JP2890275B2 - Image processing method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a method for representing a natural still image in which each pixel is represented by any one of N colors by using M colors less than N colors without deteriorating the image quality of the natural still image. In addition, the present invention relates to an image processing method for performing a process of thinning N colors into M colors.

[0002]

2. Description of the Related Art A navigation study group defines a format relating to a navigation device, and also defines a color representing a natural still image displayed on a display device. As is well known, the color of each pixel constituting the natural still image is R
(Red), G (green), and B (blue). According to the above navigation study group,
The natural still image displayed on the display device of the navigation device is quantized by 8 bits for each of R, G, and B, so that (2 ⁸ × 2 ⁸ × 2 ⁸ ) colors, that is, about 1
It is to be expressed in 6.77 million colors. However, in order to store such image data representing a natural still image having a large number of colors in an image memory, it is necessary to use an image memory having a large storage capacity.

Various methods have been proposed for reducing the storage capacity of an image memory, in other words, for expressing a natural still image with a small number of colors.

As one of the methods, a dither method is well known. By employing this dither method, the number of colors used can be reduced. Here, the colors to be used are limited to a predetermined number. In the dither method,
A limited number of color combinations, for example, 4 × 4 = 1
Six pixels express not only the colors used but also other colors. However, the dither method has a drawback that a high-resolution screen is required as a display device in order to make a displayed image look natural without deterioration. Therefore,
Although the dither method can reduce the number of colors used, the storage capacity of the image memory cannot be reduced too much.

In another method, there is a method of simply thinning out colors to be displayed. Here, "simple thinning"
Means that R, G and B are each 3 bits × 3 bits × 3
This means that video is coarsely quantized by bits and the image is represented by 2 ³ × 2 ³ × 2 ³ colors, that is, 512 colors. Here, 51
The two colors are fixed colors. However, although this method can reduce the storage capacity of the image memory, it has a disadvantage that the image quality of a natural still image is deteriorated.

Further, another method is disclosed in
There is a limited color display method disclosed in Japanese Patent No. 1374. In this limited color display method, a hue axis of a display candidate color is registered in advance according to an input image, a chromaticity appearance frequency of an image converted from the input image to a hue image is obtained, and each of the hue axes is registered with respect to the registered hue axis. By examining the chromaticity appearance frequency around the axis of, the hue having the highest chromaticity appearance frequency in the local area is converted. Here, the number of display candidate colors is, for example, 6
Very few with pieces. Therefore, when a natural still image is displayed in a color limited by this limited color display method, the image quality naturally deteriorates. In other words, this limited color display method
It is not suitable to use for displaying a natural still image to be displayed on a display device of a navigation device.

Therefore, an object of the present invention is to provide an image processing method capable of expressing a natural still image with a small number of colors without deteriorating the image quality and reducing the storage capacity of an image memory to be used. It is in.

[0008]

According to the image processing method of the present invention, a natural still image in which each pixel is represented by any one of N colors can be used without deteriorating the image quality of the natural still image. This is a method of performing a process of thinning out the N colors to the M colors in order to represent M colors less than N colors.

The image processing method according to the first aspect of the present invention comprises the steps of: (a) coarsely quantizing each pixel representing the natural still image so that each pixel is less than the N colors but the M colors; Get a coarse still image in any of the more colors,
(B) examining the frequency distribution representing the frequency of the color representing the coarse still image, and (c) sorting the color in the order of the color having the highest frequency,
(D) A color equal to the frequently used M color is determined as a color to be used among colors representing the coarse still image, and (e) the remaining colors of the color representing the coarse still image other than the color to be used. For, look for similar colors in the colors used above,
Converting the remaining colors into the similar colors; and (f) adding the colors in which the similar colors are not found among the remaining colors to the colors to be used, and calculating the frequency from the colors to be used. Removing the least number of colors and adding to the remaining colors; and (g) repeating steps (e) and (f) until all of the remaining colors are converted to the similar colors. It is characterized by.

[0010] In the image processing method of the first aspect, the N, for example, equal to ^{2 8 × 2 8 × 2 8} ,
Wherein M is, for example, equal to 2 ^8.

According to a second aspect of the present invention, there is provided an image processing method comprising the steps of: (a) displaying a screen of the natural still image with a first area having a gradual color change; (B) determining the L color that is more frequent and less than the M in the first area as a color to be used by the first group; and (c) determining the second group. For the remaining colors of the first group other than the colors used by the first group in the first area, a similar color of the first group is searched for among the colors used by the first group. Converting the remaining colors of the first group into similar colors of the first group, and (d) converting the remaining colors of the first group without the similar colors of the first group into the second area. (E) If there is a color similar to the color used by the first group in the color representing the second area, the similar color is represented by the first color. (F) each pixel representing the natural still image in the second area is coarsely quantized to make the pixel smaller than the N color. A coarse still image represented by any one of a small number of colors larger than the (ML) color is obtained, and (g) a frequently used (ML) color is selected from the colors representing the coarse still image by the second color. (H) determining the colors used by the second group for the remaining colors of the second group representing the coarse still image other than the colors used by the second group; Searching for a similar color in the second group, and converting the remaining colors in the second group into similar colors in the second group;
(I) In addition to the colors used in the second group, the colors in which the similar colors in the second group are not found among the remaining colors in the second group are added to the colors used in the second group. Removing the least frequent color and adding it to the remaining colors of the second group;
Repeating steps (h) and (i) until all of the remaining colors in the group have been converted to similar colors in said second group.

In the image processing method according to the second aspect, N is, for example, equal to 2 ⁸ × 2 ⁸ × 2 ⁸ ;
Wherein M is, for example, equal to 2 ^8, wherein L is, for example, equal to 2 ^7.

[0013]

Embodiments of the present invention will be described below with reference to the drawings.

In the image processing method of the present invention, a natural still image in which each pixel is represented by one of N colors is represented by M colors less than N colors without deteriorating the image quality of the natural still image. In addition, a method of thinning out N colors to M colors. In this embodiment, N is 2 ⁸ × 2 ⁸ × 2 ⁸ , that is,
About 1677 equals ten thousand, M is 2 ^8, i.e., equal to 256.

In this embodiment, the present image processing method is applied to a natural still image displayed on a display device of a navigation device. The display device of the navigation device is (320 ×
210) Assume that a screen is composed of pixels. Therefore,
The colors actually displayed on the display device are at most 67,20.
0 color. That is, although there are about 16,770,000 colors that can be used for display, the number of colors actually displayed on the display screen is 67,200 or less. In this embodiment, the number of colors displayed on the display screen is limited to 256.

Once the 256 colors to be used are determined by the image processing method, a natural still image is displayed on a display device using a palette well known in the art. That is, 256 colors out of about 16.77 million colors are allocated to the palette memory as palette data. The 256 addresses indicate the number of pallets used on the screen, and each pallet address indicates a pallet number. The image memory stores pixel data composed of color data using the pallet data. When displaying a natural still image on a display device, each pixel data read from the image memory is given to the palette memory as an address signal,
Pallet data representing a color is read from the address specified by the address signal and sent to the display device. The detailed operation of the display method using the pallet is described in, for example, Japanese Patent Application No. 152,202, filed on May 29, 1991 by the applicant of the present invention, entitled "Image Display Apparatus". Please refer to the specification and drawings attached to the application form.

Referring to FIGS. 1 and 2, a first embodiment of the present invention will be described.
An image processing method according to the embodiment will be described. The image processing method according to the present embodiment includes a coarse quantization step S1, a use color determination step S2, a color conversion step S3, a color replacement step S4, and a processing end determination step S5.

In the coarse quantization step S1, each pixel representing a natural still image is coarsely quantized so that each pixel is represented by one of less than about 16.770,000 colors but more than 256 colors. Obtain a still image. In this embodiment, R,
G and B are coarsely quantized by 4 bits × 4 bits × 3 bits, respectively, and 2 ⁴ × 2 ⁴ × 2 ³ colors, that is, 2048
Obtain a coarse still image expressed in color. Coarse quantization step S
After 1, the process proceeds to the used color determination step S2.

In the used color determination step S2, 256 colors that are frequently used are determined as colors to be used from the 2048 colors representing the coarse still image obtained in the coarse quantization step S1. More specifically, a use color determination step S2
Comprises a frequency check substep S2-1, a sorting substep S2-2, and a color determination substep S2-3. In the frequency investigation sub-step S2-1, a frequency distribution representing the frequency of a color indicating a coarse still image is examined (FIG. 2).
(A)). In the sorting sub-step S2-2, sorting, that is, rearrangement is performed in the order of color having the highest frequency (see FIG. 2B). Here, a well-known method in this technical field is employed for sorting.
For example, a binary sorting method or a quick sorting method is used.
In the color determination sub-step S2-3, 256 colors that are frequently used are determined as colors to be used from among colors representing coarse still images. Color conversion step S after use color determination step S2
Proceed to 3.

In the color conversion step S3, as shown by the arrow A in FIG. 2B, the remaining colors representing the coarse still image other than the 256 colors used are similar among the 256 colors used. Find the color that exists and convert the remaining colors to similar colors. After the color conversion step S3, the color replacement step S4
Move on to

In the color replacement step S4, as shown by the arrow B in FIG. 2B, this color is used in addition to the 256 colors that are not found in the similar colors among the remaining colors. The least frequent color is removed from the 256 colors and added to the remaining colors. After the color replacement step S4, the process proceeds to the processing end determination step S5.

In the processing end determination step S5, it is determined whether all the remaining colors have been converted to similar colors.
If all of the remaining colors have not been converted to similar colors (NO in step S5), the process returns to color conversion step S3,
If it has been converted (YES in step S5), the process ends. In this way, as shown in FIG. 2C, 256 colors to be used are determined.

After determining the 256 colors to be used by the image processing method according to the first embodiment, the present inventor tried to display a natural still image on a display device using the above-mentioned palette. As a result, although the image quality was degraded to some extent, an image that was somehow durable was obtained. However, when the inventor observed the image obtained in this manner in detail, it was confirmed that the image quality was degraded, particularly in an area where the color changes gradually, such as the sky. This is because each pixel representing a natural still image is coarsely quantized in the coarse quantization step S1 without distinguishing whether the color change is gentle or not. Probable cause. To prevent the image quality from deteriorating in the area where the color changes gradually,
The inventor has devised a second embodiment which is an improvement on the first embodiment.

Referring to FIGS. 3 and 4, the second embodiment of the present invention will be described.
An image processing method according to the embodiment will be described. The image processing method according to the present embodiment includes an area dividing step Sa1 and a first
A group use color determination step Sa2, a first group color conversion step Sa3, a color replacement step Sa4, a coarse quantization step Sa5, a second group use color determination step Sa6,
Second group color conversion step Sa7 and color replacement step Sa
8 and a process end determination step Sa9.

In the area dividing step Sa1, the screen of the natural still image (FIG. 4A) is divided into a first area having a gentle color change and the remaining area of the natural still image screen excluding the first area. It is divided into a certain second area (see FIG. 4B).
Here, the first area is called an area A, and the second area is called an area B.

Hereinafter, the operation of the area dividing step Sa1 will be described in detail. The area A is extracted from the natural still image as described below.

First, the RGB values are differentiated in all the horizontal directions of the screen, and if the differentiated value is larger than a predetermined first value, it is determined that there is a change and the pixel is deleted.

Next, a value obtained by dividing the absolute value of the difference between the RGB values between the start point and the end point of the remaining gently changing line segment by the length of the line segment is a predetermined second value. If larger,
It is assumed that the line segment is to be left. This operation is for removing a line segment that does not change. At this time, the same operation is performed on the remaining line segments in the vertical direction. A portion determined to be erased in both the vertical and horizontal directions is determined as a region to be erased, and a region determined to be left in either one is determined as a portion to be left. The region thus left is a region that changes gently.

With respect to the remaining area obtained in this way, the remaining area less than a predetermined length in the vertical and horizontal directions is deleted. Here, the predetermined length is, for example, a length corresponding to 10 pixels. This operation is performed twice in order to erase a region having a certain area or less. That is, this operation is an operation for erasing an area that cannot be recognized by human eyes. Here, this operation is performed twice because, if the operation is performed only once, the "whisker" may remain without erasing the area in terms of area.

The area obtained in this way is selected as a gentle area A, and the unselected area is set as an area B. The region grouping step Sa1 is followed by a first group use color determining step Sa2.

In the first group used color determining step Sa2,
The 128 colors that are frequently used in the area A are determined as the colors used by the first group. Here, in order to quickly perform the determination operation in step Sa2, coarse quantization may be performed on each pixel to the extent that the image is not deteriorated before executing step Sa2. For example, each of R, G, and B is slightly coarsely quantized by 6 bits × 6 bits × 5 bits to obtain a somewhat coarse still image expressed in 2 ⁶ × 2 ⁶ × ²⁵ colors, that is, about 130,000 colors. An image obtained by such coarse quantization cannot be distinguished from a natural still image by human eyes. The detailed operation in this step Sa2 is the same as the used color determining step S2 in the first embodiment, and thus the description thereof is omitted. After the first group use color determining step Sa2, the process proceeds to the first group color conversion step Sa3.

In the first group color conversion step Sa3, for the remaining colors of the first group other than the colors used by the first group in the area A, a similar color is searched from the colors used by the first group. Then, the remaining colors of the first group are converted to similar colors of the first group. The first group color conversion step Sa3 is replaced with a color replacement step S
a4 follows.

The color replacement step Sa4 includes a color addition sub-step Sa4-1 and a color assignment sub-step Sa4-2. In the color addition sub-step Sa4-1, the remaining colors (of the first group) having no similar colors among the colors used in the first group are added to the colors representing the area B. Color assignment sub-step Sa4-
In 2, if there is a color similar to the color used by the first group in the color representing the area B, the similar color is assigned to the color used by the first group and removed from the area B. This is to avoid overlapping of the color used in the area A with the color used in the area B.
When the processing in the color replacement step Sa4 is completed, 128 colors to be used in the area A are determined. Color replacement step Sa
After 4, the process proceeds to the coarse quantization step Sa5.

In the coarse quantization step Sa5, each pixel representing a natural still image in the area B is coarsely quantized so that each pixel is less than about 16.77 million colors but less than 128.
Obtain a coarse still image represented in one of more colors.
In this embodiment, each of R, G, and B is roughly quantized by 4 bits × 4 bits × 3 bits, and 2 ⁴ × 2 ⁴ × 2 ³ colors,
That is, a coarse still image expressed in 2048 colors is obtained.
An image obtained by such coarse quantization is an image that can be seen by human eyes. Here, B is quantized more coarsely than other R and G because it is known that the color of B does not significantly affect the sensitivity of human eyes. Here, the reason why the quantization is roughened with respect to the region B is that the region B is a region where the color changes sharply as compared with the region A. This is because the difference cannot be discriminated as compared with the case where the quantization is not coarsened. After the coarse quantization step Sa5, the process proceeds to the second group use color determination step Sa6.

In the second group used color determining step Sa6,
Of the colors representing the coarse still image, 128 colors that are frequently used are determined as the colors used by the second group in the area B. Second
The group use color determination step Sa6 is followed by a second group color conversion step Sa7.

In the second group color conversion step Sa7, the second group
For the remaining colors of the second group representing the coarse still images other than the colors used by the group, a similar color of the second group is searched for among the colors used by the second group, and the remaining colors of the second group are searched. To a second group of similar colors. After the second group color conversion step Sa7, the process proceeds to the color replacement step Sa8.

In the color replacement step Sa8, the second group is used in addition to the colors used in the second group, in which the similar colors of the second group are not found among the remaining colors in the second group. The least frequent color is removed from the colors and added to the remaining colors in the second group. Processing end determination step Sa in color replacement step Sa8
9 follows.

In the processing end determination step Sa9, it is determined whether or not all of the remaining colors of the second group have been converted to similar colors of the second group. If all of the remaining colors in the second group have not been converted to similar colors in the second group (step Sa9)
NO), the process returns to the second group color conversion step Sa7, and if the conversion has been performed (YES in step Sa9), the processing for the area B ends. This processing end determination step Sa9
Is completed, 128 colors to be used in the area B are determined.

In this way, 256 colors including the 128 colors used for the area A and the 128 colors used for the area B are assigned to the palette, and the screen divided into the areas A and B is synthesized to form a natural still image. Is displayed on the display device.

When the inventor observed a natural still image displayed on the display device, it was confirmed that the image was good without deterioration in image quality.

The specific numerical values such as the number of quantization bits used in the above-described embodiment are merely examples, and the present invention is not limited to these. That is, these numerical values are appropriately selected depending on the display medium, the number of pallets that can be used, and the like.

In the embodiment of the second embodiment, the number of colors to be used for each of the first and second areas is set to a half of the desired color. If the number of colors in the two areas is equal to the desired number of colors, the number of colors may be assigned to the first and second areas.

In the above description of the embodiment, the hardware is not described at all. However, at least an image memory for storing a natural still image to be subjected to the inputted processing, and a program for performing the above processing operation are provided. It is needless to say that a processor for processing a natural still image stored in the image memory in accordance with a program memory storing the image memory and a program stored in the program memory is required.

[0044]

As is apparent from the above description, according to the present invention, when a natural still image is displayed on a display device, about 16.770,000 colors which conventionally required a large-capacity image memory were used. For example, there is an effect that a natural still image can be displayed without deteriorating the image quality by limiting the image to very few colors such as 256 colors and using an image memory having a small storage capacity.

In one embodiment of the present invention, the first and second regions are divided into a first region where the change of RGB is gentle in the vertical and horizontal directions of the display screen and a second region where the change is not so. The number of colors to be performed is stored in L and (ML) of the desired number of colors M, the number of colors in the first and second areas is combined to obtain the desired number of colors, and the colors determined in this way are assigned to a palette. Since the screens divided into the first and second areas are combined and displayed, there is also an advantage that an image having good image quality can be obtained.

[Brief description of the drawings]

FIG. 1 is a flowchart illustrating an image processing method according to a first embodiment of the present invention.

FIG. 2 is a diagram for explaining the image processing method of FIG. 1;
(A) shows the frequency distribution obtained in the frequency research substep, (b) shows the frequency distribution after sorting in the sorting substep, and (c) shows the frequency distribution of the color finally determined. Show.

FIG. 3 is a flowchart illustrating an image processing method according to a second embodiment of the present invention.

4A and 4B are diagrams for explaining the operation of an area division step in FIG. 3, wherein FIG. 4A shows a screen before area division;
(B) shows the screen after area division.

[Explanation of symbols]

 S1 Coarse quantization step S2 Use color determination step S2-1 Frequency investigation sub-step S2-2 Sorting sub-step S2-3 Color determination sub-step S3 Color conversion step S4 Color replacement step S5 Processing end determination step Sa1 Area division step Sa2 First Group use color determination step Sa3 First group color conversion step Sa4 Color replacement step Sa4-1 Color addition substep Sa4-2 Color allocation substep Sa5 Coarse quantization step Sa6 Second group use color determination step Sa7 Second group Color conversion step Sa8 Color replacement step Sa9 Processing end step

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁶ , DB name) G06T 1/00 G06T 5/00-5/50 G09G 5/00-5/40 H04N 1/46-1 / 60

Claims (4)

(57) [Claims] 1. A method for displaying a natural still image in which each pixel is represented by one of N colors with M colors smaller than the N colors without deteriorating the image quality of the natural still image. An image processing method for thinning a color into the M colors includes the steps of: (a) coarsely quantizing each pixel representing the natural still image so that each pixel is smaller than the N colors, (B) examining a frequency distribution indicating the frequency of the color representing the coarse still image, (c) sorting in the order of color having the highest frequency, (d) From the colors representing the coarse still image, a color equal to the frequently used M color is determined as a color to be used, and (e) the remaining colors of the coarse still image other than the color to be used are Look for similar colors in the colors you use and replace the remaining colors with (F) adding the color in which the similar color is not found among the remaining colors to the color to be used, and removing the least frequently used color from the color to be used. And (g) repeating steps (e) and (f) until all of the remaining colors are converted to similar colors. . 2. The image processing method according to claim 1, wherein said N is equal to 2 ⁸ × 2 ⁸ × 2 ⁸ and said M is equal to 2 ⁸ . 3. The method according to claim 1, wherein each of the pixels is represented by one of the N colors by using M colors smaller than the N colors without deteriorating the image quality of the natural still image. An image processing method for performing a process of thinning colors to M colors, wherein: (a) a screen of the natural still image includes a first area having a gradual change in color, and the natural still image excluding the first area. And (b) determining L colors, which are more frequent and less than the M colors in the first area, as colors to be used in the first group, and (c) For the remaining colors of the first group other than the colors used by the first group in the first area, a similar color of the first group is searched for among the colors used by the first group. Converting the remaining colors of the group to similar colors of the first group; and (d) converting the remaining colors of the first group to Adding a group of dissimilar colors to the color representing the second region, and (e) if the color representing the second region has a color similar to the color used by the first group, Assigning the similar colors to the colors used by the first group and removing them from the second area; and (f) quantizing each pixel representing the natural still image in the second area by quantizing the pixels. Coarse, each pixel
Obtaining a coarse still image represented by any of the colors less than the colors but more than the (ML) color; and (g) extracting the (ML) color having the higher frequency from the colors representing the coarse still image as And (h) using the second group for the remaining colors of the second group representing the coarse still image other than the colors used in the second group. Looking for similar colors in the second group among the colors, converting the remaining colors in the second group into similar colors in the second group, and (i) converting the remaining colors in the second group. In addition, the colors of the second group, in which similar colors are not found, are added to the colors used by the second group, and the least frequently used colors are removed from the colors used by the second group. (J) the steps (h) and (h) until all of the remaining colors of the second group are converted to similar colors of the second group; And (i) are repeated. 4. The method according to claim 3, wherein said N is equal to 2 ⁸ × 2 ⁸ × 2 ⁸ , said M is equal to 2 ⁸ , and said L is equal to 2 ⁷ .