JPH08202327A - Image processor - Google Patents

Abstract

PURPOSE: To assign the number of colors to pallet of respective partial color spaces appropriately even in the case a color whose appearance frequency is dominated in a partial color space is present in an image processor forming pallets based on the image data of a color original image. CONSTITUTION: In a frequency modifying means 13, frequencies of respective color included in respective partial color space histograms are inspected and in the case frequencies are larger than the average appearance frequency of all appearance colors, the frequencies are modified on the basis of the average appearance frequency. In the number of pallet registered color determining means 14, the number of colors assigned to pallet of respective partial color spaces is determined based on modified partial color space histograms. In a pallet registering means 15, registration to pallets is performed successively in the order of a color having higher frequency of appearance every partial color space histogram according to the number of determined assignment colors.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a pallet used for outputting color information of a full-color original image in a limited color to an image output device such as a display or a printer. The present invention relates to an image processing apparatus that creates a palette based on pixel data of a color original image (hereinafter, also referred to as an original image).

[0002]

2. Description of the Related Art In recent years, full-color display is becoming common in image output devices such as displays and printers. When performing full-color display, if one pixel is represented by 24 bits, it is possible to display approximately 16.7 million colors at the same time, but the amount of data becomes enormous. Therefore, by expressing one pixel by 8 bits, the number of colors that can be used at the same time is set to 256 colors, and color information of these 256 colors is registered in a palette and an image is often output. However, since about 16.7 million colors and 256 colors are selected and registered in the palette, it is necessary to select the colors so that the deterioration of the image quality is not felt.

As a general method of creating a palette, there is a method of examining the appearance color and its frequency in the original image from the color original image, and registering in the palette in order from the appearance color with the highest frequency. However, in this method, if there is a similar color that changes subtly, or if there is a color that varies due to noise during scanning, etc., the frequency of that color is dispersed,
There was a problem that the necessary colors would not be registered in the palette. The problems of pallet creation by this conventional method will be described below.

FIG. 13 shows an example of all appearance color histograms obtained by scanning all pixels of a color original image, where the horizontal axis represents the appearance color and the vertical axis represents the appearance frequency. It should be noted that four colors are registered in the palette here for the sake of simplicity.

Normally, as shown in FIG. 13, R,
The four colors of G, B, and Y will be registered in the palette as limited colors. However, as shown in FIG. 14, when the frequency of B color is dispersed to b1, b2, b3 due to noise or the like, P color becomes higher in frequency instead of B color, so that the palette has limited colors. Since R, G, B, and P are registered, the color tone of the output result will be uncomfortable. Such a problem also occurs when similar colors are dispersed.

On the other hand, in Japanese Unexamined Patent Publication No. Hei 4-257984, the color of the partial color space is divided into several partial color spaces, and the partial color space is assigned to a palette according to the total ratio of the appearance frequencies of the partial color spaces. The number of colors is determined, the palette is created by allocating as many colors as the determined number of colors in order from the highest frequency of colors in each partial color space, and by using this palette, the output result of the hue close to the original image is obtained. There has been proposed an image output device capable of obtaining the following.

[0007]

However, in this image output apparatus, if there is a color having a prominent appearance frequency in the original image, the limited color more than necessary is assigned to the partial color space to which the color belongs. There was a problem that the output result was different from the image. Hereinafter, this problem will be described in detail.

Now, assume that an all appearance color histogram as shown in FIG. 15 has been obtained from the color original image. To simplify the description, it is assumed here that 10 colors are registered in the palette and 4 partial color spaces are formed.

Now, if the all appearance color histogram of FIG. 15 is equally divided into color spaces, as shown in FIG. 16, partial color spaces of R partial color space, G partial color space, B partial color space and Y partial color space are obtained. A histogram is obtained. Then, as shown in FIG. 17, the ratio of the appearance frequency meter (hereinafter, referred to as frequency meter) of each partial color space histogram to the total frequency is calculated, and from the obtained ratio, each part that can be registered in the palette is calculated. Determine the number of colors assigned to the color space. Here, the limited number of colors is 10, and the number of assigned colors of each partial color space is R ... 3,
G ... 2, B ... 2, Y ... 1. If all the appearance color histograms of FIG. 15 are color-coded based on the assigned number of colors, FIG.
It becomes like 8.

By the way, in the partial color space histogram of B of FIG. 16, there is a b1 color (frequency of 12) having a prominent frequency. Therefore, the assigned number of colors of the B partial color space for this b1 color. And the b2 color is also selected from the B partial color space. As a result, as shown in FIG.
The two colors are registered in the palette, and the y2 color that should be originally selected is removed from the registration.

As described above, according to the conventional palette creating method,
If there is a color with an extremely high frequency, the allocation of the number of colors to be assigned to the palette will be improper due to the influence of that color, and as a result, an output result with a discomfort in the hue etc. with respect to the original color image can be obtained. become.

SUMMARY OF THE INVENTION It is an object of the present invention to provide an image processing apparatus capable of appropriately distributing the number of colors to be assigned to a palette for each partial color space even when there is a color whose frequency is outstanding. .

[0013]

In order to solve the above problems, an image processing apparatus according to the present invention comprises an all appearance color histogram creating means for creating an all appearance color histogram based on pixel data of a color original image. A partial color space histogram creating means for creating a partial color space histogram for each partial color space obtained by equally dividing a color space based on the all appearance color histogram created by the all appearance color histogram creating means; Based on the all appearance color histogram created by the color histogram creating means, the average appearance frequency of all appearance colors is obtained, and the frequency of each color included in each partial color space histogram is equal to or more than the average appearance frequency. , Modify the frequency to the average appearance frequency,
Frequency correction means for creating a modified partial color space histogram for each partial color space histogram, and a frequency meter for the colors in each partial color space is obtained from the modified partial color space histogram created by the frequency correction means, and this frequency meter And a palette registration color number determining means for determining the ratio of the frequency of appearance colors in each partial color space, and determining the number of colors assigned to the palette for each partial color space based on this ratio, According to the number of assigned colors determined by the palette registration color number determining means, there is provided palette registration means for performing registration in the palette in order from the color having the highest frequency for each of the partial color space histograms.

[0014]

According to the above-mentioned image processing apparatus, the frequency correction means inspects the frequency of each color included in each partial color space histogram, and if the frequency is equal to or higher than the average appearance frequency of all appearance colors, the frequency is averaged. It is modified based on the appearance frequency. Then, the palette registration color number determining means determines the number of colors to be assigned to the palette for each partial color space based on the corrected partial color space histogram.

Therefore, even when there is a color having a prominent frequency in a specific partial color space, the colors in the partial color space are allocated more than necessary to the palette when determining the number of colors to be allocated. Therefore, it is possible to obtain an output result in which there is little discomfort in the hue and the like with respect to the original image.

[0016]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of an image processing apparatus according to the present invention will be described below with reference to the drawings.

The block diagram of FIG. 1 shows the functional configuration of the image processing apparatus in this embodiment. The image processing apparatus 10 includes an all appearance color histogram creating unit 11, a subspace histogram creating unit 12, and a frequency correcting unit 13.
And a palette registration color number determining means 14 and a palette registration means 15.

The all appearance color histogram creating means 11 comprises an image input means (not shown). The image input means scans the color original image to input the pixel data of all pixels, and based on the pixel data. Create a histogram of all appearance colors.

The partial color space histogram creating means 12 is
Based on the all-appearance color histogram created by the all-appearance color histogram creating means 11, a partial color space histogram is created for each partial color space, which is obtained by equally dividing the color space.

The frequency correction means 13 is based on the all-appearance color histogram created by the all-appearance color histogram creating means 11 and based on the average appearance frequency of all the appearance colors, the modification frequency R.
Ask for. Then, the frequency of each color included in each of the partial color space histograms is inspected in order, and when the frequency exceeds the correction frequency R, the frequency of the color is corrected to the correction frequency R. By making such a correction, a corrected partial color space histogram is created for each partial color space histogram.

The palette registration color number determination means 14 obtains a color frequency meter for each partial color space from the corrected partial color space histogram created by the frequency correction means 13. And
Based on this frequency meter, the ratio of the frequency of appearance colors in each partial color space is obtained, and the number of colors assigned to the palette for each partial color space is determined based on this ratio.

The palette registration means 15 determines the number of assigned colors determined by the palette registration color number determination means 14 according to
For each of the partial color space histograms, registration is performed in the palette in order from the color with the highest frequency.

The image processing apparatus 10 described above is applied to, for example, a computer system having a color display such as a workstation or a personal computer, a printer apparatus capable of color printing, and the like.

FIG. 2 shows a concrete example for realizing the image processing means 10 shown in FIG. 1, and shows, for example, the hardware configuration of a workstation having an image processing function.

The display 21 is a color display device composed of a CRT and displays various documents as well as image images such as figures and tables on the screen. The display on the display 21 is controlled by the display control unit 22.

The keyboard 23 is an input device for inputting data such as commands and character strings. The contents of the selection instruction input from the mouse 24 are the keyboard / mouse control unit 25.
Through a processor unit 30 described later.

The hard disk 26 is composed of a secondary storage device such as a magnetic disk, and stores various data in a file format.

The real memory 27 is a buffer memory composed of a memory device such as a RAM, and temporarily stores various programs and various data and commands input from the keyboard 23 and the mouse 24 in addition to various programs.

The pixel data storage memory 28 is a bit map memory in which pixel data to be displayed on the display 21 is written, and pixel data of a color original image read by an image scanner 29 described later is accumulated.

The image scanner 29 is an image input means for scanning a target color original image and converting it into pixel data.

The processor section 30 is a central processing unit composed of a CPU and its peripheral circuits, and controls / manages the operations of the above sections in accordance with a control program. This processor unit 30 is configured as shown in FIG.
The functions of the all appearance color histogram creating means 11, the partial space histogram creating means 12, the frequency correcting means 13, the palette registration color number determining means 14, and the palette registration means 15 are realized.

Next, an outline of the palette creation processing in the image processing apparatus 10 described above will be described with reference to the flowchart of FIG. 3 and the specific examples of FIGS.

First, all appearance color histogram creating means 11
At, while scanning all pixels of the color original image,
Based on the pixel data, an all appearance color histogram is created (step 101). In the partial color space histogram creating means 12, the all appearance color histogram creating means 1 is
A partial color space histogram for each partial color space obtained by equally dividing the color space is created based on the all appearance color histogram created in step 1 (step 102).

As described above, a partial color space histogram is created for the all appearance color histogram shown in FIG.
This is equivalent to obtaining a partial color space histogram for each partial color space of R, G, B, and Y as shown in FIG.

In the frequency correction means 13, the average appearance frequency of all the appearance colors is obtained based on the all appearance color histogram created by the all appearance color histogram creation means 11, and the correction frequency R is obtained, and at the same time, each part The frequency of each color included in the color space histogram is sequentially checked. Then, when the frequency is equal to or higher than the correction frequency R, the frequency of the color is corrected to the correction frequency R to create a corrected partial color space histogram (step 103).

The correction frequency R in the partial color space histogram shown in FIG. 4 is as follows: correction frequency R = (total appearance frequency) ÷ (sum of frequency meters of all partial color spaces) = 61 ÷ (5 + 4 + 4 + 3) = 61 ÷ 16 = 3.81∴R = 4 Therefore, in FIG.
The color, g1 color, g2 color, and b1 color are the correction targets. Then, by correcting the frequency of these colors with the average appearance frequency 4, a corrected partial color space histogram as shown in FIG. 5 is obtained. In the modified partial color space histogram of FIG. 5,
Colors with a high frequency in the partial color space, such as b1 color, have been corrected.

The palette registration color number determining means 14 obtains a color frequency meter in each partial color space from the corrected partial color space histogram created by the frequency correcting means 13. Then, based on this frequency meter, the ratio of the frequency of appearance colors in each partial color space to the total appearance frequency (total appearance frequency) is calculated, and the number of colors assigned to the palette for each partial color space is calculated. It is determined (step 104).

Here, when the frequency meter and ratio of each partial color space are obtained from the corrected partial color space histogram of FIG. 5, the contents are as shown in FIG. 6, and the number of assigned colors of each partial color space in the limited number of colors 10 is obtained. Becomes R ... 3, G ... 2, B ... 1, Y ... 1.

Next, the palette registration means 15 performs registration in the palette in order from the color with the highest frequency for each of the partial color space histograms according to the assigned color number determined by the palette registration color number determination means 14. (Step 10
5).

FIG. 7 shows a color-coded histogram of all appearing colors in FIG. 15 based on the number of assigned colors determined in step 104. As described above, in the conventional example of FIG. 18, the b2 color that is less frequently registered is registered in the palette, and the y2 color that should be originally selected is removed from the registration. However, in this embodiment, since the frequencies of the r1 color, b1 color, etc., whose frequencies are prominent in the partial color space are corrected, the y2 color is registered in the palette as shown in FIG.
Two colors have been removed from the palette registration. Therefore, an output result that does not cause discomfort in the hue or the like with respect to the original image is prevented from being assigned to the palette of the partial color space more than necessary due to a part of the colors with the protruding frequency existing in the partial color space. Can be obtained.

Next, a concrete example of the above-described palette creating process will be described with reference to the flowcharts of FIGS. 8 and 9 and FIGS.
Will be described in more detail with reference to.

First, the above-mentioned processing in the first half of FIG. 3 will be described. FIG. 8 is a flowchart showing the procedure until the all appearance color histogram and the partial color space histogram are created, which corresponds to steps 101 and 102 in FIG.

First, when creating the all appearance color histogram and the partial color space histogram, these histograms are initialized (step 201). The initialization of each histogram is performed by the all appearance color histogram creating means 11 in FIG.
Is executed by the partial color space histogram creating means 12.

The next steps 202 to 205 are
The loop when creating the all appearance color histogram is shown and corresponds to step 101 in FIG. Here, the all-appearing-color histogram creating unit 11 scans the pixels of the color original image in the order shown in FIG. 10, and accumulates the colors of the pixels and their frequencies to create the all-appearing-color histogram.
These processes are performed by accumulating the pixel data of the scanned original image in a memory (not shown).

First, the start pixel (upper left corner) of the original image (original image) is set as the scan target pixel G (step 202). Then, the color of the scan target pixel G is checked, and if the color is already registered in the all appearance color histogram, the frequency is counted up by one. If the color is not registered in the all appearance color histogram, the frequency is set to 1 and the RGB of that color is set.
The value is registered in the all appearance color histogram (step 20).
3).

Subsequently, it is determined whether the scanning target pixel G is the end pixel (step 204). Here, as shown in FIG. 10, the scan target pixel G is the end of the lower right corner of the original image.
If it is a pixel, it exits the loop and proceeds to step 206, where en
If it is not the d pixel, the next pixel is set as the scan target pixel G (step 205), and the process returns to step 203 to continue the loop processing. By the processing from step 203 to step 205, the all appearance color histogram having the contents shown in FIG. 11 is obtained. In FIG. 11, index represents a number given in descending order of frequency.

The next steps 206 to 208 are
It is checked whether the total number N of appearance colors is less than the number L of colors that can be registered in the palette, and the steps when the number is small are shown.
This step is a part not described in the outline flowchart of FIG.

In step 204, the scanning target pixel G is end.
When it becomes a pixel, the number of colors registered in the total appearance color histogram, that is, the total appearance color number N is obtained (step 20).
6). This total appearance color number N corresponds to N of index in FIG. Then, it is determined whether the total number N of appearance colors is less than or equal to the number L of colors that can be registered in the palette (step 207). Here, when the total number N of appearance colors N ≦ the number L of colors that can be registered in the palette (step 207: Y), the total number N of appearance colors can be registered in the palette and it is necessary to allocate the number of colors to be assigned to the palette. Therefore, the total number N of appearance colors is registered in the palette (step 208). This completes the palette creation process. Further, when the total number N of colors that appear is greater than the number L of colors that can be registered in the palette (step 207: N), it is necessary to distribute the number of colors to be assigned to the palette. Proceed to the step of color registration to the palette using the appearance frequency meter.

First, the subspace histogram creating means 12
Then, the correction frequency R is obtained based on the entire appearance color histogram (step 209). The correction frequency R is a value obtained by rounding up the fractional part of the average appearance frequency of all appearance colors as described with reference to FIG.

The next steps 210 to 213 are
The loop when creating each partial color space histogram is shown, and corresponds to step 102 in FIG. By rotating around this loop, the partial color space histogram is obtained in the form as shown in FIG. FIG. 12 schematically shows the relationship between the all appearance color histogram and each partial color space histogram.
It shows that four partial color spaces are created.

First, the index "k" for the loop is initially set to "k ← 1" (step 210). This "k" means the index value of the all appearance color histogram shown in FIG. Next, for the k-th color of the all appearance color histogram, the upper 2 bits of the RGB value are checked, and the frequency and the RGB value of the color are registered in the corresponding partial color space histogram. Thereafter, it is determined whether k = N, that is, whether all the appearance colors have been registered in the partial color space histogram (step 212). Where k
When ≠ N (step 212: N), k is incremented by 1, that is, the value of index is moved to the next color of the all appearance color histogram, and the process returns to step 211 and loops. When k = N (step 212:
Y), proceed to the step of modifying the next partial color space histogram.

Next, the latter half processing of FIG. 3 will be described. FIG. 9 is a flowchart showing the procedure for correcting the partial color space histogram and registering it in the palette, which corresponds to steps 103 to 105 in FIG.

Steps 214 to 217 in FIG.
Shows a frequency correction loop of each partial color space histogram, and corresponds to steps 103 and 104 in FIG.

The frequency correction means 13 checks the registered colors of each partial color space histogram in descending order of frequency, and step 209
If there is a color with a frequency greater than the correction frequency R obtained in
The frequency of the color is corrected to the correction frequency R (step 21).
4). Here, the correction process is repeated until there is no more color than the correction frequency R for each partial color space histogram, and at the time when there is no color, the process moves to the next partial color space histogram and the same operation is performed. By this correction processing, a corrected partial color space histogram (not shown) is created for each partial color space histogram.

The palette registration color number determining means 14 determines from the corrected partial color space histogram obtained in step 214
A frequency meter hi (h0, h1, h2, ..., h63) of the colors in each partial color space is obtained (step 215), and from this frequency meter hi, the appearance color in each partial color space for all appearance frequencies is calculated. Frequency ratio
% i (% 0,% 1,% 2, ...,% 63) is calculated (step 216). At this time, the part after the decimal point is discarded (see FIG. 6). Next, from the ratio% i of the frequency of appearance colors in each partial color space, the number of colors li (l0, l) assigned to the palette for each partial color space
, L2, ..., L63) are determined (step 217). At this time, the part after the decimal point is discarded (see FIG. 6).

The following steps 218 to 222 are
FIG. 10 shows a loop for registering from the partial color space histogram to the palette according to the determined number of colors assigned to the palette for each partial color space, and step 105 in FIG.
Is equivalent to.

First, the index "i" for the loop is initially set to "i ← 0" (step 218). This "i" indicates the unillustrated indexes (0 to 63) of the 64 partial color space histograms shown in FIG.

The palette registration means 15 registers, from the histogram of the partial color space i, the number li of colors to be assigned to the palette determined in step 217, in order from the color with the highest frequency to the palette (step 219). Subsequently, the color newly registered in the palette in the previous step 219 is deleted from the all appearance color histograms (step 220). This is done to prepare for pallet replenishment when unregistered pallets occur. Then, it is determined whether i = 63, that is, whether palette registration has been completed for all 64 partial color space histograms (step 221). Here, when i <63 (step 221: N), i is incremented by 1 to shift to the next partial color space histogram, and the process returns to step 219 to loop. When i = 63 (step 221: Y), the process proceeds to the step for replenishing the next unregistered palette.

The next steps 223 to 225 are
This is a step of color replenishment when there is an unregistered portion remaining in the palette, and is a portion not described in the outline flowchart of FIG.

First, in order to check whether or not there is an unregistered part in the palette, the number of colors that can be registered in the palette L is assigned to the palette for each partial color space (l0, l1, l2, ..., l63).
Is subtracted from the variable m (step 223), and m> 0
It is determined whether or not (step 224). Where m =
When it is 0 (step 224: N), there is no unregistered portion in the pallet, so the process ends. When m> 0 (step 224: Y), m colors have not been registered in the palette, so the color of the palette is supplemented.

As described above, since the color newly registered in the palette is deleted from the all-appearing color histogram in step 222, the all-appearing color histogram is a histogram in the frequency order of colors not registered in the palette. Has been fixed. Therefore, m colors are registered in the palette from the highest frequency of all the corrected appearance histograms (step 225). As a result, colors are registered in all parts on the palette.

According to the flowcharts shown in FIGS. 8 to 9, the number of colors to be assigned to the palette can be appropriately distributed even when there is a color with an extremely high frequency. In addition, even if there is a color that is varied due to noise or the like, or if there is a similar color that changes subtly, the necessary color can be reliably registered in the palette.

[0063]

As described above, in the image processing apparatus according to the present invention, when the frequency of each color included in each partial color space histogram is higher than the average appearance frequency of all appearance colors, the frequency is changed. Since the number of colors assigned to the palette for each partial color space is determined based on the corrected partial color space histogram based on this corrected partial color space histogram, there are colors with a frequency that is prominent in the partial color space. Even in such a case, it is possible to appropriately allocate the number of colors to be allocated to the palette without excessive allocation from the partial color space to the palette.

According to this, since the number of registered palettes can be effectively used and the colors can be registered in a well-balanced manner from each partial color space, it is possible to obtain an output result with less discomfort in the hue and the like with respect to the original image. It will be possible.

[Brief description of drawings]

FIG. 1 is a block diagram showing a functional configuration of an image processing apparatus according to an embodiment.

FIG. 2 is a configuration diagram showing a specific example for realizing an image processing apparatus.

FIG. 3 is a flowchart showing an outline of palette creation processing.

FIG. 4 is an explanatory diagram of a partial color space histogram.

FIG. 5 is an explanatory diagram of a modified partial color space histogram.

FIG. 6 is an explanatory diagram showing contents obtained from a corrected partial color space histogram.

FIG. 7 is an explanatory diagram of an all-appearance color histogram that is color-coded based on the number of assigned colors.

FIG. 8 is a flowchart showing a specific example of palette creation processing.

FIG. 9 is a flowchart showing a specific example of palette creation processing.

FIG. 10 is an explanatory diagram showing a scanning order of pixels.

FIG. 11 is an explanatory diagram showing the contents of the all appearance color histogram.

FIG. 12 is a schematic diagram showing the relationship between the all appearance color histogram and each partial color space histogram.

FIG. 13 is an explanatory diagram showing an example of a histogram of all appearance colors in a conventional example.

FIG. 14 is an explanatory diagram showing a histogram of all appearance colors in which frequencies are dispersed due to noise.

FIG. 15 is an explanatory diagram showing an example of a histogram of all appearance colors in a conventional example.

FIG. 16 is an explanatory diagram of a partial color space histogram in the conventional example.

FIG. 17 is an explanatory diagram showing contents obtained from a conventional partial color space histogram.

FIG. 18 is an explanatory diagram of an all-appearance color histogram that is color-coded based on the number of assigned colors in the conventional example.

[Explanation of symbols]

11 ... All appearance color histogram creation means, 12 ... Partial color space histogram creation means, 13 ... Frequency correction means, 14 ...
Palette registration color number determination means, 15 ... Palette registration means

Claims (1)

[Claims] 1. An all-appearance color histogram creating means for creating an all-appearance color histogram on the basis of pixel data of a color original image, and an all-appearing color histogram created by the all-appearance color histogram creating means. , A partial color space histogram creating means for creating a partial color space histogram for each partial color space obtained by equally dividing the color space, and all appearance colors based on the all appearance color histogram created by the all appearance color histogram creating means. While determining the average appearance frequency of the, the frequency of each color included in the partial color space histogram is equal to or more than the average appearance frequency, the frequency is corrected to the average appearance frequency, for each partial color space histogram From the frequency correction means for creating the modified partial color space histogram, and the modified partial color space histogram created by the frequency correction means. Obtain an appearance frequency meter of the color in the partial color space, and based on this appearance frequency meter, obtain the ratio of the frequency of the appearance color in each partial color space, and based on this ratio, the palette for each partial color space. The number of colors to be assigned to the palette for determining the number of colors to be assigned to the palette, and according to the number of colors to be assigned to be determined by the means for determining the number of colors to be registered in the palette, colors are assigned to the palette in order from the color with the highest frequency for each of the partial color space histograms. An image processing apparatus comprising: a pallet registration unit that performs registration.