JP2713298B2 - Color image signal decoding device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [0001] BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to the decoding of color image signals.
Data compression decoding apparatus. [0002] 2. Description of the Related Art Conventionally, documents and drawings made of black and white have been scanned.
Black and white binary facsimile signal obtained by scanning with a canner
Run-length as a method for digitally compressing and encoding signals
There is a known coding scheme. This run-length code
The encoding scheme encodes a continuous length of white or black pixels.
Therefore, if white and black pixels occur continuously,
The compression efficiency is high because the number of components is reduced as a whole. Normal
This run-length encoding method
Facsimile coding country
Modified Huffman (Modi
modified Huffman) method and Modified Re
A Modified READ method is defined,
This has been adopted. On the other hand, as a method of displaying or recording a color image,
Are R (red), G (green),
B (blue) additive three primary colors, C in printer recording
(Cyan), M (Magenta) and Y (Yellow)
Each color component of primary colors (sometimes four colors by adding black)
Expressed by superimposing images. Express color halftones
In this case, each of these color component images is
Pixels) and colorless pixels (pixels without color)
The color that expresses the halftone in the proportion occupied by these pixels
-Pseudo halftone method (mesh method and dither method are typical examples
There is), and color continuous halftone method expressed by multi-valued color pixels
There is a law. [0004] However, color pseudo intermediate
According to the gray scale method, the halftone image signal changes periodically.
Since the image is binarized using the binarization threshold, colored pixels and colorless images
Many patterns occur in which elements repeat periodically. Therefore,
The run length is short and the number of runs is large, so the run length
The disadvantage is that the compression efficiency by encoding is extremely poor.
Was. In the color continuous halftone method, the difference
Differential encoding (Differential PCM) and conversion
A completely different coding method called encoding is used.
International standard run-length coding is not applicable
There was an inconvenience. An object of the present invention is to provide a color pseudo halftone image,
Perform simple preprocessing on color continuous halftone images
Modified Huffman (hereinafter referred to as MH)
Efficient data by run-length encoding such as
A decoding device for decoding the output of the encoding device to be compressed is provided.
To provide. [0007] SUMMARY OF THE INVENTION A color image signal of the present invention
The decoding device of
Run-length decoding of length-encoded prediction error signal
Means and run-length decoded for each block
The prediction error signal is recorded in units of
And write alternately in ascending order and descending order, and then
If the signal is GOOD, the prediction error signal
Signal is read out, and if the predicted state signal is BAD,
Read prediction error signal from memory in descending order and perform reverse array conversion
Means for predicting the color component image signal currently being decoded
The same color component image signal that has already been decoded
And the predicted state signal generated using the
Input to the reverse arrangement conversion means, and
Prediction error signal output from the column conversion means and the prediction signal
Is used to decode the prediction error signal into the original color component image signal.
And means to be performed. (Operation) Next, color image signal decoding of the present invention
Encoding device and the corresponding encoding device with reference to the drawings.
This will be described in detail. Hereinafter, C (cyan), M (magenta), Y (yes)
The following description focuses on the encoding of the three color component images
However, for R (red), G (green), and B (blue) color component images,
Can be encoded in the same way. FIG. 1 is a block diagram showing an example of the configuration of the present invention.
FIG. In the figure, the color component of the color image is
An image signal X (hereinafter abbreviated as an image signal) is input. Color
The input order of the minute image signal X is the same scanning of the color image.
Cyan, magenta, and yellow image signals for the line
May be input in units of one line (for example, cyan
One line, one line for magenta, one line for yellow
(Input in order), may be input in units of one pixel (for example,
For example, one pixel of cyan, one pixel of magenta, and one pixel of yellow
Enter). Also, for each screen, first, one screen of cyan
Screen, and then enter one screen for magenta and yellow in order.
You may force. Terminal 101 receives a timing signal.
It is. The timing signal is a synchronous signal for the image signal.
Signal and clock pulse. This timing signal
Can also distinguish the color of the image signal currently input from
Wear. The image signal X is input to the predictive conversion circuit 1
The image signal (the same color component as the image signal X)
An image signal and an image signal of a color component different from the image signal X
Is predicted as a reference pixel, and the prediction error
It is converted into a signal Y and a prediction state signal S. Prediction error signal Y
Is a 1-bit signal consisting of 0 and 1
0, and 1 when the prediction is incorrect. The predicted state signal is
1-bit signal indicating the degree of medium
GOOD when the condition is easy, BAD when the condition is easy
Each is represented by 0,1. Therefore, the predicted state signal
More predictive error signals are labeled GOOD or BAD
It is. Here, the configuration of the predictive conversion circuit is the object of encoding.
Image signal is a color pseudo-halftone signal,
These are different from the color continuous halftone signals of
Will be described later. The prediction error signal Y is generated by the array conversion circuit 2.
Array is converted for each block, and the
It becomes the difference signal U. This array conversion controls the prediction state signal S.
Run under Examples of one block are cyan and ma.
Prediction error corresponding to one scan line for each of Zenda and Yellow
It may be a signal, or a combination of these three lines
Block. The specific method of array conversion will be described later.
Labeling the prediction error signal GOOD and BAD
When While the prediction hit continues, the same label continues. The label changes as much as possible every time a forecast is missed
You. Array conversion is performed under the following rule. The array-transformed prediction error signal U is smoothed
The signal is converted into a smoothed signal V by the path 3. The smoothing circuit 3
For example, exclusive with the register 32 as shown in the broken line in the figure.
Signal OR, and the signal U has
The sign of the signal V is changed from 1 to 0 or from 0 to 1 each time it occurs.
It works to reverse. By this conversion, the signal U is reserved.
The interval at which mismeasurement occurs assumes that signal V is a monochrome binary signal.
Will now correspond to the run length when the The smoothed signal V is output from the run-length encoder 4.
Run-length code using different run-length codes
Is encoded. The international standard M is used as the run-length encoder.
An H encoder can be used. This run-length code
Of the prediction error signal labeled GOOD
Signal sequence is likely to be encoded with white run-length code
The prediction error signal sequence labeled BAD
Since the probability of being encoded with run-length codes increases,
Design run-length codes for each statistical property
Then, efficient run-length encoding is performed. Compressed code obtained by run-length encoding
The signal C is transmitted to the transmission line or stored in the file memory.
Is stored in Reference numeral 5 represents any of these.
Compression taken from transmission line or file memory
Code C 'is a decoding composed of reference numerals 6, 7, 8, and 9.
The original image signal X undergoes an inverse transform process of the encoding by the encoding device.
And output to the terminal 200. The compression code C 'is first supplied to a run-length decoder
6, and is converted into a signal V 'by run-length decoding.
From which the black and white transition points are extracted by the inverse smoothing circuit 7
Converted to U '. The inverse smoothing circuit 7 is, for example,
As shown in FIG.
Is done. The output of register 72 is
Signal, the black-and-white change point is displayed in the exclusive OR output.
Becomes 1, which corresponds to a misprediction. The signal U 'is
Further, the array is inversely transformed by the array inverse transformation circuit 8 and the prediction error signal is obtained.
No. Y '. Array inverse transform circuit is predictive inverse transform circuit
9 under the control of the predicted state signal S 'supplied from
Then, the signal U 'is rearranged by the inverse logic of the encoding.
Then, it is converted into a prediction error signal Y '. The prediction error signal Y 'is
An image signal already decoded by the prediction inverse transform circuit 9
(An image signal of the same color component as the image signal currently being decoded
And / or image signal of the color component
Is converted to an image signal X 'based on the predicted value. What
The timing signal is decoded from the run-length decoder 6.
To the output terminal 201
Will be issued. Here, errors in the transmission path and file memory
Otherwise, C '= C and the encoding and decoding operations are correct.
V '= V, U' = U, S '= S, Y'
= Y, X '= X. [0019] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG.
2 shows an embodiment of a predictive conversion circuit for signals. Also color simulation
The halftone method is used as the halftone method, and
An, magenta, and yellow, with one line as the unit
FIG. 2 shows an example of a reference pixel arrangement when inputting and encoding.
You. The coded pixel (the pixel to be coded)
In the case of cyan, other pixels on the scan line containing the coded pixel
The color component image signals (magenta, yellow) are still encoded
Since only the cyan image signal is used as the reference pixel
Used. Pixel X indicates a coded pixel, and pixel H is
Halftone dot L on the same line (between centers of adjacent halftone dots)
Only a distance away. Pixels G, H, I, J are pixels
Pixels on the same line as X, and pixels A, B, C, D,
E and F are pixels on one line. Encoded pixels are magenta
In the case of, the cyan image signal already on the same line
Since it has been coded, it can be used as a reference pixel. Shi
Ann's reference pixel K is the same as the magenta coded pixel X and the original pixel.
Pixels at the same position on the color image, and N is the same as H
The pixel at the position. When the coded pixel is yellow
Is the sign of the cyan and magenta image signals on the same line.
Since it has been converted, it can be used as a reference pixel. Maze
, The reference pixel P is at the same position as the coded pixel X, and Q is
Pixels at the same position. In any case, the number of reference pixels
Also has 10 pixels. In the example of FIG. 2, when the coded pixel is cyan,
Does not use magenta and yellow as reference pixels,
No yellow is used as a reference pixel in the case of
However, in either case, all pixels on the previous line are
Refer to this pixel because all color components have been encoded
It can also be used as a pixel. Also cyan, magenta,
Image signal X is input and encoded in pixel units in the order of yellow
If the color component pixel is different from the
On the same line as the coded pixel and located before the coded pixel.
All of the pixels can be used as reference pixels. Note that
When using the dither method as the color pseudo halftone method
Select the reference pixel corresponding to the dither matrix threshold
Do. Also, the number of reference pixels can be set arbitrarily. In FIG. 3A, the image signal X is turned off.
The memory 1 for cyan according to the color of X through the conversion circuit 11
2. Memory 13 for magenta, memory 14 for yellow
Is written to. When using the reference pixel of FIG.
Needs a capacity to store image signals for two lines.
You. The read-out switching circuit 15 draws a picture according to the color of the image signal X.
Read the reference pixel as illustrated in 2 from each memory
You. These reference pixels are stored in a prediction ROM (Read Onl).
y Memory) 16 is applied to the output and the prediction signal [0023] [Outside 1] And the predicted state signal S is extracted. The method of making the prediction ROM is as follows. Real
All the reference pixel pairs for each color component image signal
2 (in the example of FIG. 2, 2^Ten= 1024)
The appearance probabilities of the signal levels 0 and 1 of the image signal X with respect to
Using a plurality of test images selected in advance,
The signal level at which the probability becomes 0.5 or more is predicted
Measurement). Further, the predicted state signal S is the plurality of test states.
All combinations of reference pixels obtained using the default image
(In the example of FIG. 2, 2^Ten= 1024 ways)
One of the appearance probabilities of the signal levels 0 and 1 of the image signal X
One of them is more than a predetermined threshold (for example, 0.94)
The predicted state signal to GOOD,
The predicted state signal for the combination below the threshold is BAD
Is determined. The image signal X is a prediction signal [0027] [Outside 2] By [0029] (Equation 1) Is converted into a prediction error signal Y by the logical operation of
It is. However, [0031] [Outside 3] Represents an exclusive OR, which is shown in FIG.
The exclusive OR circuit is indicated by reference numeral 16.
The timing signal T supplied from the terminal 101 is written
Switching and read-out switching circuit and memory
Clock for reading and writing, and initial value of memory
Used for setting. FIG. 3B shows the prediction conversion circuit of FIG.
9 is an embodiment of a pair of predictive inverse transform circuits. See in the figure
Reference numeral 91 is a write switching circuit, and 95 is a read switching circuit.
Road, 96 is a prediction ROM, and 92, 93 and 94 are
Anne memory, Magenta memory, Yellow memo
And the same as those used in the predictive conversion circuit.
Work. Reference numeral 97 denotes an exclusive OR circuit.
is there. Therefore, the decoded signal X ′ is [0035] (Equation 2) Is given by At the beginning of a scan line, a timing signal
It is predicted that the initial state of each memory is set by T '.
Prediction signal output by OM97 [0038] [Outside 4] Is [0040] (Equation 3) ## EQU1 ## Also, if the prediction error signal Y 'is correctly
If decoded, Y '= Y. Therefore, equation (1) and
From equation (2) [0042] (Equation 4) Thus, the original image signal X is decoded. Return
The image signal which has been output is predicted through the readout switching circuit 95.
Predicted value applied to ROM [0044] [Outside 5] And outputs a predicted state signal S '. Therefore,
In the initial state, [0046] (Equation 5)If S '= S, then Y' = Y
If X '= X, decoding proceeds correctly. Next, array conversion and array inverse conversion will be described.
I will tell. FIG. 4A is a block diagram showing an embodiment of an array conversion circuit.
FIG. In the figure, the prediction error signal Y is one scanning line.
Is written to the memory 21 and then read out every time
The converted signal U is output. Array conversion is memory
Memory during the writing and reading process,
Address control corresponds to the array conversion method. FIG. 5 shows an example of the array conversion method. Figure smell
A1 indicates a time series of the prediction error signal of one scanning line. This
In this case, the number of pixels in one scanning line is set to 10 for simplicity.
Y_iThe subscript i indicates the time series number of the pixel.
Indicates a shift. Therefore, in the figure, Y₁ ~ Y_Three , Y₆
~ Y₈ , Y_TenIs 0, Y_Four , Y_Five , Y₉ Is 1. A2
Represents the time series of the prediction state signal S corresponding to the prediction error signal Y.
Show. G indicates GOOD and B indicates BAD. Prediction error signal
Yi is GOOD or not depending on the predicted state Si.
Labeled either BAD. That is, the figure
In the example, Y₁, Y_Two , Y_Four , Y₆ , Y₇ , Y₈ , Y₉ Is
GOOD, Y_Three , Y_Five , Y_TenIs BAD. FIG. 5A3 shows the prediction error written in the memory.
Indicates a signal. The addresses of the memory are 1, 2, ... from left to right.
Address 10 And in writing, GOOD
Write the labeled prediction errors in order from address 1,
The prediction error labeled on the BAD is changed to address 10 (one scan
Write at the last address of the line) or in the opposite direction. Write like this
In other words, the prediction error between GOOD and BAD is stored in the memory.
Each can be written separately. In the figure, note
The prediction error of GOOD is in addresses 1-7 of Lee, and addresses 8-10
A BAD prediction error is written in the ground. Next, the prediction error signal in the memory is read.
However, the read time-series signal is shown in FIG. 5A4. First,
Read in order from address 1 in the memory until a prediction error occurs.
put out. In the example shown, Y written at address 3_Four Should be expected
So, Y₁ , Y_Two , Y_FourIs read. Expected
When this occurs, the prediction error is calculated from the last address of the memory.
Again, reading is performed until a prediction error occurs. In the example shown, 9
Address Y₉ Is unexpected, so Y_Three And Y_Four Is read
You. Next, since GOOD is read again,₆, Y₇ ,
Y₈ , Y₉ Is read out and Y₉ Is not as expected, so
Is Y for BAD_TenIs read. In this way, the array
The conversion is completed. The array-converted prediction error signal U is shown in FIG.
A5 shows a signal obtained by smoothing the signal.
V is shown. In other words, when a prediction error occurs, white
A process of inverting black has been performed. A6 in FIG.
Run length when V is black and white run length encoded
Is shown. In the figure, W2 is a white run length 2
And B2 represents a black run length 2. Smoothing processing
The four signal runs W2, B2, W4, and B2
is there. On the other hand, the signal U (FIG.
Assuming that A4) is directly run-length encoded, the run is
W2, B1, W1, B1, W3, B1, W1
Therefore, the number of runs is significantly reduced by the smoothing process.
You can see that it is. The rules for array conversion are (1) Make the same label as continuous as possible (2) Change the label as much as possible for each misprediction
When It is the point of coding efficiency that satisfies the two conditions of
It is a good idea from. Of course, it is possible to restore
It is needless to say that is a precondition. here,
"As much as possible" means "if you can't do it,
Use other promises possible. " For example, as shown in FIG.
In this example, the prediction error signal is given by A1,
The state signal is Y₁ ~ Y_TenIf all were GOOD
Let's assume a match. Then, the memory will be as shown in Fig. 5B3.
Y as shown₁ ~ Y_TenAre written in the order of.
In this case, all data is labeled GOOD
It is not possible to read BAD next to GOOD
I can't. Read the memory as described above
Take ascending order and descending order, and switch between them for each prediction error.
Then, the prediction error signal after the array conversion is shown in FIG.
I will be. In this array conversion,
The label has not changed, but can the array be inverted?
In the present invention, this is permitted. Returning to FIG. 4A, the operation theory of the array conversion circuit
Make a light. In the figures, reference numerals 23 and 24 are respectively
Ascending address counter (up counter) and descending
This is a forward address counter (down counter). References
The character 22 is a logic circuit, and when writing, the predicted state signal S
Therefore, when S = 0 (GOOD), the up-count pulse
Is generated on the line 123, and when S = 1 (BAD),
A count pulse is generated on line 124. Also multi
The switching signal of the multiplexer 25 is supplied through the line 122.
You. This switching signal is used when the predicted state signal S
Things. Up counter 23 and down counter
24 is a timing pulse at the beginning of a scan line.
1 and 10 are loaded, respectively.
Memory address by address pulse
Let it. Ad estimated by two address counters
Is switched by multiplexer 25 and line 121 is
The data is supplied to the address lines of the memory 21 through the memory. This
As a result, the GOOD prediction errors are stored in the memory 21 in ascending order.
, BAD prediction errors are written in descending order. One scan line
When writing is completed, the image signal of the next scanning line arrives
Before reading the memory. Note that the next scanning line
If image signals arrive continuously, double the array conversion circuit.
And operate alternately. In reading
The read signal U is supplied to the flip-flop 26,
= 1 each time the flip-flop is inverted. Frizz
The output of the flop is provided to the logic circuit 22 via line 126.
Count pulse at reading and reading,
For generating pulse signals and switching signals for multiplexers
Can be. FIG. 4 (B) shows the array conversion shown in FIG. 4 (A).
2 shows an array inverse conversion circuit corresponding to the circuit. Reference numeral 81 is
Memory, 82 a logic circuit, 83 an up counter, 8
4 is a down counter, 85 is a multiplexer, 86 is
These are flip-flops used in the array conversion circuit.
It works the same with the same function. However, write and read
The operation of protrusion is opposite to that of the array conversion circuit. That is,
The array-converted prediction error signal U 'is first stored in the memory 81
The write address is controlled by U '= 1.
Output of flip-flop 86 which inverts the output every time
This is performed using a force signal. Also, the control of the read address is
Using the predicted state signal S 'supplied from the predictive inversion circuit
Do it. These address control methods use an array conversion circuit and
Is the same. The array conversion and its inverse conversion operation have been described above.
However, the array conversion method and circuit are in ascending memory address order.
・ Other than the descending order, various things can be considered. For example
Two memories are prepared, and the prediction error of GOOD is stored in the first memory.
Molly stores the BAD prediction error in the second memory in order.
The memory to be read out every time a prediction error occurs
May be changed to As described above, the color pseudo halftone signal
Is run-length
As the grease becomes longer and the number of runs is greatly reduced,
The compression efficiency is greatly improved. Next, the present invention is applied to a multivalued color continuous halftone image.
An embodiment when applied to a signal will be described. FIG.
1 shows an embodiment of a predictive conversion circuit for a multi-level image signal. Figure
, The multi-valued color image signal applied to the terminal 100
(Assumed to be n bits) is Bi by the gray converter 111.
from the “nary Code” to the “Gray Code”
You. Binary → Gray conversion increases data compression rate
The conversion table for 4 bits is shown below.
Things. [0059]                 Binary Code Gray Code Bit plane 1234 1234                       0000 0000                       0001 0001                       0010 0011                       0011 0010                       0100 0110                       0101 0111                       0110 0101                       0111 0100                       1000 1100                       1001 1101                       1010 1111                       1011 1110                       1100 1010                       1101 1011                       1110 1001                       1111 1000 Gray-converted multi-valued color image signal is written
Input to the path, if it is a cyan signal, cyan memory 1
13, a magenta memory 11 for a magenta signal.
4, if it is a yellow signal, a yellow memory 115
Is written. Each memory stores each image signal
bit (1st bit to nth bit)
It is configured to be readable. Image signal written
The signal is sequentially read out for each bit plane, and the prediction conversion is performed.
Done. That is, the encoded pixel readout circuit 116
The first bit is cyan, magenta, and yellow
The data is read in units of one scanning line, the next is the second bit, and the last is the
An n-bit signal is read. Corresponding to the coded pixel
Reference pixels are sequentially read out by the reference pixel readout circuit.
Is done. The arrangement of reference pixels is shown in Fig. 2 for each bit plane.
Can be used, but already coded images
It can be set arbitrarily from the prime. These reference pixels
A prediction signal applied to the prediction ROM 118 and output to the prediction ROM 118 [0060] [Outside 6]Then, the prediction state signal S is taken out. Again
The encoded pixel signal X is predicted by the exclusive OR circuit 119.
signal [0062] [Outside 7] Is converted to a prediction error signal Y.
You. The prediction error signal Y is the same as the color pseudo halftone signal.
Run-length encoding after array transformation and smoothing
Is done. One scan line of cyan, magenta, and yellow
After one bit is encoded, the second bit and finally the nth bit
is encoded, and encoding of the multi-level signal for one scanning line is completed.
Complete. However, the coding order of each pixel is
(113, 114, 115).
Can be set arbitrarily. Also prediction ROM
The contents of 118 are designed separately for each bit plane.
Therefore, it is better to set the value for all planes.
Also increases compression efficiency. FIG. 6B shows a prediction for a color multi-valued image signal.
6A shows an inverse conversion circuit, and includes a prediction conversion circuit shown in FIG.
Pair. The prediction error signal Y 'for each bit plane is excluded.
When applied to the other OR circuit 999, the corresponding reference image
Predicted signal predicted from elementary [0065] [Outside 8] By the above operation, the bit plane signal X 'is obtained.
Decrypted. The decoded bit plane signal X 'is written
Memory for cyan when cyan
-993, in the case of magenta, the memory for magenta,
For yellow, the corresponding bit in the yellow memory
The data is written in the portions (1 to n). 1st to nth bits
Read-out switching when prediction reverse conversion is completed for plane
By the circuit 992, in order of cyan, magenta, and yellow
1st to nth bits are read from memory in one scanning unit
And a binary signal by the binary converter 991
And the multi-level image signal Xa 'is taken out to the terminal 200.
Is done. The principles and features of the present invention can be summarized as follows.
become. First, in the present invention, first, prediction of an image signal is called.
Means that the color component image signal currently being encoded is already
Encoded image signal (image of the same color component as the encoded pixel
Signal and / or image signal of different color component)
To predict and convert it to a prediction error signal. This conversion allows
Thus, the redundancy of the image signal is reduced. Same color as coded pixel
Different from using only the image signal of the component as the reference pixel
It is better to use image signals of different color components as reference pixels.
Higher reduction efficiency and higher compression efficiency. Joke
The image signal of reduced length, that is, the prediction error signal
In order to represent the code with as few bits as possible,
Guess coding is used. Where the run-length encoding
The average run length is compared as in MH coding.
Code and average run length suitable for white run
Are two types of black run codes suitable for
Prepare length codes and use them as white, black, white, black, etc.
The run length code is alternately changed. The present invention
In order to adapt to such run-length coding,
The prediction error signal is converted. Two types of conversion
The first is that a prediction error is caused by a prediction state signal.
The difference signal is predicted by the group (GOOD)
Two groups of hard-to-hit groups (BAD)
This is a means for attaching and converting the array. By this means,
GOOD is a group with a relatively long average run length, B
AD falls into groups with relatively short average run lengths
Can be The second means should predict the prediction error signal
This is a means for encoding the
In the meantime, the prediction error signal is
Apply a lubrication process. GOOD and BAD are exchanged for each misprediction
GOOD of GOOD is realized by array conversion and smoothing
Runs are white run length codes, BAD runs are black run lengths.
The probability of being encoded with long-length codes,
It can be higher. [0068] According to the present invention, the conversion process is performed using standard MH codes.
MH coding by applying before and after the encoder and decoder
Color pseudo-halftone images without any algorithm changes
And efficient data compression of color multi-valued images
It can be applied to various purposes.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram showing a configuration of the present invention. FIG. 2 is a schematic diagram illustrating an example of a reference pixel arrangement. FIGS. 3A and 3B are block diagrams showing an example of a prediction conversion circuit and a prediction inverse conversion circuit for a color pseudo halftone image signal, respectively. FIGS. 4A and 4B are block diagrams illustrating examples of an array conversion circuit and an array inverse conversion circuit, respectively. FIG. 5 is a schematic diagram for explaining an array conversion operation. FIGS. 6A and 6B are block diagrams showing an example of a predictive conversion circuit and an inverse conversion circuit for a color multilevel image signal, respectively. [Description of Code] 1 prediction conversion circuit 2 array conversion circuit 3 smoothing circuit 4 run-length encoder 5 file memory or transmission line 6 run-length decoder 7 inverse smoothing circuit 8 array inverse conversion circuit 9 prediction inverse conversion circuit 31, 71, 87, 97, 119, 999 Exclusive OR circuit 32, 72 Register 11, 91 Write switch circuit 15, 95 Read switch circuit 16, 96 Predictive ROM 12, 92 Cyan memory 13, 93 Magenta memory 14, 94 Yellow memory 21, 81 Memory 22, 82 Logic circuit 23, 83 Up counter 24, 84 Down counter 25, 85 Multiplexer 26, 86 Flip-flop 111 Gray converter 991 Binary converter 112 Write switch circuit 996 Write switch circuit 992 Read Switching circuit 113, 993 Memory for cyan 114,994 Memory for magenta 115,995 Memory for yellow 116 Encoded pixel readout circuit 117,997 Reference pixel readout circuit 118,998 Prediction ROM

Claims (1)

(57) [Claims] Means for run-length decoding the run-length coded prediction error signal with the prediction loss as a break of the run, and ascending order of the memory with the run-length decoded prediction error signal for each block in units of the prediction loss And in the descending order, and the predicted state signal is GOO
Means for reading prediction error signals from the memory in ascending order in the case of D, and reading prediction error signals in descending order from the memory in the case of a prediction state signal of BAD, and performing reverse array conversion;
A prediction signal and a prediction state signal for the color component image signal currently being decoded are generated using the already decoded same color component image signal and other color component image signals, and the prediction state signal is input to the inverse arrangement conversion means. A color image signal comprising: a prediction error signal output from the inverse arrangement conversion means; and means for decoding the prediction error signal into an original color component image signal using the prediction signal. Decoding device. 2. As an inverse array conversion unit, a prediction error signal, which is run-length decoded for each block, is written alternately in the first memory and the second memory in units of up to prediction error, and when the prediction state signal is GOOD, 2. A means for reading a prediction error signal from said first memory and reading out the prediction error signal from said second memory and performing reverse array conversion when the prediction state signal is BAD. An apparatus for decoding a color image signal according to any one of the preceding claims.