11", "13", and "15". Therefore, the larger the data D_MAX is, the weaker the filter is, and the easier it is to select the data D4. Conversely, the smaller the data D_MAX is, the stronger the effect of the filter is, and the more difficult it is to select the data D4.
[0073]
The raw image data output from the adaptive median filter 70 is supplied to a color separation circuit 78 via a white balance adjustment circuit 74 and a gamma correction circuit 76, and each pixel has an interpolation having all the R, G, and B color information. Converted to image data. The converted interpolated image data is then output via the YUV conversion circuit 80.
[0074]
As can be understood from the above description, the noise component included in the raw image data is removed by the adaptive median filter 70. The brightness / discrimination circuit 72 evaluates the brightness of the image based on the data D_MAX related to the target pixel, and changes the tap coefficient T based on the evaluation result. As a result, the brightness of the image is reflected in the change in the filter characteristics of the adaptive median filter 70. The raw image data is data in which each pixel has only R, G, or G color information. By performing filter processing on such image data, color noise is suppressed. Thereby, good filter processing is realized.
[0075]
In this embodiment, an improved type of the median filter is used, but another filter such as a Gaussian filter may be used as long as the characteristics of the filter can be changed. In this embodiment, a digital camera has been described, but the present invention can also be applied to an image signal processing device such as a television receiver or a video recorder. Further, in this embodiment, the data having the largest numerical value among the data D0 to D8 is given as the data D_MAX, and the data D_MAX is given to the comparators 58c-6 to 58c-9 shown in FIG. Alternatively, data D4 may be provided to comparators 58c-6 to 58c-9. In this case, the brightness of the target pixel is always evaluated.
[Brief description of the drawings]
FIG. 1 is a block diagram showing one embodiment of the present invention.
FIG. 2 is an illustrative view showing one example of a configuration of a color filter applied to the embodiment in FIG. 1;
FIG. 3 is a block diagram illustrating an example of a configuration of a signal processing circuit applied to the embodiment in FIG. 1;
FIG. 4 is a block diagram illustrating an example of a configuration of a blocking circuit applied to the embodiment in FIG. 3;
FIG. 5 is a block diagram showing an example of a configuration of an adaptive median filter applied to the embodiment in FIG. 3;
FIG. 6 is an illustrative view showing one portion of an operation of the adaptive median filter shown in FIG. 5;
FIG. 7 is an illustrative view showing another portion of the operation of the adaptive median filter shown in FIG. 5;
FIG. 8 is a block diagram illustrating an example of a configuration of a region determination / coefficient determination circuit applied to the embodiment in FIG. 3;
FIG. 9 is an illustrative view showing one portion of an operation of the area determining / coefficient determining circuit shown in FIG. 8;
FIG. 10 is an illustrative view showing another portion of the operation of the area determining / coefficient determining circuit shown in FIG. 8;
FIG. 11 is an illustrative view showing another portion of the operation of the area determination / coefficient determination circuit shown in FIG. 8;
FIG. 12 is an illustrative view showing yet another portion of the operation of the area determination / coefficient determination circuit shown in FIG. 8;
FIG. 13 is a block diagram showing an example of a configuration of a signal processing circuit applied to another embodiment of the present invention.
FIG. 14 is a graph showing a standard relative luminosity curve.
FIG. 15 is an illustrative view showing one portion of an operation of the embodiment in FIG. 13;
FIG. 16 is an illustrative view showing another portion of the operation of the embodiment in FIG. 13;
[Explanation of symbols]
10. Digital camera
20 ... Signal processing circuit
42, 76 ... γ correction circuit
46, 80 ... YUV conversion circuit
52, 70 ... Adaptive median filter
56: Saturation / hue calculation circuit
58: Area discrimination / coefficient determination circuit

Claims (9)

Filter means for removing noise components from the image signal,
Color evaluation means for evaluating a color component of the image signal, and changing means for changing a parameter value related to the effectiveness of the filter means so as to be inversely proportional to human relative luminous efficiency based on the evaluation result of the color evaluation means An image signal processing device comprising: 2. The image signal processing device according to claim 1, wherein the color evaluation unit includes a saturation detection unit that detects a saturation of the image signal, and a hue detection unit that detects a hue of the image signal. Brightness evaluation means for evaluating a brightness component of the image signal, and a determination means for determining the effectiveness of the filter means based on the evaluation result of the brightness evaluation means,
The image signal processing device according to claim 1, wherein the parameter value is a reference value used for evaluation by the brightness evaluation unit. 4. The image signal processing apparatus according to claim 3, wherein the determining unit increases the effectiveness of the filter unit as the brightness component of the image signal decreases. 5. The image signal processing device according to claim 4, further comprising a gamma correction unit that performs gamma correction on the image signal. It further comprises a photographing means for photographing the subject,
The image signal processing device according to claim 1, wherein the image signal is a signal output from the photographing unit. Filter means for removing noise components from the image signal,
An image signal processing apparatus, comprising: a brightness evaluation unit that evaluates a brightness component of the image signal; and a change unit that changes the effectiveness of the filter unit based on an evaluation result of the brightness evaluation unit. The image signal is a signal in which each pixel has only color information of one of a plurality of colors. The image signal from which a noise component has been removed by the filter means is subjected to color separation, and the color information of the plurality of colors is applied to each pixel. The image signal processing apparatus according to claim 7, further comprising a color separation unit that assigns a color. 9. The image signal processing device according to claim 7, wherein the plurality of colors are R, G, and B.

Images