JPS62281064A - Color image storage device - Google Patents

Abstract

PURPOSE:To attain compression and to decrease the capacity of a memory part without reducing an image quality by storing an image signal in the shape of luminance, hue and saturation. CONSTITUTION:An RGB signal, after it is A/D-converted, is stored respectively to image memories 5, 6 and 7, matrix-calculated by a matrix arithmetic circuit 8 and converted to Y to show luminance and R-Y and B-Y as color difference signal. For these luminance and color difference signals, a prescribed conversion is executed by converting tables 9 and 10 and the signals are converted to luminance Y', hue H and saturation C. For the luminance conversion of the converting table 9, in order to match the dynamic range of an input image signal and the reproducing range (for example, ink) of an output, the luminance Y is normalized, the luminance Y' is obtained, and normalization is executed by the conversion from the Y to the Y' and compression is executed. The color difference conversion of the converting table 10 is the conversion to prevent the color reproducibility from being lost and compress the dynamic range in the saturation C direction.

Description

Detailed Description of the Invention 3. Detailed Description of the Invention [Field of Industrial Application] The present invention relates to a color image storage device, such as a frame memory, for storing color image signals.

[Prior Art] In conventional color video printers, for example, C, M,
Y, the signal was mainly processed. An example is shown in FIG. The A/D converted RGB signal is LOG converted and becomes cy (cyan).

It is converted into M (magenta), Y, (yellow) signals, and is further subjected to masking correction to produce C,', M'.

The signal becomes Y0° and is input to the head driver, which drives the head and prints. Masking correction is based on the following matrix calculation.

However, the drawback of this method is that good color reproduction cannot be obtained. In other words, due to the difference between the dynamic range of the input image signal and the reproduction range of the output ink, simply performing logarithmic conversion may result in a density of 3.0 or more, and in such cases, it is natural that the printing The maximum concentration will be exceeded. In addition, a method called gamma conversion is used to correct the density range, but this has the disadvantage that the saturation changes and the color reproduction differs from the original image.

Furthermore, not only the above-mentioned color printer but also a color display device requires a frame buffer for each image signal in order to display in color as shown in FIG. 13. In this case, the problem is what kind of image signal should be stored in the frame buffer. That is, in the conventional example shown in FIG. 13, the RGB signal is converted into a luminance signal Y and two color difference signals (R-Y, B-Y) by matrix conversion,
, RY, and BY signals are stored, subjected to necessary image processing, and converted back into RGB signals for display on a CRT device (not shown), for example, and then output. The problem here is storing the color image signal in a frame buffer, for example, but if it is not possible to reduce the number of pixels in order to save storage capacity, then the bits in the depth direction of the color image signal There is no choice but to reduce it. However, conventionally, for example, each of the two color difference signals has six bits in the depth direction.
If we use bits, there is a problem that the color reproduction will already change from the original image, and for this reason, 8 bits are absolutely necessary to maintain good color reproduction. Under these circumstances, the inability to reduce the amount of memory resulted in an increase in the memory portion and increased costs.

[Problems to be Solved by the Invention] The present invention was proposed to solve the above-mentioned problems of the prior art, and its purpose is to reduce the capacity of the image area without degrading the image quality. The object of the present invention is to propose a color image storage device.

[Means for Solving the Problems] The configuration of the present invention for achieving the above-mentioned problems is such that, regarding a color image signal consisting of a luminance signal and two color difference signals, the two color difference signals are converted into a hue signal and a saturation signal. , a compression means for compressing the dynamic range of the image signal, and a storage means for storing the image signal.

[Operation] In the present invention having the above configuration, the compression means independently compresses and outputs the dynamic range of either or both of the luminance signal or the chroma signal, and the storage means compresses and outputs the dynamic range of either or both of the luminance signal and the saturation signal, and the storage means compresses and outputs the dynamic range of either or both of the luminance signal and the saturation signal, and the storage means compresses and outputs the dynamic range of either or both of the luminance signal and the saturation signal, and the storage means Try to remember.

[Examples] Examples according to the present invention will be described in detail below with reference to the accompanying drawings.

(Outline of Embodiment) As a color processing method that achieves good color reproduction, there is an image processing method that processes brightness and color difference. FIG. 1 shows an embodiment in which the basic processing block configuration is applied to a color video printer. After the RGB signals are A/D converted, they are stored in the image memory 5.6゜7, and are subjected to rematrix calculation by the matrix calculation circuit 8 to produce Y1 which indicates the brightness.
It is converted into R-Y and B-Y as color difference signals.

These luminance and color difference signals are subjected to a predetermined conversion described later using conversion tables 9 and 10, and are converted into luminance (Y"), hue (H), and saturation (C).

The brightness conversion in step 9 changes the brightness Y to match the dynamic range of the input image signal and the reproduction range of the output (for example, ink).
is normalized to obtain the brightness Y°. In the conversion from Y to Y', normalization is performed and compression is also performed. or,
The color difference conversion No. 10 is a conversion that compresses the dynamic range in the chroma C direction without losing color reproducibility. In this embodiment, both luminance conversion and color difference conversion are performed as shown in Fig. 1, but from the viewpoint of maintaining color reproducibility and compressing the dynamic range, only one of them can be performed independently. It will become clear later in the explanation that the same effect can be obtained from each of these methods.

Now, the luminance (Y'), hue (H), and saturation (Co) are determined by the color correction conversion table 11 into Cy and M.

The signal is converted to Ye, further subjected to masking correction for asymmetric color component correction of the ink, inputted to the head driver 12, and printed. The outline of this embodiment is as above. Therefore, details of the embodiment will be explained next.

<Conversion to luminance and one color difference> After the RGB signals are A/D converted, they are stored in a memory 5°6.7 and subjected to matrix calculation. The matrix operation is a 3×3 matrix operation shown in the following equation.

The main reason for storing the image signal in the memory is that the luminance conversion described later requires one screen worth of images in order to examine the dynamic range in the input image signal of one screen. Therefore, in the embodiment shown in FIG.
However, if higher-speed processing is required, the memories 5 and 6.7 can be moved to the stage after the matrix operation 8 and the input RGB signals can be converted directly. Matrix calculation may be performed to convert the luminance into two color differences, and this one luminance color difference may be stored in the memories 5, 6.7.

The freezing of the image in the memory as described above is triggered by the freeze switch 21 on the console 20.

<Brightness Signal Conversion (Gradation Conversion)> As described above, the brightness signal conversion 9 is performed to fill the difference between the input dynamic range and the output reproducible range. However, when compensating for this difference, it is desirable not to narrow the input dynamic range as much as possible. To this end, in this embodiment, the dynamic range of the luminance Y signal of the input image signal is investigated, and the conversion characteristic most suitable for the input is selected.

The image fixed by the freeze switch 21 is examined by the CPU 2 for its dynamic range regarding brightness. This can be checked by sequentially reading data from the memory 5.6.7 and taking a brightness histogram as shown in FIG. The histogram is stored in the RAMI, but in this case, it takes time to scan all the pixels in the memory 5, so it may be thinned out, for example every 5 pixels, according to the sampling shown in FIG. 9(a). . Since one screen must be captured within 1/30 seconds, the ROM constitutes the A/D converter, conversion table, etc.

Memory needs to operate at high speed.

The dynamic range of brightness is captured by the highlight point (Hl) and dark point (DP) in the histogram. The highlight point (HP) is defined as the brightness of a point 1% from the brightest point, and the dark point (DP) is similarly defined as the brightness of a point 1% from the darkest point.

On the other hand, the luminance conversion table 9 allows conversion of several types of gradation conversion characteristics as shown in FIG. Which characteristic to choose is done as follows. If the highlight point (HP) and dark point (
DP) are close to HPl + DPI in Figure 4, the most appropriate conversion characteristic that does not impair the dynamic range of the input image is a conversion characteristic that connects HPI - DPI in Figure 4. be. On the contrary, HP4
With characteristics that connect DPI, reproduction will be lost on the higher luminance side. The brightness conversion table 9 having several types of characteristics is formed of, for example, a ROM, and the number of conversion tables can be increased or decreased depending on its capacity. Then, the ROM may be appropriately addressed based on the values of the highlight point (Hp) and dark point (DP) of the histogram.

Characteristics of luminance conversion> In this way, the dynamic range in the luminance direction is properly compressed (so-called normalization processing), and a faithful output image that cannot be obtained by conventional unified logIR conversion is reproduced. Ru. In other words, gradation correction has been performed on the output ink.

Also, since this luminance converted luminance signal Y° is compressed, if it is necessary to store the image signal in a frame memory, etc., the luminance signal Y° after the luminance conversion is better than the luminance Y after the matrix calculation 8. The luminance signal Y° has a superior feature in that the storage capacity can be saved.

Furthermore, if only the luminance is compressed without compressing the dynamic range in the chroma direction, which will be described below, the chroma C can be better preserved and good color reproduction can be achieved. Furthermore, since table conversion is used, it also has features such as being able to prevent enlargement of the circuit scale.

Although the above embodiment has been described using a video printer with a frame memory as an example, it can be implemented without a frame memory if a line memory is used.

<Conversion from color difference to 7 levels of saturation> As shown in Figure 5, the dynamic range of the RGB input signal is considerably larger than the reproduction range of the saturation direction of the C, M, and Y signal inks. It is necessary to compress the image signal in the direction of saturation. For this purpose, in this example,
In the color difference conversion table 10, the color difference (R-Y, B-Y) is converted to saturation C1 hue H, and further compression is performed in the direction of saturation to obtain Co from C. First, conversion from color difference to hue and saturation is performed using the following equation.

However, if R'-Y=0, H=O.

C (chroma)=R−Y+B−Y The calculation results of the above formula for each combination of values of RY and BY are stored in the conversion table 1o. However, as mentioned above, the dynamic range in the direction of the RGB input signal is compared to the reproduction range in the chroma direction of the cy (cyan) signal, M (magenta) signal, Y, , (yellow) signal ink, and the J: saturation direction. is quite large, so even if it is compressed, the color reproducibility during printing will not deteriorate that much. The simplest compression characteristic is shown in FIG. This is based on the fact that in a normal image iJ that exists in the natural world, most of the images are concentrated in the low-saturation areas, so the high-saturation parts are clipped. Note that Co in the figure is obtained by further converting the above-mentioned C (saturation)=RY+BY using the conversion table 10.

If the above conversion compresses all of the input chroma signal C, which is larger than the output chroma signal C', into Co, a problem arises because it becomes impossible to express the difference in chroma in that part. In order to prevent this, it may be possible to provide a conversion characteristic as shown in FIG. 7, but this would reduce the saturation of all pixels, making it impossible to obtain good color reproduction. Therefore, it is conceivable to provide nonlinear characteristics as shown in FIG. This prevents the saturation from decreasing in most pixels, that is, the reproducibility of the saturation is not lost. Furthermore, the chroma signal C' can be generated without losing the difference in chroma in the high chroma portion. This is because in a normal natural image, the image is concentrated in a region of relatively low saturation.

<Normalization of Saturation> In order to compress to a more appropriate dynamic range, the normalization described in the luminance conversion may also be used for compression of saturation. That is, a saturation histogram is obtained from the memory 6.7.

Suppose that the histogram is as shown in FIG. 10(a), for example. The maximum chroma obtained in the same manner as in the case of luminance compression is set as C□, and the minimum chroma is set as CL. Based on these values, select from among the multiple characteristics shown in Fig. 10(b). Be sure to choose the optimal compression/conversion characteristics. In this way, optimal color reproduction can be obtained. however,
Since selecting appropriate conversion characteristics from the histogram in this way tends to slow down the processing, a high-speed memory element is required. Incidentally, examples of sampling color difference signals for obtaining a histogram at high speed are shown in FIGS. 9(b) and 9(c).

<Color correction conversion> The Y', H, and C' signals obtained in this way are input to the color correction conversion table 11 that receives these three signals as input, and the C'
, M', Y' times signal conversion. The method of calculating the values stored in this table is basically the inverse of the above-mentioned conversion, and is as follows.

(RY)'=C' xCos H (B-Y)'=C' X5in H Furthermore, R' = (RY)'+Y'G' = -(0,3(RY)' +0 .11 (B-Y
)') +Y"B' = (B-Y)'+Y'
Next, C,=-1ogR'M=-1ogG' Y,=-1og B' Then, matrix approximation for correction of ink asymmetric color components is performed. Here, 811 to a33 are constants.

<Characteristics of Saturation Compression> According to the above embodiment, not only can the difference in saturation between input and output be effectively compensated for, but also the maximum amount of saturation signal can be reduced, so even with the same bit allocation, it is possible to It is quantized and has the effect of increasing efficiency. That is, storage efficiency due to compression increases.

In the above embodiment, a video printer was used as an example, but any system related to color image correction may be used, such as a configuration system for printing. In addition, in this example, after A/D conversion, the signals are converted into Y, RY, B-Y signals by matrix calculation, but after performing matrix conversion before A/D conversion, Y, RY, The BY signal may be A/D converted and used as a digital signal.

<Color image storage device> From what has been described above, the embodiment shown in FIG. It is understood that the problem ``For example, if the bits in the depth direction of two color difference signals are set to 6 bits each, the original image and color reproduction will change'' will be solved at the same time. . That is, even if memory is saved, the savings are compensated for by compression, and for the reasons mentioned above, no problems occur in color reproduction.

FIG. 11 shows an embodiment from the viewpoint of saving memory. The major difference from the embodiment shown in FIG.
~31. These memories 30 to 31 are, for example, frame memories (frame buffers) for image output.

As mentioned above, conventionally there was a problem in reducing the memory in this part, but by compressing the luminance signal or chroma signal in this embodiment, the memory can be saved.

By the way, there have been several examples of luminance signal compression as mentioned above, but since the memory 5 is essentially necessary to obtain a histogram, it is necessary to fix the characteristics as shown in Figures 6 and 7, for example. If you choose wisely, you won't need the memory 5. For the same reason, the memory 6.7 is also unnecessary if the compression does not require a saturation histogram.

Further, in this embodiment, the number of pixels is set to the standard classification in NTSC (640×480), but it is of course possible to reduce the number of pixels. In particular, if the color correction table 11 in FIG. 1 or 11 is configured as shown in FIG. 12,
It should be possible to significantly reduce color information.

[Effects of the Invention] As explained above, according to the present invention, the image signal is
By storing images in the form of hue and saturation, compression becomes possible, and as a result, the capacity of the memory unit can be reduced without deteriorating image quality.

[Brief Description of the Drawings] Figure 1 is a block diagram of an embodiment in which the present invention is applied to a color video printer, Figures 2 and 13 are diagrams explaining conventional examples, and Figure 3 is a diagram of a luminance signal. Figure 4 is a diagram showing an example of a histogram, Figure 4 is a characteristic diagram of various conversion characteristics regarding luminance signals, Figure 5 is a diagram explaining the difference in dynamic range between RGB system and Y, MC, system image signals, Figure 6 , Fig. 7, Fig. 8
The figure is a characteristic diagram of saturation compression characteristics. Figures 9 (a) to (C) are diagrams explaining pixel sampling when obtaining a histogram. Figures 10 (a) and (b) are about saturation. A diagram explaining the case of taking and compressing a histogram, FIG. 11 is a block diagram of another embodiment from the viewpoint of reducing the storage area, and FIG. 12 is a circuit diagram of another example of the configuration of the color correction conversion table. It is. In the diagram, 1.RAM, 2.CPU, 3.ROM, 4.Bus,
5, 6, 7, 30, 31. 32...Memory, 8...
Matrix calculation unit, 9... Brightness conversion table, 10
. . . color difference conversion table, 11 . . . color correction conversion table, 12 . . . head driver, 13 . . . head. Patent Applicant: Canon Co., Ltd. Han P Award Anatomy CR, O's bite # capital 1 @ buds 2 44〉- 0 koreuguri Esu

Claims (2)

[Claims] (1) Concerning a color image signal consisting of a luminance signal and two color difference signals, a conversion means for converting the two color difference signals into a hue signal and a saturation signal, and either one of the luminance signal or the saturation signal, or A color image storage device comprising: compression means for independently compressing and outputting both dynamic ranges; and storage means for storing the hue signal and the output of the compression means. (2) The color image processing apparatus according to claim 1, wherein the dynamic range is compressed based on the frequency distribution of the input.