JPS63190474A - Color picture data coder - Google Patents

Abstract

PURPOSE:To code a color picture data with multi-color with efficient compression by providing two coding means adopting different coding systems. CONSTITUTION:The tilted device consist of a buffer memory 1 dividing a picture data into 3X3 blocks, a binary color picture coding circuit 100 suitable for coding a binary picture such as a line drawing or a character or a picture mainly, a multi-level color picture coding circuit 101 suitable for coding a picture such as a photograph or a picture, and a decision and selection circuit 102 deciding by which of the circuit 101 or 100 a color picture data in a block is to be coded and selecting the circuit. A picture data is inputted to a gradient calculation circuit 6 of the circuit 101 and the change in the color picture data level is calculated as to 4 directions and the sign and change amount in the direction having the largest change are outputted and one representative color, the direction taking a maximum gradient in the output of a gradient calculation circuit 6 and the gradient are used as one block code. Thus, efficient compression coding is applied.

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Field of Application] The present invention relates to a color image data encoding device 1. In particular, the present invention relates to a block encoding device that divides a color image into adjacent blocks and encodes the divided blocks.

[Prior Art] Conventionally, this type of device has been configured using one type of encoding method. For example, as shown in Figure 2(a), the color image data contained in a 4x4 pixel block (CO8 to C33) is divided into two representative colors (CA) determined by that block.
, CB), approximate these representative colors for each pixel, and replace it with a signal indicating which of the above two representative colors it has been classified into.

With this encoding method, there is no problem if there is little variation in the color of pixels within a block, as shown in FIG. 2(a), but
For example, as shown in FIG. 2(b), cx +
If there are three colors c, + cz, and this block is encoded into two representative colors CxC2°, C1 will be classified and converted to C2', so the reproduced pixel block will be The disadvantage was that only two color signals were reproduced and the quality was poor. Therefore, if the number of representative colors is simply increased, the reproduced image will certainly be reproduced more faithfully, but on the contrary, the compression efficiency will decrease. In this way, the fact that three or more colors are included in a block is especially true when, for example, line drawings such as characters of multiple colors are included.

[Problems to be Solved by the Invention] The present invention was proposed in order to eliminate the drawbacks of the above-mentioned conventional example, and its purpose is to solve the problem even when a block contains multicolor color signals. It is an object of the present invention to provide a color image data encoding device that efficiently performs compression encoding, and furthermore, a color image data encoding device that enables encoding and reproduction faithfully to the original image, and at the same time has improved encoding efficiency.

[Means for Solving the Problems] The configuration of the present invention for realizing the above-mentioned problems includes a dividing means for dividing two-dimensional color image data into blocks each consisting of a plurality of adjacent pixels, and a dividing means for dividing two-dimensional color image data into blocks each consisting of a plurality of adjacent pixels; a first code that selects representative color image data of at least one representative color from the image data and encodes it into a first encoding code by associating each pixel in the block with the representative color image data of the at least one color; and
Select representative color image data of one color that represents the color image data of the block, calculate the gradient of the color image data value within the block, and perform second encoding of the representative color image data of one color and the gradient. and second encoding means for encoding into a code.

Furthermore, another configuration of the present invention for achieving the above object includes a dividing means for dividing two-dimensional color image data into blocks each consisting of a plurality of adjacent pixels; a first encoding means for selecting representative color image data of one color and encoding it into a first encoding code by associating each pixel in the block with the representative color image data of the at least one color; Select representative color image data of one color that represents the color image data, calculate the gradient of the color image data value within the block, and encode the representative color image data of one color and the gradient into a second encoding code. a second encoding means for determining the image area within the block; a third encoding means for encoding the determination result of the determination means into a third encoding code; selecting means for selecting the first encoded code or the second encoded code according to the selected encoded code and outputting the selected first or second encoded code and the third encoded code; Be prepared.

[Operation] According to the present invention having the above configuration, when there is a plurality of representative colors in an image block, the first encoding means encodes the representative colors without losing them, and the second encoding means encodes the representative colors within the image block. If one representative color is sufficient for a given image, that representative color is used for encoding.

According to another configuration of the present invention, the determining means recognizes the image area within the image block, and the selecting means selects the first image area according to the recognition.
Since the encoding code or the second encoding code is selected, compression efficiency is improved and there is no image deterioration during playback.

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

Fig. 3 shows a color image data encoding device according to an embodiment of the present invention, and its configuration is roughly divided into four components. The four component parts are image data 3x3
Buffer memory 1 divided into blocks and line drawing 1
A binary color image encoding circuit 100 suitable for encoding binary images such as character images, and a multilevel color image encoding circuit 101 suitable for encoding images such as photographs and paintings.
Then, it is determined and selected which of the binary color image data encoding circuit 100 or the multilevel color image data encoding circuit 101 is optimal for encoding the color image data in the block. It consists of a judgment selection circuit 102. These components will be explained below in order.

<Binary color image data encoding circuit> The configuration of this binary color image data encoding circuit 100 includes a representative color selection circuit 2 that selects two types of representative colors, and a representative color selection circuit 2 that selects two types of representative colors, and a representative color selection circuit 2 that selects each pixel in a block by selecting these representative colors. It consists of a binarization circuit 3 that classifies the color into one of the following and outputs the classification result as a binary value, and an encoding circuit 4 that performs block encoding of the representative color and the binary code of the classification result.

The general operation of this binary color image data encoding circuit 100 is as follows. Now, it is assumed that the 3×3 block shown in FIG. 3 is to be encoded. In FIG. 1, color image data collected into 3×3=9 pixels by a buffer memory 1 is input to a representative color selection circuit 2, where two representative colors are determined. The color image data of the two representative colors determined in this way and the color image data of nine pixels are manually input to the binarization circuit 3.

This circuit 3 classifies the nine color image data into one of the two representative colors by determining the data distance between the representative color and each color image data. Then, the two representative colors and the binary code indicating which of these colors is classified are converted into one block code by the encoding circuit 4. Needless to say, this binary code becomes multi-valued if the representative color selection circuit 2 selects three or more representative colors.

Representative color selection circuit 2> This circuit 2 will be explained according to FIG.

Blocked color image data is processed by a signal separation circuit 41
The components are separated by Here, the separated signal components are represented by Y (luminance), 1. Let it be Q (color difference). The component-separated and shaped color image data is input to signal calculation circuits 42, 43, and 44 for Y, I, and Q, respectively.

Since the signal calculation circuits 42, 43, and 44 are the same in circuit terms, the Y signal calculation circuit 42 will be described below. This Y signal calculation circuit 42 calculates and holds the maximum value YeaK, the minimum value Yl@ln, the difference between the maximum value and the minimum value, etc. in the block for the Y color image data.

As for color image data I and Q, similarly to Y, the maximum value, minimum value, and the difference between them are held in each latch. The maximum and minimum values for each of the Y signal, ■ signal, and Q signal are manually input to the signal selection circuit 46. On the other hand, each Y,
The difference between the maximum and minimum values for I and Q is input to the conversion circuit 45. In the conversion circuit 45, the difference signals are compared after being appropriately weighted, and a signal representing the component giving the largest difference among them is output.

The outputs of the signal calculation circuits 42, 43, and 44 (y-ax.

Y, , n, I Output as color image data that gives maximum and minimum values. That is, the representative color selection circuit 2 determines the component with the largest color change from among the color component signals of the nine pixel blocks, and outputs the maximum and minimum values for that color component. The determined maximum value (hereinafter referred to as the first representative color) and minimum value (second representative color) are the two representative colors. In the example of FIG. 2, this means outputting the representative color as CAC6 or ax CY cz.

<Binarization circuit 3> An example of the configuration of the binarization circuit 3 is shown in FIG. Since there are two representative colors in this embodiment, the binarization circuit 3 determines which of the above two representative colors each pixel is closer to for each pixel in the block, and outputs the determination result. Output as binary value. In FIG. 5, 51.52 is the first . This is a register that holds color image data of the second representative color for each color component.

The subtraction circuits 53 and 54 are circuits for determining, for each pixel in the block, which of the two determined representative colors is less likely to have a different color. That is, 53.54 is a subtraction circuit that calculates the difference for each component between the color image data of one pixel in the block from buffer memory 1 and the representative color, and 55.56 is a subtraction circuit that weights each component with respect to the difference. The weighting circuit is a circuit, and 57.58 is a circuit that calculates the sum of squares. In this way, the "image data distance" between the color image data of one pixel in the block and the representative color is calculated. When the amount related to the image data distance between one pixel and two representative colors is calculated, the comparison circuit 59 outputs a signal indicating the representative color with the smaller distance, that is, the smaller sum of squares.

For example, if you want to select the first representative color, select °゛1゛.

If the second representative color is selected, 0'' is output.The above sequence is repeated for each pixel, and all pixels in the block are associated with the representative color.Color image data of the representative color and a code indicating which representative color each pixel is classified into are formatted as inseparable codes by the encoding circuit 4. That is, the encoding circuit 4
The output is as shown in FIG. The format is a first representative color signal, a second representative color signal, and 00°+
These are the classification results for pixels of CI O"""22, etc.

<Outline of multivalued color image encoding circuit> 101 in Figure 1
is a multilevel color image encoding circuit. This encoding circuit is for efficiently encoding color images with little color change. One block of nine color image data is manually input to the representative color selection circuit 5, and one representative color is selected.

For example, since a color image with little color change is handled, this representative color is an intermediate color among the nine color image data in the block. The nine color image data are sent to the gradient calculation circuit 6.
, the change in color image data level is calculated for the four directions shown in FIG. 7, and the sign and amount of change in the direction with the largest change are output. However, the encoding order may be determined for all of the four directions, and the gradients may be encoded in each class. In this way, in the encoding circuit 7, one representative color and the direction and gradient in which the gradient outputted by the gradient calculating circuit 6 is the maximum form one block code. The above is an outline of the multivalued color image data encoding circuit 101.

It will be explained in more detail. First, the representative color selection circuit 5 will be explained.

<Representative Color Selection Circuit 5> FIG. 8 shows an example of the configuration of the representative color selection circuit 5. In FIG. 8, color image data divided into nine pixels per block is processed by a signal separation circuit 61 into three components Y (luminance), 1. It is separated into Q (color difference). These separated color image data are processed by arithmetic circuits 62, 63 and 64, respectively.
, the maximum value of each component (Y, 1., □
8, etc.) and the minimum value (Y + aln + I + nln, etc.)
is detected, and the median value of the maximum and minimum values (Y□8
÷Y1n/2) etc. is given. The signal synthesis circuit 65 synthesizes the median value of each component to obtain color image data as a representative color.

<Gradient Calculation Circuit> FIG. 9 shows a configuration example of the slope calculation circuit 6. This circuit extracts the tendency of the color gradient of the block color image data cut out by the buffer memory l, and 71 separates the blocked color image data into components similar to the separation circuit 61 of FIG. (for example, YIQ), and 72.73.74 is an arithmetic circuit that calculates the gradient of each component. The arithmetic circuits 72, 73, and 74 are arithmetic circuits 75, 74, respectively, which calculate the horizontal inclination, the vertical inclination, the downward slope to the right, and the upward slope to the right for the block shown in FIG. 76, 77, and 78, and an encoding circuit 79 that outputs the gradient of the component of the color image data within the block as a code.

First, the signals of nine pixels of one block (Cao to C2□
) is separated by the signal separation circuit 71 into a component for Y (
cI. . (l□.,..., C'22) and the component for ■ (C2oo, C2+o, ・'', C
22□) and the component for Q (C'Go + C31
0+ +C322), and the component for Y is manually inputted to an arithmetic circuit 72, the component for (2) to an arithmetic circuit 73, and the component for Q to an arithmetic circuit 74, respectively. Inside each arithmetic circuit, taking the component of Y as an example and explaining it, according to the notation in Fig. 7, the horizontal slope ((C'oo + C'or + C'0
2) - (C'2G + C'21 + C'z2)
), vertical slope ((C'oo + C'lo +
C'2G) - (C'02 + C'+2 + C
'22)), downward slope to the right ((c'o.+
01°, 10C'. )(C'22 + C' +2 +
C'211), upward slope to the right ((C1゜2 +
C'or + C'+2) - (C'2o+ C'to
+C'2+)1 is calculated. The same applies to the other I and Q components. These four gradient values represent the gradient tendency of the color image data within the block.

The encoding circuits 79, 80, etc. each select the largest gradient among the four directions and encode it. That is, the direction and slope encoded codes are output. Therefore, the slope calculation circuit 6
For the three YMQ components, the direction of the maximum gradient and its value are encoded and output. These encoded codes are output to the encoding circuit 7 in FIG. In this circuit 7, the three components of YIQ are converted into color image data of the final color system, and the direction of the maximum gradient and its gradient value in that color system are encoded. As shown in FIG. 2, a unique identification code indicating that the block has been encoded by a multivalued color image encoding circuit and color image data of the representative color within the block are encoded.

Incidentally, FIG. 10(b) is a modification example in which encoding is performed for all four directions without selecting the maximum gradient direction from among the four directions.

These encoded codes are input manually to the signal selection circuit 9 in FIG.

As described above, by the two code circuits 100 and 101,
Two encoded codes encoded separately using different encoding methods are output.

<Selection of encoded code> What is output from these encoding circuits are encoded codes as shown in FIG. 6 or 10, and these encoded codes are input to the signal selection circuit 9. FIG. 11 shows a schematic representation of reproduced images according to each encoding.

Now, one of these two block encoding codes is selected by the signal output from the encoding method selection circuit 8. Information for distinguishing which one has been selected is added to the block code output from the signal selection circuit 9.

<Encoding method determination circuit> Selection signal 12 for selecting one of the above two encoding codes
is output from the encoding method determination circuit 8. This encoding method determination circuit 8 is a circuit that determines the image tone or image area of pixels within a block. That is, it is determined whether the block should be encoded by the binary color image data encoding circuit 100 or the multilevel color image data encoding circuit 101.

This encoding determination circuit calculates the variance σ2 of color image data. The calculation formula is as follows.

σ2=Σα(CI-C')・(CI-C)/9 Here, CI is the color image vector of the i-th pixel in the block, and when expressed in RGB system, C+=(RI, G+
, B I). Further, C is the intra-block average value of color image data, and α is a diagonal matrix component for weighting. If you want to find the variance 02 of luminance Y, if you want to find the variance 02 of color difference ■ or Q, you can decide as follows. This is RGB signal and NTSC
This was determined using the relationship with the signal. More easily, when finding the dispersion using the G component of the RGB signal,
It is also possible to set α to .

FIG. 11 shows an example of the encoding determination circuit. In the circuit of FIG. 11, dispersion calculation circuits 91, 92, and 93 calculate the dispersion for each component of the color image data separated into, for example, YIQ by the signal separation circuit 90. These three variances are input to a comparison circuit 95, where they are compared with a predetermined threshold value. In this case, the comparator circuit 9
In step 5, the selection signal 12 is generated by, for example, weighting the Y (luminance) component so that the variance value is emphasized. Dispersion indicates the degree of variation in color image data within a block. Therefore, in a binary image such as a line drawing, the value is large, and in a multivalued color image with little change, it is small. The result of the determination is reflected in either the multivalued image. Therefore, this selection signal 12 is input to the signal selection circuit 9, and the selected encoding code is encoded by the encoding method most suitable for encoding that block.

In addition, as a method for determining the image area or image tone within a block,
In addition to finding the variance, we also find the data distance between two adjacent pixels, find out how many pixel bears where the distance is greater than a certain threshold exist in the block, and then calculate the There is also a method of recognizing the image area or image tone of the block.

<Decoding> Next, decoding will be briefly explained. At the time of decoding, which encoding method has been used to encode the encoded code to be decoded is known from the identification code shown in FIG. 6, FIG. 10(a), or FIG. 10(b). If encoded by the binary color image data encoding circuit 100 (code shown in FIG. 6), each pixel is sequentially assigned to one of two representative colors according to its type code and decoded. If it is encoded by the multilevel color image data encoding circuit 101 (code in FIG. 10), the central pixel of the block (C++)
Assign color image data of representative colors to . For pixels around the center, the gradient is decoded and proportionally distributed from the gradient to obtain color image data for each pixel.

Modification> In the embodiment described above, the signal separation circuits 41, 61, 71 .
Although the YIQ color system was used in 90, the color system is not limited thereto, and may be the L*a*b* system or the L*u*v* system.

Furthermore, in the binary color image data encoding circuit 100,
Although the description has been made using two representative colors, this is not limited to two colors, and can be considered as n colors. This is because even in the same line drawing, lines of three or more colors may exist in the background color.

In addition, in the binary color image data encoding circuit 100, the distance between the pixel color image data and the representative color image data is calculated using the sum of squares, but due to the nature of the signal, the sum of absolute values is sufficient. It is also possible to put it away. In this case, the advantage is that the circuit configuration becomes simple. Further, although the standard for determining the representative color is that the maximum and minimum components are far apart from each other, the pixels may be divided into several groups and the average of the groups may be used as the representative color.

Furthermore, in the multivalued color image data encoding circuit 101,
Although the representative color takes the median value of each component, this may also be the value of a pixel at a specific position within the block.

Both of the above two encoding methods have been described on the assumption that 3×3 blocking is performed, but it is also possible to adopt 4×4 or other blocking.

[Effects of the Invention] As explained above, according to the present invention, multicolor color image data can be efficiently processed by providing two encoding means that operate in parallel and employ different encoding methods. Can be compressed and encoded.

Further, according to another invention, since encoding is performed after determining which of the above two encoding means should be used for encoding, it is possible to faithfully reproduce the original picture at the time of decoding.

[Brief explanation of the drawing]

Fig. 1 is a diagram showing the overall outline of the encoding device of the embodiment according to the present invention, Fig. 2 (a) and (b) are diagrams explaining the drawbacks of the conventional example, and Fig. 3 is used in the embodiment. 4 is a block diagram of the representative color selection circuit 2, FIG. 5 is a block diagram of a binarization circuit that associates pixels with representative colors, and FIG. 6 is a block diagram of binary color image data. A format diagram of the encoded code by the encoding circuit, FIG. 7 is a diagram showing the direction of the gradient calculated by the gradient calculation circuit 6, FIG. 8 is a configuration diagram of the gradient calculation circuit 6, and FIG. 9 is a multivalued color A configuration diagram of the representative color selection circuit 5 used in the image data encoding circuit, FIGS. 10(a) and (b)
11 is a diagram showing an example of the format of the encoded code encoded by the multilevel color image data encoding circuit, FIG. 11 is a circuit diagram of an example of the encoding determination circuit, and FIG. 12 (
a) to (C) are diagrams schematically illustrating the differences between two encoding methods used in this embodiment. In the figure, 1... Buffer memory, 2.5... Representative color selection circuit, 3... Binarization circuit, 4... Encoding circuit, 6...
Gradient calculation circuit, 7... Encoding circuit, 8... Encoding method determination circuit, 9... Signal selection circuit, 10.11...
Block code, 12... selection signal. Patent applicant Canon Co., Ltd.

Claims (4)

[Claims] (1) A dividing means that divides two-dimensional color image data into blocks each consisting of a plurality of adjacent pixels, and selects representative color image data of at least one color from the blocks of color image data, and A first method for encoding each pixel into a first encoding code by associating each pixel with representative color image data of the at least one color.
an encoding means for selecting representative color image data of one color representing the color image data of the block, calculating the gradient of the color image data value within the block, and combining the representative color image data of the one color and the gradient; A color image data encoding device comprising: second encoding means for encoding into a second encoded code. (2) A dividing means that divides two-dimensional color image data into blocks each consisting of a plurality of adjacent pixels, and selects representative color image data of at least one color from the blocks of color image data, and A first method for encoding each pixel into a first encoding code by associating each pixel with representative color image data of the at least one color.
an encoding means for selecting representative color image data of one color representing the color image data of the block, calculating the gradient of the color image data value within the block, and combining the representative color image data of the one color and the gradient; a second encoding means for encoding into a second encoding code, a determining means for determining the image area within the block, and a third encoding means for encoding the determination result of the determining means into a third encoding code. 3 encoding means, selecting the first encoding code or the second encoding code according to the determination result, and combining the selected first or second encoding code with the third encoding code; A color image data encoding device comprising a selection means for outputting an encoded code. (3) The determining means determines whether the image area of the block is a first image block representing characters/line drawings or a second image block representing photographs/paintings based on the dispersion value of the color image data in the block. The selecting means selects the first encoded code when the determining means determines that the image block is a first image block, and selects the first encoded code when the determining means determines that the image block is a second image block. 2. The color image data encoding device according to claim 1, wherein two encoding codes are selected. (4) The determining means determines the number of two adjacent pixels in the block whose color image data distance is greater than a predetermined threshold; Color image data encoding according to claim 1, characterized in that when the number is large, the first encoding code is selected, and when the number is small, the second encoding code is selected. Device.