JPH08331373A - Color image reduction device and color image reduction method - Google Patents

Abstract

PURPOSE: To provide the color image reduction device and the color image reduction method in which a reduced color image is obtained while keeping color information of an original color image. CONSTITUTION: A representative color generating means 24 has a histogram generating means 26 and a range selection output means 28. The histogram generating means 26 counts number of picture elements having a color belonging to each range block obtained by dividing a color space as to picture elements forming a division image. The range selection output means 28 selects the range block including most of picture elements among range blocks as a representative range block and provides an output of a color corresponding to the representative range block as a representative color of a division image. A reduced image generating means 30 replaces one division image with one picture element and generates a reduced color image by using the representative color of the division image as a color of one picture element. Thus, a color close to colors appearing most in the division image is used for a representative color to reduce the image.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a color image reducing apparatus and a color image reducing method, and more particularly to a technique for reducing an image while maintaining the color of an original color image.

[0002]

2. Description of the Related Art Color image reduction processing is performed for the purpose of preprocessing for image processing. A block diagram of color image reduction processing by a conventional color image reduction device is shown in FIG. 15A. The conventional color image reduction processing is composed of image division processing 4, thinning processing 6, and reduced image creation processing 8.

As shown in FIG. 16, first, one original color image 10 is divided into a predetermined number of divided images 12a, 12b, ...
., 12j. This is the image division processing 4 (see FIG.
5A). Here, the divided images 12a, 12b,
.., 12j are each composed of a plurality of pixels. Also, each pixel has its own color.

Next, the divided images 12a, 12b, ...
, 12j, one pixel is extracted from each of the divided images 12a, 12b, ..., 1
The representative color of 2j. In the example of FIG. 16, each divided image 12
Pixel 1 located at the upper left of a, 12b, ..., 12j
4a1, 14b1, ..., 14j1 are extracted. This is the thinning process 6 (see FIG. 15A).

Next, the extracted pixels 14a1 and 14b
Pixels 16 having the same color as 1, ..., 14j1
A reduced color image 18 is created using a, 16b, ..., 16j as elements. This is the reduced image creation processing 8 (see FIG. 15A).

.., 12j of the original color image 10 and the pixels 16a, 16b ,. The image 10 can be reduced.

[0007]

However, the color image reduction processing by such a conventional color image reduction apparatus has the following problems. In the color image reduction processing shown in FIG. 15A, the divided images 12a, 12
When the representative color of b, ..., 12j is determined, the pixels 14a1, 14b1, ..., 14j1 located at the upper left of each divided image 12a, 12b ,. There is. Therefore, there is an advantage that the process of determining the representative color is extremely simple.

However, the divided images 12a, 12b, ...
.., pixels 14a1 and 14b located at the upper left of 12j
Even if there is meaningless noise on any one of 1, ..., 14j1, the color of the pixel with noise is determined to be the representative color of the divided image. was there. In this case, the reduced color image 18 cannot be created faithfully to the original color image 10 and the post-processing is hindered.

To solve this problem, a color image reduction process shown in FIG. 15B has been proposed. In the color image reduction process shown in FIG. 15B, the thinning process 6 (see FIG.
5A), averaging process 2 is adopted.

Averaging process 2 is performed as follows. For example, all the pixels 14a1 and 14a forming the divided image 12a
The average value of the colors a2, ..., 14ak is calculated, and this average value is used as the representative color of the divided image 12a. Other split image 1
For 2b, ..., 12j, the representative color is determined by the same method.

As described above, by adopting the averaging process 2 in determining the representative color, even if a part of the divided image has meaningless noise, the noise is almost ignored. You can Therefore, it is possible to create a highly reliable reduced color image 18 that is hardly affected by noise.

However, in the color image reduction processing shown in FIG. 15B, for example, the average value of the colors of all the pixels 14a1, 14a2, ..., 14ak forming the divided image 12a is used as the representative color of the divided image 12a. Therefore, there is a problem that the representative color does not always reproduce the original color of the divided image 12a.

For example, 80 out of all the pixels 14a1, 14a2, ..., 14ak forming the divided image 12a.
When the color of% of the pixels is “white” and the color of 20% of the pixels is “black”, the representative color is not “white” which occupies most of the divided image 12a, but “black” which occupies a part thereof. But not "ash"
Becomes

Therefore, when the post-processing subsequent to the color image reduction processing is processing in which the color of the original color image 10 is used as a reference, for example, "" white "is recognized as" wall clock ". If such processing is performed, inconvenience will occur.

The present invention solves the problems of the conventional color image reducing apparatus, and it is possible to obtain a reduced color image while retaining the color information of the original color image, and a color image reducing method. The purpose is to provide.

[0016]

[Means for Solving the Problems]

[0017]

[Technical idea devised to solve the problem] To provide a color image reducing apparatus and a color image reducing method capable of obtaining a reduced color image while retaining color information of an original color image, a divided image is configured. The reduced color image is created by using the color with the maximum frequency among the colors of the pixels to be represented as the representative color of the divided image.

That is, as described in FIG. 1 showing the configuration of the invention described in the claims, the color image reducing apparatus according to claim 1 makes it possible to obtain a plurality of original color images stored in the image storage area 20. Image dividing means 22 for dividing into divided images
Among the colors of the pixels forming the divided image, the representative color generating unit 24 that makes the color with the maximum frequency the representative color of the divided image, replaces one divided image with one pixel, and sets the representative color of the divided image to 1 A reduced image creating means 30 for creating a reduced color image is provided as the color of the pixel.

A color image reducing apparatus according to a second aspect is the color image reducing apparatus according to the first aspect, in which the representative color generating means 24 is provided.
Is a histogram creating means 26 for counting the number of pixels having a color belonging to each range block obtained by dividing the color space among the pixels constituting the divided image, and the range block containing the most pixels among the range blocks is represented. It is characterized in that range selection output means 28 for selecting the range block and outputting the color corresponding to the representative range block as the representative color of the divided image is provided.

According to a third aspect of the color image reduction method of the present invention, the original color image is divided into a plurality of divided images, and among the colors of the pixels constituting the divided image, the color having the maximum frequency is used as a representative color of the divided image. One of the divided images is replaced with one pixel, and a reduced color image is created by using the representative color of the divided image as the color of the one pixel.

A color image reducing method according to a fourth aspect is the color image reducing method according to the third aspect, in which when defining the representative color, the pixels constituting the divided image belong to each range block obtained by dividing the color space. The number of pixels having a color is counted, and the range block that includes the most pixels among the range blocks is selected as the representative range block, and
The color corresponding to the representative range block is output as the representative color of the divided image.

[0022]

[Definition of terms] The concept of terms in the claims expressing the technical idea devised to solve the problems will be defined as follows, and the relationship between the terms and examples will be described.

"Color with maximum frequency": A color with the highest appearance frequency or a color close to the color with the highest appearance frequency. In the embodiment, the in-block pixel average value Mij calculated in the pixel average value calculation process of FIG. 3 corresponds.

"Color space": A space having coordinate axes of elements used for expressing colors. Therefore, the color
When defined as a combination of R (red component), G (green component), and B (blue component), the color space is a three-dimensional space. Similarly, when defining as a combination of Y (yellow), M (magenta) and C (cyan), H (hue), S
When defining as a combination of (saturation) and V (brightness),
It becomes a three-dimensional space. In the embodiment, the RGB space shown in FIG. 11 is applicable.

"Range block": Refers to each of the divided blocks when the color space is divided into a plurality of blocks. Therefore, when the RGB space is a rectangular coordinate system, a cubic range block can be obtained by equally dividing each of the R, G, and B components. In the embodiment, the first
This corresponds to the range block Bs in FIG.

[0026]

A color image reducing apparatus according to claim 1 and claim 3
The color image reduction method is characterized in that the reduced color image is created by using the color of the maximum frequency among the colors of the pixels forming the divided image as the representative color of the divided image. Therefore, it is possible to reduce the size of the image with the color closest to the color that appears most in the divided images as the representative color.

The color image reducing apparatus according to claim 2 and the color image reducing method according to claim 4 further count the number of pixels having a color belonging to each range block obtained by dividing the color space, It is characterized in that the color corresponding to the representative range block including the largest number of pixels is output as the representative color of the divided image. Therefore, the size of the range block can be set according to the request for the post-processing.

[0028]

FIG. 2 shows a hardware configuration when each function of the color image reducing apparatus shown in FIG. 1 is realized by using a CPU. This color image reducing device is generally composed of a color television camera 32, a workstation WS, and a CRT 34.
It has. The workstation WS includes an image input board 36, a keyboard 38, a mouse 40, a CPU 42,
The RAM 44, the ROM 46, and the interface 48 are provided.

The color television camera 32 photographs a landscape or the like, converts the obtained optical information into an analog electric signal, and outputs the analog electric signal. The analog electric signal output from the color television camera 32 is taken into the RAM 44 as a digital electric signal via the image input board 36. RAM44
The digital electric signal captured by is called image data.

The CPU 42 reduces the image data taken into the RAM 44 according to the program stored in the ROM 46. The processing result is interface 4
It is displayed on the CRT 34 via 8. Keyboard 38
The mouse 40 is also used as an input means for commands and the like.

The RAM 44 corresponds to the image storage area 20 in FIG. The CPU 42 also includes the image dividing unit 22, the histogram creating unit 26, and
It corresponds to the range selection output means 28 and the reduced image creation means 30.

Next, FIG. 3 shows a block diagram of the color image reduction processing by the color image reduction apparatus shown in FIG.
Color image reduction processing includes color image setting processing 50, image division processing 52, histogram creation processing 54, range selection processing 56, pixel average value calculation processing 58, and reduced image creation processing 6.
It consists of zero.

Here, the image division processing 52 corresponds to the image division means 22 in FIG. The histogram creation processing 54 corresponds to the histogram creation means 26. The range selection process 56 and the pixel average value calculation process 58 correspond to the range selection output means 28. The reduced image creation processing 60 corresponds to the reduced image creation means 30.

FIG. 4 shows a time-series flowchart of the color image reduction processing shown in FIG. The flow of the color image reduction process will be described based on FIG. 4 while referring to the hardware configuration of FIG.

First, the CPU 42 converts the image data, which has been temporarily stored in the RAM 44 as a digital electric signal, into a predetermined format via the color television camera 32 and the image input board 36, and then, again, the image storage area. It is stored in the RAM 44. The image data thus converted into the predetermined format and stored in the RAM 44 is referred to as a sample color image (original color image) G. CP
U42 creates the sample color image G in this way (step S2).

As shown in FIG. 5, a sample color image G
Is a pixel with a full-color pixel value D [g, h] that has RGB components (red component, green component, blue component) as elements, and n rows and m
It is a matrix arranged orthogonally in columns, that is, a full-color image of size n × m pixels. The format of the sample color image G is shown below.

[0037]

However, G: sample color image D [g, h]; pixel value (full color pixel) g; vertical pixel number (= 1,2,3, ..., n) n; vertical pixel Number h; pixel number in horizontal direction (= 1,2,3, ..., m) m; number of pixels in horizontal direction n * m; total number of pixels.

FIG. 6 shows the data format of the sample color image G. Note that step S2 corresponds to the color image setting process 50 shown in FIG.

In the above example, the sample color image G is created by converting the image data taken into the RAM 44 via the color television camera 32 or the like into a predetermined format. The data is not limited to that input from the color television camera 32, and may be input from a magnetic disk (not shown) on which an image is recorded. Furthermore, the image data converted in advance into a predetermined format can be loaded into the RAM 44.

Next, the CPU 42 loads the reduction ratio input by the operator from the keyboard 38 into the RAM 44. In this embodiment, it is assumed that the vertical and horizontal reduction rates are the same, and this is 1 / k. Therefore, the overall reduction ratio is 1 / k * k.

That is, 1 / k * k; the reduction ratio of the sample color image. For example, if k = 4, the reduction ratio is 1/1
It becomes 6. In this way, the reduction rate is set (step S4).

Next, the CPU 42 equally divides the sample color image G into n / k (= ei) pieces in the vertical direction and m / k (= ej) pieces in the horizontal direction, as shown in FIG. Each of the divided images is called a divided image Gij. The size of the divided image Gij is k ×
There are k pixels, which can be n * m / (k * k). In this way, the sample color image G is divided (step S
6). The format of the divided image Gij is shown below.

Gij = {Dij [1,1], Dij [1,2], Dij [1,3], ..., Dij [1, k], Dij [2,1], Dij [2,2 ], Dij [2,3], ..., Dij [2, k], ... Dij [x, y], ... Dij [k, 1], Dij [k, 2], Dij [k , 3], ..., Dij [k, k]}.

However, Gij; divided color image (divided image) Dij [x, y]; pixel value (full color pixel) i; vertical division number (i = 1,2,3, ..., ei) ) ei; Vertical division number (= n / k) n; Vertical pixel number j; Horizontal division number (j = 1,2,3, ..., ej) ej; Horizontal division number (= m / k) m; number of pixels in the horizontal direction k; reciprocal of vertical (horizontal) reduction ratio of image x; vertical pixel number of divided image (x = 1,2,3, ...,
k) y; horizontal pixel number of the divided image (y = 1,2,3, ...,
k) x * y; The total number of pixels in the divided image.

FIG. 8 shows the data format of the divided image Gij. Note that step S6 is the image division processing 5 shown in FIG.
Corresponds to 2.

Next, the CPU 42 extracts RGB components from each full color pixel (k × k pixel) of the divided image Gij as shown in FIG. 9 (step S8). That is,
From the pixel value Dij [x, y], 256-level red component DRij [x, y]
, Green component DGij [x, y], and blue component DBij [x, y] are extracted. The format of the components of the pixel value Dij [x, y] is shown below.

Dij [x, y] = {DRij [x, y], DGij [x, y], DBij [x, y]} where Dij [x, y]; pixel value (full color pixel) DRij [x , y]; red component of pixel value (256 gradations) DGij [x, y]; green component of pixel value (256 gradations) DBij [x, y]; blue component of pixel value (256 gradations) i; Vertical division number (i = 1,2,3, ..., ei) ei; Vertical division number (= n / k) n; Vertical pixel count k; Vertical image (horizontal direction) Reciprocal of reduction ratio j; horizontal division number (j = 1,2,3, ..., ej) ej; horizontal division number (= m / k) m; horizontal pixel number x; divided image Vertical pixel number (x = 1,2,3, ...,
k) y; horizontal pixel number of the divided image (y = 1,2,3, ...,
k) x * y; The total number of pixels in the divided image. Note that FIG. 10 shows a data format of pixels.

Next, the CPU 42 causes the number L * of range blocks input by the operator from the keyboard 38.
L * L is loaded into the RAM 44. As a result, L * L * L range blocks Bs are created in the RGB color space (step S10).

Color is generally determined by three independent factors. Therefore, if these three elements are taken as three-dimensional coordinate axes, all colors are represented as points in the space defined by these three-dimensional coordinates. Generally, a space having coordinate axes of elements used for expressing colors is called a color space. In this embodiment, a color is defined as a combination of R (red component), G (green component) and B (blue component).
Therefore, if R, G, and B are taken as three-dimensional orthogonal coordinate axes, the color space is defined as a three-dimensional orthogonal space. This color space is RG
Called B space.

When the RGB space is divided into a plurality of blocks, each divided block is called a range block Bs (see FIG. 11). In this embodiment, each is finite (2
Each of the R, G, and B components of 56 gradations is divided into L parts to obtain L * L * L range blocks Bs.

Therefore, the range block Bs is a cube, and the size of the range block Bs is 256 / L × 256 / L × 256.
/ L. The range of L and the data format of the range block Bs are shown below.

1 <= L <= 256 (However, 256 / L must be a positive integer.) Bs = {r [s], g [s], b [s], dotnum [ s]} where L is the number of RGB component divisions Bs is the range block s is the block number of the range block (= 1,2, ..., L
* L * L) r [s] ； R component division number (= 1,2, ..., L) g [s] ； G component division number (= 1,2, ..., L) b [s]; B component division number (= 1,2, ..., L) dotnum [s]; counter value of the number of pixels in the block (initial value is 0).

The block number s and the range block Bs
The association with the contents of is performed as follows. B1 = (1,1,1,0), B2 = (1,1,2,0), B3 = (1,1,3,0) ..., BL = (1,1, L, 0) , BL + 1 = (1,2,1,0), ..., BL * L = (1, L, L, 0), BL * L + 1 = (2,1,1,0) ,. .., BL * L * L = (L, L, L, 0) That is, the B component is counted up in order from the G component to the R component.

The formula for calculating the division number of each RGB component from the block number s is shown below. r [s] = ceil (s / (L * L)) g [s] = ceil ((s-(r [s] -1) * L * L) / L) b [s] = s-(r [s] -1) * L * L-(g [s] -1) * L.

However, L: the number of divisions of each RBG component Bs; range block s; block number (= 1,2, ..., L * L * L) r [s]; division number of R component (= 1 , 2, ..., L) g [s] ； G component division number (= 1,2, ..., L) b [s] ； B component division number (= 1,2, ...) , L) ceil (x) ； It is a function that returns the smallest integer not less than x. In addition, the range block Bs is shown in the column (a) of FIG.
Shows the data format of.

Next, the CPU 42 sets, for each divided image Gij, three pixels in the RGB space of the pixels forming the divided image Gij.
A dimension histogram is created (step S12). The three-dimensional histogram is created as follows.

For each pixel forming the divided image Gij,
The range block Bs including the pixel value (3 component values of RGB) is searched for, and the pixel number counter value d of the range block Bs is searched.
Increment otnum [s]. K * k in the divided image Gij
The above processing is repeated until the processing of all the pixels is completed. The formula for calculating the three-dimensional histogram is shown below.

Each pixel value of the divided image: Dij [x, y] = {DRij
[x, y], DGij [x, y], DBij [x, y]} Range block: Bs = {r [s], g [s], b [s], dot
In num [s]}, 256 / L * (r [s] -1) <DRij [x, y] <= 256 / L * r [s] 256 / L * (g [s] -1) <DGij Block number that satisfies the condition of [x, y] <= 256 / L * g [s] 256 / L * (b [s] -1) <DBij [x, y] <= 256 / L * b [s] find s, dotnum [s] =
Use dotnum [s] + 1. 1 <= x, y <= k
Do for all x, y in the range.

However, Dij [x, y]; pixel value (full color pixel) DRij [x, y]; red component of pixel value (256 gradations) DGij [x, y]; green component of pixel value (256th floor) Key) DBij [x, y]; blue component of pixel value (256 gradations) i; vertical division number (i = 1,2,3, ..., ei) ei; vertical division number (= n / k) n ； Number of vertical pixels k ； Reciprocal of vertical (horizontal) reduction ratio of image j ； Horizontal division number (j = 1,2,3, ..., ej) ej ； Horizontal Number of divisions in direction (= m / k) m; number of pixels in horizontal direction x; vertical pixel number of divided image (x = 1,2,3, ...,
k) y; horizontal pixel number of the divided image (y = 1,2,3, ...,
k) L; number of divisions of each RGB component Bs; range block s; block number (= 1,2, ..., L * L * L) r [s]; division number of R component (= 1,2, ..., L) g [s] ； G component division number (= 1,2, ..., L) b [s] ； B component division number (= 1,2, ..., L). dotnum [s]; It is the pixel number counter value in the block (initial value is 0).

FIG. 11 shows the three-dimensional histogram created in this way. Note that step S12 corresponds to the histogram creation processing 54 shown in FIG.

Next, the CPU 42 selects the range block Bs having the largest pixel number counter value dotnum [s] (the block having the highest pixel appearance frequency) from all the range blocks Bs, and the range block Bs is selected. Is a representative range block Bc (step S14). The formula for obtaining the representative range block Bc is shown below.

Find c such that dotnum [c] = max (dotnum [s]). The range block Bs = c that gives c becomes the representative range block Bc.

However, s; block number (= 1,2, ..., L * L * L) L; number of divisions of each RGB component c; block number of representative range block dotnum [s]; pixel in block Number counter value max (dotnum [s]); It is an operator that finds the maximum dotnum [s] of dotnum [1] to dotnum [L * L * L].

FIG. 11 shows the representative range block Bc.
The data format of the representative range block Bc is shown in the column (b) of FIG. Note that step S14 corresponds to the range selection processing 56 shown in FIG.

Next, the CPU 42 causes the representative range block B
The average value of all the pixels included in c is calculated (step S16).

First, all the pixels (k * k) included in the representative range block Bc are searched, and the total sum of the values of the pixels included in the representative range block Bc is calculated. The total value is
Divide by the pixel number counter dotnum [c] of the representative block,
The average value is Mij. The procedure for calculating the average value is shown below.

Representative range block: Bc = {r [c], g [c]
, b [c], dotnum [c]} Each pixel of the divided image: Dij [x, y] = {DRij [x, y], DGij [x,
y], DBij [x, y]} Sum table: Dsum = {DRsum, DGsum, DBsum} (initial values are all 0) Mean value: Mij = {MRij, MGij, MBij}, 256 / L * ( r [c] -1) <DRij [x, y] <= 256 / L * r [c] 256 / L * (g [c] -1) <DGij [x, y] <= 256 / L * g If a pixel that satisfies [c] 256 / L * (b [c] -1) <DBij [x, y] <= 256 / L * b [c] exists, its component value (DRij [x,
y], DGij [x, y], DBij [x, y]) is the sum value (DRsum, DGsum,
DBsum). 1 <= x, y <= k
Do for all x, y in the range.

Finally, the sum value (Dsum) for each RGB component is
Divide by the total number of pixels in the divided image.

MRij = DRsum / dotnum [c] MGij = DGsum / dotnum [c] MBij = DBsum / dotnum [c].

However, Dij [x, y]; pixel value (full color pixel) DRij [x, y]; red component of pixel value (256 gradations) DGij [x, y]; green component of pixel value (256th floor) Key) DBij [x, y]; blue component of pixel value (256 gradations) i; vertical division number (i = 1,2,3, ..., ei) ei; vertical division number (= n / k) n ； Number of vertical pixels k ； Reciprocal of vertical (horizontal) reduction ratio of image j ； Horizontal division number (j = 1,2,3, ..., ej) ej ； Horizontal Number of divisions in direction (= m / k) m; number of pixels in horizontal direction x; vertical pixel number of divided image (x = 1,2,3, ...,
k) y; horizontal pixel number of the divided image (y = 1,2,3, ...,
k) L: Number of divisions of each RGB component Bc; Representative range block c; Block number of representative range block r [c]; Division number of R component of representative range block g [c]; Division of G component of representative range block Number b [c]; Division number of B component of representative range block dotnum [s]; Total number of pixels included in representative range block Dsum; Sum total value of pixels in block DRsum; Sum total value of R components of pixel in block DGsum; Sum of values of G component of pixel in block DBsum; Sum of value of B component of pixel in block Mij; Average pixel value of block MRij; R component of pixel average value of block MGij; G component of pixel average value of block MBij; Block It is the B component of the inner pixel average value.

The average value Mij thus calculated becomes the representative color of the divided area Gij. Note that step S16
This corresponds to the pixel average value calculation processing 58 shown in FIG.

In this way, the CPU 42 creates a three-dimensional histogram for each divided image Gij (step S12) and obtains the representative range block Bc (step S1).
4), average value Mij is calculated (step S16). In this way, the representative color is obtained for each divided image Gij.

Next, the CPU 42 makes the reduced color image M in which the average value Mij obtained for each divided image Gij is 1 dot.
Is created (step S18). This reduced color image M
The size of is ei × ej. Below, the reduced color image M
The format of the pixel value (full color pixel) Mij of is shown.

Mij = {MRij, MGij, MBij} where Mij is the pixel value MRij of the reduced color image M MRij is the R component of the pixel value of the reduced color image MGij is the G component of the pixel value of the reduced color image MBij is the reduced color image B component of pixel value of i; vertical division number (i = 1,2,3, ..., ei) ei; vertical division number (= n / k) n; vertical pixel number k; Reciprocal of vertical (horizontal) reduction ratio of image j; horizontal division number (j = 1,2,3, ..., ej) ej; horizontal division number (= m / k) m; horizontal The number of pixels in the direction.

FIG. 13 shows the reduced color image M. Further, FIG. 14 shows the data format of the reduced color image M. Note that step S18 is the reduced image creation processing 6 shown in FIG.
Corresponds to 0.

Although the RGB space is used as the color space in the above-described embodiment, the YMC having Y (yellow), M (magenta), and C (cyan) as elements is used as the color space.
It is also possible to use a space or an HSV space having H (hue), S (saturation), and V (lightness) as elements.

Although the average value Mij of the pixel values belonging to the representative range block Bc is adopted as the representative color of the divided area Gij, the representative range block Bc defined by the following equation is used as the representative color of the divided area Gij. The median value Cij may be adopted.

Median: Cij = {CRij, CGij, CBij} where CRij [x, y] = 256 / L * (r [c] -1/2) CGij [x, y] = 256 / L * ( g [c] -1/2) CBij [x, y] = 256 / L * (b [c] -1/2).

By using the median value Cij as the representative color of the divided area Gij, the time for calculating the average value Mij can be saved and the processing speed of the color image reduction processing can be shortened.

Further, the RGB space is divided into range blocks Bs, a three-dimensional histogram in the RGB space is created for each divided image Gij, and the representative range block Bc with the highest frequency of appearance is selected, whereby the divided image is divided. Although the representative color of Gij is calculated, if the gradation of each of the R, G, and B components (256 gradations in each component in the embodiment) is small (for example, 32 gradations), the RGB space is set. It is also possible to directly count the appearance frequency of each color without dividing into the range block Bs and use the color having a high appearance frequency as the representative color of the divided image Gij. With this configuration, it is not necessary to create a three-dimensional histogram, and the processing time can be shortened.

In the above embodiment, the CPU 4
Although the case where each function of the color image reducing apparatus shown in FIG. 1 is realized by using 2 is described as an example, a part or all of each function can be realized by hardware logic.

[0083]

According to the color image reducing apparatus of the first aspect and the color image reducing method of the third aspect, among the colors of the pixels forming the divided image, the color having the maximum frequency is used as the representative color of the divided image to form the reduced color image. It is characterized by creating.

Therefore, it is possible to reduce the size of an image by using a color that is closest to the color that appears most in the divided images as a representative color. That is, it is possible to obtain a reduced color image while retaining the color information of the original color image.

The color image reducing apparatus according to claim 2 and the color image reducing method according to claim 4 further count the number of pixels having a color belonging to each range block obtained by dividing the color space, It is characterized in that the color corresponding to the representative range block including the largest number of pixels is output as the representative color of the divided image.

Therefore, the size of the range block can be set in accordance with the request for post-processing. That is, it is possible to obtain a reduced color image while retaining the color information of the original color image with arbitrary accuracy.

[Brief description of drawings]

FIG. 1 is a diagram showing a configuration of a color image reducing device described in claims.

FIG. 2 is a diagram showing an example of a hardware configuration when each function of the color image reducing device according to the present invention is realized by using a CPU.

FIG. 3 is a block diagram of color image reduction processing by the color image reduction apparatus according to the embodiment of the present invention.

FIG. 4 is a time-series flowchart showing color image reduction processing by the color image reduction apparatus according to the embodiment of the present invention.

FIG. 5 is a diagram showing a sample color image in the color image reduction processing according to the embodiment of the present invention.

FIG. 6 is a diagram showing a data format of a sample color image in the color image reduction processing according to the embodiment of the present invention.

FIG. 7 is a diagram showing divided images in a color image reduction process according to an embodiment of the present invention.

FIG. 8 is a diagram showing a data format of divided images in a color image reduction process according to an embodiment of the present invention.

FIG. 9 is a diagram showing pixels forming a divided image in the color image reduction processing according to the embodiment of the present invention.

FIG. 10 is a diagram showing a data format of pixels forming a divided image in the color image reduction processing according to the embodiment of the present invention.

FIG. 11 is a drawing showing a three-dimensional histogram in a color image reduction process according to an embodiment of the present invention.

FIG. 12 is a diagram showing a data format of a range block and a representative range block in the color image reduction processing according to the embodiment of the present invention.

FIG. 13 is a diagram showing a reduced color image in the color image reduction processing according to the embodiment of the present invention.

FIG. 14 is a drawing showing a data format of a reduced color image in a color image reduction process according to an embodiment of the present invention.

FIG. 15 is a block diagram of color image reduction processing by a conventional color image reduction device.

FIG. 16 is a diagram showing a divided image and a reduced color image in a conventional color image reduction process.

[Explanation of symbols]

 24 ... Representative color generation means 26 ... Histogram creation means 28 ... Range selection output means 30 ... Reduced image creation means

Claims (4)

[Claims] 1. An image dividing means for dividing an original color image stored in an image storage area into a plurality of divided images, and a color having a maximum frequency among colors of pixels constituting the divided image is set as a representative color of the divided image. Representative color generating means for replacing one divided image with one pixel, and a reduced image generating means for generating a reduced color image using the representative color of the divided image as the color of the one pixel. Color image reduction device. 2. The color image reducing apparatus according to claim 1, wherein
The representative color generation means counts the number of pixels having a color belonging to each range block obtained by dividing the color space among the pixels forming the divided image, and a range including the largest number of pixels in the range block. A color image reduction device comprising: a range selection output unit that selects a block as a representative range block and outputs a color corresponding to the representative range block as a representative color of a divided image. 3. An original color image is divided into a plurality of divided images, and among the colors of the pixels forming the divided image, the color with the maximum frequency is used as a representative color of the divided image, and one divided image is replaced with one pixel. And a reduced color image is created by using the representative color of the divided image as the color of the one pixel. 4. The color image reduction method according to claim 3,
When determining the representative color, the number of pixels that have a color belonging to each range block obtained by dividing the color space is counted for the pixels that make up the divided image, and the range block that contains the largest number of pixels A color image reducing method characterized by selecting a range block and outputting a color corresponding to the representative range block as a representative color of a divided image.