JPH04263577A - Color picture processor - Google Patents

Abstract

PURPOSE:To prevent picture quality from deteriorating by detecting the amount of signal change based on a chrominance signal and the average color signal of this color and correcting the brightness of the average value chrominance signal of the other color by using the change amount. CONSTITUTION:Original picture information read after photoelectric conversion by a CCD line sensor 1 is corrected in a shading correction circuit 3, turned to be a digital signal in an A/D conversion circuit 4, and inputted to a color separation circuit 5. An R signal outputted from the circuit 5 is inputted to shift registers 13 and 14 in order. When the R signal inputted to the register 13 is for the noted picture element, the R signal before one picture element is in the register 14, and the R signal with one picture element is obtained from the circuit 5. The average value R signal with three picture elements is obtained by the arithmetic circuit 19, and the noted picture elements and the average values of B and G signals are similarly obtained. Further, the ratio of the G signal of the noted picture element and the average value G signal calculated by an arithmetic circuit 20 is obtained. The picture quality is improved by correcting the signal ratio by proportionating it to the average value R signal in the arithmetic circuit 23.

Description

[Detailed description of the invention]

0001

[Industrial Application Field] The present invention relates to a color image reading device attached to the output side of a color image reading device, such as a digital color copying machine or a color scanner, which separates one pixel into a plurality of color signals and reads them. The present invention relates to an image processing device.

0002

2. Description of the Related Art Generally, in this type of color image reading apparatus, it is necessary to separate a color image by some method and extract it as a signal of each color component. In this case, in order to make the reading optical system smaller in size, there is a color image reading device that uses a color line image sensor in which a color separation filter is directly attached to the front surface of the line sensor. Specifically, the filter forming one pixel is R
It is a color point sequential system that sequentially arranges three colors: (red), G (green), and B (blue). However, with such a filter configuration, one pixel is represented by three light-receiving elements on the sensor, so the reading resolution is 1/3 of the resolution of a normal black-and-white image reading device.

[0003] In the image sensor system in which a filter is directly attached, in order to avoid such a decrease in resolution, R, G,
Among the B filters, for example, the arrangement density of the G filter is increased to every other light receiving element, and the remaining R filters and B filters are arranged by color, such as reducing the number and arranging them alternately between the G filters. JP-A-58-173962 discloses materials with different densities. According to this, the arrangement density of the G filter is twice that of the R filter and the B filter, and if sampling is performed for each G filter, the resolution of the color point sequential method described above is 3/2 times as large. It can be read at high resolution.

0004

[Problem to be Solved by the Invention] According to the above-mentioned publication system, although the resolution of the G signal compatible with the G filter becomes high,
In image edge portions, color shift may occur due to differences in the positions of the R filter and the B filter relative to the pixel of interest. That is, although the G signal is an actual signal, the R signal and the B signal are actually interpolated by the arithmetic mean of two adjacent signals in the signal sequence. As a result, the image quality deteriorates due to the occurrence of colors different from the actual colors in the reproduced image.

[0005]

[Means for Solving the Problem] In the invention as claimed in claim 1, color separation filters of a plurality of different colors are arranged with different arrangement densities depending on the color, with each filter covering each light receiving element. In a color image processing device that obtains each color signal of a pixel of interest from information read by a line image sensor installed, the average of each color signal is calculated based on each color signal of the pixel of interest itself and each color signal of pixels surrounding the pixel of interest. An average value calculation means for calculating a value color signal is provided, and the color signal of the pixel of interest output from the light receiving elements compatible with color separation filters arranged in high density is output as is, and the average value of this color signal and this color is output as is. A specific color signal processing means is provided for detecting a signal change amount based on the color signal, and the brightness of each of the average value color signals of other colors is corrected using the signal change amount to output a color signal of that color. A brightness correction means was provided.

More specifically, three primary colors R, G, and B are used as each color separation filter, and as in the second aspect of the invention, the G filter among the R, G, and B filters is used as one light receiving element. R filters and B filters are arranged alternately, and
In a color image processing device that uses line image sensors arranged alternately with respect to light-receiving elements between filters to obtain each R, G, and B signal of a pixel of interest based on the arrangement density of G filters, the pixel of interest itself An average value calculation means is provided for calculating an average value color signal of each color based on each R, G, B signal of each pixel located before and after the pixel of interest, and each R, G, B signal of pixels located before and after the pixel of interest. A G color signal processing means is provided that outputs the signal as is and detects luminance information from the ratio of this G signal and the average value G signal, and uses this luminance information to determine the luminance of the average value R signal and the average value B signal. Correct R
A brightness correction means for outputting the signal and the B signal was provided.

On the other hand, in the invention as claimed in claim 3, color separation filters of a plurality of different colors are read by line image sensors arranged periodically, with each filter covering each light receiving element. In a color image processing device that obtains each color signal of a pixel of interest from the information obtained, a detection means for detecting the amount of change in the color signal of the pixel of interest is provided, and the number of pixels before and after the pixel of interest is determined according to the amount of change. An average value calculation means is provided for calculating an average value color signal based on the color signal of the pixel of interest and the color signals of pixels before and after the pixel of interest by varying the pixel of interest.

In this case, in the invention as set forth in claim 4, the detection means detects the magnitude of the difference between color signals of one specific color obtained from two pixels located before and after the pixel of interest; In the described invention, the detection means detects the amount of change in the color signal of one specific color among the color signals of the pixel of interest,
The number of pixels before and after the pixel of interest used for calculation by the average value calculating means is varied so that it increases when the amount of change is greater than a predetermined value and decreases when the amount of change is smaller than a predetermined value.

Furthermore, in the invention as set forth in claim 6, the color separation filters are arrayed with different array densities depending on the color;
A detection means is provided for color signals corresponding to color separation filters with high arrangement density.

0010

[Operation] Human visual characteristics have low resolution for color information and high resolution for luminance information. Furthermore, indiscernible high-frequency information has a spatial integration effect. Therefore, claim 1
Alternatively, according to the invention described in 2, the color signal of the pixel of interest is obtained as an average value color signal of the color signals of pixels before and after the pixel of interest, and color signal information corresponding to a color separation filter with a high arrangement density that regulates resolution is obtained. The amount of change in the signal based on is regarded as brightness information, and the brightness is reproduced by performing brightness correction that changes the brightness of the average color signal corresponding to the color separation filter with low array density, while maintaining the resolution. , the color shift that occurs at the edge of the image is reduced.

In the invention as claimed in claim 3, the number of surrounding pixels used when calculating the average value is made variable by the average value calculation means, the amount of change in the color signal of the pixel of interest is detected by the detection means, and the amount of change in the color signal of the pixel of interest is detected according to the detection result. For example, claim 5
As in the invention described above, the number of pixels is increased in sharp image edge parts where the amount of change is large, and the average value over a wide range of the previous and subsequent images is taken to obtain the intermediate color between the previous and subsequent images, making color shift less visible and reducing the amount of change. In a smooth image part with a small number of pixels, the average signal of a small number of pixels or the signal of the target pixel itself is output, and a clear color is reproduced in the output of a pixel close to a pixel containing an edge.

0012

[Embodiment] An embodiment of the invention according to claims 1 and 2 is shown in FIG.
This will be explained based on FIGS. First, the CC, which is a line image sensor for color reading used in this embodiment,
FIG. 2 shows how the color separation filters are arranged in the D-line sensor 1. In the figure, the parts indicated by R, G, and B are one light receiving element each equipped with red, green, and blue transmission filters 2, respectively. In this example, the G filters are arranged at a density of every other light receiving element, while the R filters and B filters are arranged between the G filters and arranged alternately, that is, at a density of every third light receiving element. There is.

As shown in FIG. 3, on the output side of the CCD line sensor 1, a shading correction circuit 3,
An A/D conversion circuit 4 and a color separation circuit 5 are connected in sequence. Therefore, the original image information photoelectrically converted and read by the CCD line sensor 1 is subjected to electrical shading correction, converted to a digital signal, and inputted to the color separation circuit 5.

As shown in FIG. 4, the color separation circuit 5 includes latch circuits 6, 7, and 8 for latching the digitized input signal a at the timing of each of the R, G, and B filters. It is comprised of latch circuits 9, 10, and 11 that latch signals b, c, and d from latch circuits 6, 7, and 8, and output them as signals e, f, and g at the same output timing. Therefore, as shown in the timing chart of FIG. 5, the signal g is output as it is as a G signal at the period of the G filter, and the signals e and f are output as R and B signals respectively.
This signal is output as a color signal for the same pixel as the signal, and R, G, and B signals are obtained from the color separation circuit 5 at a resolution compatible with the G filter. The processing up to now has been carried out in the same manner in the above-mentioned publications.

However, with this much image processing,
The G signal can be obtained as an output as is, but the R signal
The signal and B signal are actually obtained as signals supplemented by interpolation, and for example, depending on the position of the R filter and B filter at the edge of the image, the color may be read as a completely different color from the actual color. There is. Therefore,
In this embodiment, as shown in FIG. 3, a color conversion circuit 12 is further provided on the output side of the color separation circuit 5, and the color information is converted from the average of surrounding pixels to the pixel of interest, and the luminance information is ) By extracting it as G signal information, G
This is designed to reduce color blurring at image edges while maintaining filter-compatible resolution.

The color conversion circuit 12 is constructed as shown in FIG. First, each R, separated by the color separation circuit 5,
Two-stage shift register 13 into which G and B signals are input
~18 are provided. Here, an arithmetic circuit 19 is provided as an average value calculation means which inputs the R signal from the color separation circuit 5 and the output signals from the shift registers 13 and 14 and performs an operation to obtain an average color signal. G
Arithmetic circuits 20 and 21 are provided as average value calculation means for similarly obtaining an average value color signal for the signal and the B signal. Further, an arithmetic circuit 22 is provided that calculates the ratio (signal change amount) between the output signal from the shift register 15 for the G signal and the output signal from the arithmetic circuit 20. Furthermore, an arithmetic circuit 23 is provided which performs a multiplication process on the output signals from the arithmetic circuits 19 and 22 and serves as a luminance correction means for outputting a corrected color signal R'. Similarly, the output signals from the arithmetic circuits 21 and 22 are multiplied,
An arithmetic circuit 24 serving as a brightness correction means for outputting a corrected color signal B' is also provided. The color signal G is configured such that the signal from the shift register 15 is output as is as the color signal G'. That is, the arithmetic circuit 22 constitutes a G signal processing means 25 serving as a specific color signal processing means.

In this configuration, the R signal output from the color separation circuit 5 is input to the shift registers 13 and 14 in sequence. Therefore, if the R signal stored in the shift register 13 is for the pixel of interest, the shift register 14 receives the R signal one pixel before the pixel of interest, and the color separation circuit 5 sends the R signal one pixel after the pixel of interest. An R signal is obtained. Using the R signals of the pixel of interest and the pixels before and after it thus obtained, the average value R signal is determined by the arithmetic circuit 19. For the B signal, the arithmetic circuit 21 calculates an average value B signal using the B signals of the pixel of interest and the pixels before and after it, and for the G signal, the average value G is calculated by the arithmetic circuit 20 using the G signals of the pixel of interest and the pixels before and after it. Ask for a signal. Further, the calculation circuit 22 calculates the ratio between the G signal of the pixel of interest and the G signal calculated as an average value by the calculation circuit 20. The arithmetic circuit 23 corrects the average value R signal in proportion to the G signal ratio determined by the arithmetic circuit 22, and
' Output as a signal. Even in the arithmetic circuit 24, the arithmetic circuit 22
The B average value color B signal is corrected in proportion to the G signal ratio determined by , and is output as a B' signal.

That is, if the pixel of interest is a subscript n, the previous pixel is a subscript n-1, the next pixel is a subscript n+1, and the average color signal of the G signal is G1, the color signal obtained from the color conversion circuit 12 is The color signals R', B', and G' are as follows: R'=Gn/G1*(Rn-1+Rn+Rn+1)/3
B'=Gn/G1*(Bn-1+Bn+Bn+1)/3
G'=Gn G1 =(Gn-1+Gn+Gn+1)/3.

Human visual characteristics have low resolution for color information and high resolution for luminance information. Furthermore, indiscernible high-frequency information has a spatial integration effect. Therefore, for color information, if we average surrounding information and reproduce brightness,
It is possible to reproduce colors that are close to natural while maintaining clarity even at the edges.

Next, an embodiment of the invention according to claims 3 to 6 will be described with reference to FIGS. 6 and 7. The same parts as those shown in the previous embodiment are indicated using the same reference numerals. In the case of the embodiment described above, color shift may be noticeable when the difference in brightness of the image is large or depending on the edge position, but this embodiment eliminates this problem. Therefore, as shown in FIG. 6, an arithmetic circuit 26 is connected to the output side of the color separation circuit 5.

This arithmetic circuit 26 is constructed as shown in FIG. First, if we pay attention to the R signal output from the color separation circuit 5, we can see that the four stages of shift registers 27, 28, 29
, 30 are provided, and the signal Rn of the pixel of interest is determined by delay processing.
Signals Rn-2, Rn for two pixels before and after centering on
-1, Rn+1, and Rn+2. An averager performs an operation to obtain an average value by selectively inputting the signal Rn+2 from the color separation circuit 5 and the signals Rn+1, Rn, Rn-1, and Rn-2 from these shift registers 27, 28, 29, and 30. An adder circuit 31 is provided as a value calculation means. Similarly, four stages of shift registers 32, 33, 34, and 35 are provided for the B signal, and the signal Bn+2 from the color separation circuit 5 is
and signals Bn+1, Bn, Bn-1, and Bn-2 from these shift registers 32, 33, 34, and 35 are selectively inputted to calculate an average value. It is provided. Similarly for the G signal, four stages of shift registers 37, 38, 39 are used.
, 40 are provided, and the signal Gn+2 from the color separation circuit 5 and these shift registers 37, 38, 39 are provided.
, 40 signals Gn+1, Gn, Gn-1, Gn-2
An adder circuit 41 is provided as an average value calculating means for calculating an average value by selectively inputting .

[0022]Furthermore, the change as a detection means calculates the amount of change in both signals by inputting the signals Gn+1 and Gn-1 of the two pixels before and after the pixel of interest output from the shift registers 37 and 39 for G, which have a high arrangement density. A quantity detection circuit 42 is provided. The output of this change amount detection circuit 42 is given to each addition circuit 31, 36, 41 as a calculation pixel number selection signal. These adder circuits 31, 36, and 41 vary the number of peripheral pixels used to obtain the average value in accordance with this calculation pixel number selection signal. That is, the signal Gn+1
, Gn-1, the number of peripheral pixels for determining the average value is increased or decreased depending on the magnitude of the change between the pixels.

Now, assuming that the threshold values used in the change amount detection circuit 42 are S0 and S1 (however, S0<S1), when the change amount D is small such as D<S0, the addition circuit 31
Then, the signal Rn of the pixel of interest itself is converted to the average value RL as it is.
In the case of S0<D<S1, Rn-1, Rn,
Using the signal for 3 pixels of Rn+1, (Rn-1+Rn
+Rn+1)/3=RL, and calculate the average value as D
> In the case of S1, Rn-2, Rn-1, Rn, Rn+1
, Rn+2 using signals for 5 pixels (Rn-2+R
The average value is calculated as n-1+Rn+Rn+1+Rn+2)/5=RL. The same goes for other color signals B and G, and if D<S0, the pixel of interest is S0<D<S1.
In the case of , the average values BL and GL are calculated from the three pixels before and after the pixel of interest, and when D>S1, the average values BL and GL are calculated from the five pixels before and after the pixel of interest.

A color conversion circuit 43 is connected to the output side of such arithmetic circuit 26. This color conversion circuit 43 inputs the average value RL, BL, GL signals of each color from the calculation circuit 26 as well as the signal G of the pixel of interest itself, calculates the hue from the average value RL, BL, GL signals, and calculates the hue using the average value RL, BL, GL signals. The color is newly determined using the luminance information, and the final values of each color signal R', B', and G' are determined. This color conversion circuit 43 may have the same configuration as the color conversion circuit 12 shown in the above embodiment and perform similar arithmetic processing, or may have a look-up table configuration.

As described above, according to the present embodiment, in edge image portions that exhibit sharp changes due to color density changes in the image, the average is taken from pixels in a wide range of the preceding and succeeding images, and in smooth image portions, the image is output as is. Therefore, at edge areas where color shift occurs, the color becomes intermediate between the previous and subsequent images, making it difficult to see the color shift. Frequency information above the sampling frequency is expressed as an average value, so the visual integration effect makes it possible to change the color to what humans see. Similar colors can be extracted. Since images without edges are output as is, clear colors can be reproduced when outputting pixels close to pixels containing edges.

[0026]

[Effects of the Invention] The present invention focuses on the fact that human visual characteristics have a low resolution for color information, a high resolution for luminance information, and a spatial integration effect for indiscernible high-frequency information. , since it is configured as described above, claim 1 or 2
According to the invention described above, the color signal of the pixel of interest is determined by the average value calculation means as an average value color signal of the color signals of pixels before and after the pixel in its vicinity, and at the same time, the color signal of the pixel of interest is calculated as the average value color signal of the color signals of pixels before and after the pixel in its vicinity. A signal change amount based on the color signal information is detected by a specific color signal processing means and regarded as luminance information,
The correction means performs brightness correction to change the brightness of the average color signal corresponding to the color separation filter with low array density to reproduce the brightness, so the color shift that occurs at the edge of the image is corrected while maintaining the resolution. can be reduced.

On the other hand, in the invention as set forth in claim 3, the number of peripheral pixels used when calculating the average value is made variable by the average value calculation means, the amount of change in the color signal of the pixel of interest is detected by the detection means, and the detection result is For example, as in the invention described in claim 5, the number of pixels is increased in sharp image edge portions where the amount of change is large, and the number of pixels is decreased in gentle image portions where the amount of change is small, and the number of pixels is averaged. Therefore, in edge image areas where there is a lot of color shift, the color becomes intermediate between the previous and subsequent images based on the average value over a wide range of images before and after, making it difficult to see the color shift, and high frequency information higher than the sampling frequency is represented by the average value. Due to the visual integration effect, it is possible to extract colors that are close to those seen by humans, but for images that are not edge areas, the average of a small number of pixels or the signal of the target pixel itself is output, so pixels close to pixels that include edges are output. The output can reproduce clear colors.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a color conversion circuit showing an embodiment of the invention according to claims 1 and 2;

FIG. 2 is a front view of a line image sensor including a filter array.

FIG. 3 is a block diagram showing the overall configuration.

FIG. 4 is a block diagram of a color separation circuit.

FIG. 5 is a timing chart showing the operation.

FIG. 6 is a block diagram of the overall configuration of an embodiment of the invention according to claims 3 to 6;

FIG. 7 is a block diagram of an arithmetic circuit.

[Explanation of symbols]

1 Line image sensor 2 Color separation filters 19 to 21 Average value calculation means 23, 24 Brightness correction means 25 Specific color processing means 31, 36, 41 Average value calculation means 42
Detection means

Claims (6)

[Claims] Claim 1: Based on information read by a line image sensor in which color separation filters of a plurality of different colors are arranged with different arrangement densities depending on the color, with each filter covering each light receiving element. In a color image processing device configured to obtain each color signal of a pixel of interest, an average value calculation means is provided for calculating each average value color signal based on each color signal of the pixel of interest itself and each color signal of pixels surrounding the pixel of interest. , the color signal of the pixel of interest output from the light-receiving element compatible with color separation filters arranged in high density is output as is, and the amount of change in the signal is detected based on this color signal and the average value color signal of this color. Color signal processing means is provided, and brightness correction means is provided for correcting the brightness of each of the average value color signals of other colors using the signal change amount and outputting the color signal of the color. Image processing device. [Claim 2] R, G, and B color separation filters are arranged so that each filter covers each light-receiving element, and a G filter is arranged every other light-receiving element, and the R filter and the B filter are combined into a G filter. In a color image processing device, each R, G, and B signal of a pixel of interest is obtained based on the arrangement density of G filters from information read by line image sensors arranged alternately with respect to light receiving elements between the two. An average value calculation means is provided for calculating an average value color signal of each color based on each R, G, B signal of the pixel of interest itself and each R, G, B signal of pixels located before and after the pixel of interest. A G color signal processing means is provided which outputs the G signal of the pixel as it is and detects luminance information from the ratio of this G signal and the average value G signal, and uses this luminance information to output the average value R signal and the average value B signal.
A color image processing device comprising a brightness correction means for correcting the brightness of a signal and outputting an R signal and a B signal. 3. A pixel of interest is determined from information read by a line image sensor in which color separation filters of a plurality of different colors are arranged periodically, with each filter covering each light receiving element. In a color image processing device configured to obtain each color signal, a detection means is provided for detecting the amount of change in the color signal of the pixel of interest, and the number of pixels before and after the pixel of interest is varied according to the amount of change, thereby changing the color of the pixel of interest. A color image processing device comprising an average value calculation means for calculating an average color signal based on a signal and color signals of pixels before and after the pixel of interest. 4. The color according to claim 3, wherein the detection means detects the magnitude of a difference between color signals of one specific color obtained from two pixels located before and after the pixel of interest. Image processing device. 5. The detection means detects the amount of change in the color signal of one specific color among the color signals of the pixel of interest, and the number of pixels before and after the pixel of interest used for calculation by the average value calculation means is:
4. The color image processing apparatus according to claim 3, wherein the color image processing apparatus is varied so that when the amount of change is larger than a predetermined value, the amount of change is increased and when the amount of change is smaller than a predetermined value, it is decreased. 6. The color separation filter is arranged in an array with different arrangement densities depending on the color, and a detection means is provided for a color signal corresponding to the color separation filter with a higher arrangement density. 6. The color image processing device according to 4 or 5.