JPH0121668B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an adaptive predictive decoding device used for predictively encoding halftone photographs used in newspapers and the like and decoding the transmitted code.

Conventionally, this type of adaptive predictive coding method compares the number of mispredicted pixels of multiple prediction error signals in a fixed block unit, such as the prediction method shown in Japanese Patent Application Laid-Open No. 55-41080. However, there is a method of selecting any one prediction error signal. This method has the disadvantage that the prediction effect is not sufficient if it is better to change the prediction method in the middle of a block, and the data compression rate is not sufficient because a mode code is required to indicate which prediction method was selected. It was hot.

An object of the present invention is to eliminate the drawbacks of the conventional prediction means described above, and to provide an adaptive predictive coding method that is highly effective in prediction and does not require a mode code.

The adaptive predictive decoding device of the present invention provides a plurality of predicted values P ₁ for the sample value X of the pixel x on the halftone photograph,
P ₂ , ......, P _o is generated, one predicted value Px is selected from multiple predicted values, and the predicted value Px and sample value
When decoding the code obtained by the adaptive predictive coding device that encodes the prediction error signal Y obtained by means for obtaining a sample value X of a pixel x; a means for obtaining a sample value X of the decoded pixel x; and a plurality of predicted values P ₁ , P ₂ , P _o for the sample value X of the decoded pixel x. , a plurality of pixels y ₁ having the same phase relationship with the previous pixel x on the two-dimensional mesh periodic pattern of the image with neighboring pixels of the pixel x,
y ₂ , ......, y _n and the plurality of pixels y ₁ , y ₂ , ...
..., y _n using sample values of neighboring pixels, and the plurality of predicted values P ₁ , P ₂ , ......, P _o
means for generating a selection signal using some sample values of the pixel used to generate the pixel and its neighboring pixels; and means for generating the plurality of predicted values according to the selection signal.
and means for selecting the predicted value P _x for the sample value X from P ₁ , P ₂ , . . . , P _o .

The adaptive predictive decoding device of the present invention uses means for generating a plurality of predicted values P ₁ , P ₂ , P _o on the encoding device side, and generates a plurality of predicted values P ₁ , P _{2 .} ,…
..., _{the predicted value P x for} the sample value Moreover, since there is no need for a mode code indicating which predicted value has been selected, this method has the effect of reducing the amount of information after encoding as a whole.

Next, the present invention will be explained in detail with reference to the drawings.

FIG. 1 is a diagram showing an example of reference pixels of a predictor used in the adaptive predictive decoding device of the present invention.
Note that FIG. 1 is for a 45° diagonal halftone image. In Figure 1, distance D (here D=10)
represents the period of the halftone dot, and the pixels that have the same phase relationship (on the two-dimensional mesh periodic pattern of the image) with the sample value X of the pixel x are y ₁ , y ₂ , ......y ₅ . In the figure, neighboring pixels a, b, c,
Sample values A, B, C, D, F, G of d, f, g,
Sample value Y ₁ of pixel y ₁ separated in the main scanning direction by halftone dot period and sample values _{B 1} , C ₁ , D 1 , G of its neighboring pixels b 1 , c ₁ , d ₁ , g ₁ , _{h 1 1} and H ₁ represents an example of reference pixels used in the predictor 3a ₁ in FIG. 2A. Similarly, the neighboring pixels a, b,
Sample values A, B, C, D, F, G of c, d, f, g
and the sample value Y ₂ of pixel y ₂ diagonally separated by the period of the halftone dot, and the sample values B ₂ , C ₂ , D ₂ of its neighboring pixels b ₂ , c ₂ , d ₂ , g ₂ , h ₂ , The combination of G ₂ and H ₂ is the 2nd A
This shows an example of reference pixels used in the predictor _3a2 shown in the figure, and the sample values Y _{3 , y 3} , y ₄ , y _n of pixels y 3 , y ₄ , . _Y4 ,
......, Y _n can be used to obtain an example of a reference pixel used in the predictor 3a ₃ in FIG. 2A.

FIG. 2A is a block diagram showing a first embodiment of an encoding device that produces codes to be decoded by the decoding device of the present invention, and is an example of the configuration when the reference pixels shown in FIG. 1 are used. be. In the figure, a sampled and binarized image signal is applied to terminal 1. Assuming that the sample value applied to terminal 1 is X, this sample value X is sent to line memory 2a, predictor 3a ₁ ,
3a ₂ , . . . , 3a _o and the prediction selection circuit 4. The line memory 2a delays the signal by about one line, and the delay amount for one line is set to H (=8000
sample), its output terminal will have (H-6)
A signal delayed by =7994 samples is obtained. Also, the line memories 2b ₁ , 2b ₂ , ......, 2bl each delay the input signal by one line, and their respective output terminals have 1, 2, ......, A signal delayed by l lines is obtained. Predictors 3a ₁ , 3a ₂ , ......, 3a _o are shift registers and read-only memory (ROM)
Using the combinations shown in FIG. 1 as reference pixels, the predicted values P ₁ and
Output P ₂ , ......, P _o . The prediction selection circuit 4 is
Using the sample values of the reference pixel used to generate the predicted values P ₁ , P ₂ , ......, P _o and some of its neighboring pixels, the halftone dot that has the highest similarity to the halftone dot containing the sample value X is A signal for selecting the predicted value Px obtained from the sample values of the halftone dots is output. The multiplexer 5 selects the predicted value Px pixel by pixel according to this selection signal, and the exclusive OR circuit 6 performs an exclusive OR with the sample value X to obtain a prediction error signal Y. The prediction error signal Y is then encoded by an encoder 7 (eg, a conventionally used encoder such as a run-length encoder) and outputted from an output terminal 8.

In addition, when encoding the prediction state signal in addition to the prediction error signal in the encoding of the encoder 7,
This can be achieved in a second embodiment of the encoding device according to the invention shown in FIG. 2B.

That is, the prediction state signal is a signal used to statistically examine the prediction probability for each reference pixel pattern in advance and to group the prediction error signals according to the magnitude of the prediction probability. For example, when there are two predicted states, the patterns are divided into patterns with a predicted accuracy probability of 0.95 or more and patterns with other patterns. In FIG. 2B, the predicted state signals S ₁ , S ₂ , ......, _So are the predicted values P ₁ , P ₂ , ...
..., _{P o} are output from predictors 3a' ₁ , 3a' ₂ , . The multiplexer 4′ receives the predicted value P _x
The prediction error signal S _x corresponding to the predicted value P x is selected in the same way as the predicted value P _x , and the selected prediction state signal S _x is input to the encoder 7'. The encoder 7' performs, for example, run-length encoding of the prediction error signal using the prediction state signal S _x . A run-length encoding method using such a prediction error signal and a prediction state signal is described in Japanese Patent Application Laid-Open No.
It is described in Publication No. 55-41080.

FIG. 3A shows a predictor 3a ₁ used in FIG. 2A.
FIG. 2 is a block diagram showing an example of an embodiment of the invention. terminal 1
Signals of the current scanning line and the previous scanning line are applied to 0 and 11. The sample value applied to terminal 10 is
Then shift registers 12a, 12b and 1
2c, pixels f, g, h ₁ , y ₁ and g ₁ in FIG.
Sample values F, G, H ₁ , Y ₁ and G ₁ are obtained.
A signal in which the sample value X is delayed by (H-6)=7794 sample times is applied to the terminal 11, and is further delayed by 4 sample times by the shift register 12d. In shift registers 12e, 12f and 12g, 1 to 4 and 11 to 13 sample times are delayed,
Sample values D, C, B, A, D ₁ , C ₁ and B ₁ of pixels d, c, b, a, d ₁ , c 1 and b ₁ in FIG _.
is obtained. The sample values of the reference pixels obtained in this way A, B, C, D, F, G, B ₁ , C ₁ , D ₁ ,
G ₁ , H ₁ and Y ₁ are applied as addresses to the ROM 13, and the aforementioned predicted value P ₁ is applied to its output terminal 14.
occurs. Neighboring pixels a, b, c, d, f, and g of pixel x are used as reference pixels, and pixel y ₂ and its neighboring pixels b ₂ , c ₂ , which are diagonally separated by the halftone dot period, are used as reference pixels.
The predictor 3a ₂ when using d ₂ , g ₂ , h ₂ can also be implemented with a circuit configuration using a shift register and ROM, and the following pixels y ₃ , y ₄ , . . . which have the same phase relationship as the pixel x
Predictors 3a ₃ , 3a ₄ , ... when using ..., y _n
..., 3a _o can also be implemented with a similar circuit configuration.

Above, we have described an example in which only one pixel having the same phase relationship with pixel x is selected as the reference pixel of the predictor, but multiple pixels having the same phase relationship with pixel x are simultaneously used as reference pixels. can. For example, neighboring pixels b, c, and g of x shown in FIG. 1 as reference pixels have the same phase relationship with x.
y ₁ , y ₂ , y ₃ and their neighboring pixels c ₁ , g ₁ , c ₂ , g ₂ ,
Sample values of c ₃ , g ₃ B, C, G, Y ₁ , C ₁ , G ₁ , Y ₂ ,
Of course, predictors using C ₂ , G ₂ , Y ₃ , C ₃ , G ₃ and the like are included.

FIG. 3B shows the predictor 3 used in FIG. 2B.
FIG. ₃ is a diagram showing an example of an embodiment of a'1. The difference in this figure from the predictor 3a ₁ in FIG. 3A is that a ROM for generating a prediction state signal S ₁ is added, and other aspects are exactly the same. The predicted state signal S ₁ is the predicted value
Similar to P ₁ , it is statistically determined in advance from several types of papers.

FIG. 4 is a block diagram showing a first embodiment of the prediction selection circuit 4 used in FIGS. 2A and 2B, in which halftone dots y ₁ , y ₂ and
This is an example of using three halftone dots centered on _y3 . Signals of the current scanning line, the previous scanning line, the 5th line, and the 6th previous scanning line are applied to the terminals 20, 21, 22, and 23, respectively. Assuming that the sample value applied to the terminal 20 is X, sample values G, _Y1 and _G1 of pixels g, _y1 and _g1 are obtained which are delayed by 1 to 11 samples in the shift register 24a. The sample values applied to terminals 21, 22, 23 are similarly transferred to shift registers 24b, 24c.
and sample values C, B _, C ₁ , Y 4 , G of pixels c, b, c ₁ , y _{4 , g 4} , y _{2 ,} g ₂ , c ₄ and c _{2 4} , _Y2 , _G2 , _C4 and _C2
is obtained. The sample values of the reference pixels obtained in this way B, C, G, C ₁ , G 1 , Y ₁ , C ₂ , G ₂ , _{Y 2 ,}
C ₄ , G ₄ and Y ₄ are applied as addresses to the ROM 25, and the predicted values P ₁ , P ₂ , P ₃ are applied to the output terminals.
A selection signal indicating which predicted value should be used is output. Here, the reference pixels used to generate the predictor selection signal are the sum of the reference pixels used to generate the prediction signal (in this case, three reference halftone dots are used).
30 pixels) are used. In other words,
A predictor that reduces the overall ROM capacity and does not deteriorate the prediction effect by determining which predictor to select with a small number of reference pixels and then predicting with the predictor that has the strongest correlation with the predicted pixel. can be configured. In the above example, in order to directly execute prediction using 30 pixels, a ROM of 2 ³⁰ ≈10 ⁹ bits is required, but with the configuration of the present invention, it can be implemented with a ROM of 5×2 ¹² =2×10 ⁴ bits.

Next, how this selection signal is determined will be explained. Here, as the reference pixel for selecting the predictor,
Since 12 pixels are used, ²¹² combinations are possible, and a selection signal must be found for each. First, predicted values when using the predictors 3a ₁ , 3a ₂ and 3a ₃ are obtained for all combinations of reference pixels of the predictors. The predicted value is determined based on total data such that for a certain combination of reference pixels, if the probability of black is 0.5 or more, the color is black, otherwise it is white. Next, select a predictor. When predicting using predictors 3a ₁ , 3a ₂ , and 3a ₃ for all combinations of pixels selected as reference pixels (the above-mentioned predicted value is used as the predicted value) ) is determined by the number of prediction errors or two-dimensional Markov entropy. A signal for selecting the predictor with the smallest value among them is written to the ROM. These values are determined in advance as an average of several types of paper, and are fixed.

FIG. 5 is a block diagram showing a second embodiment of the prediction selection circuit 4 used in FIGS. 2A and 2B, and similarly to FIG. 4, halftone dots y ₁ , y ₂ and This is an example of using three halftone dots centered on _y3 . Signals of the current scanning line, the previous scanning line, the 5th line, and the 6th previous scanning line are applied to the terminals 20, 21, 22, and 23, respectively. Letting the sample value applied to the terminal 20 be X,
The sample values G, F, G ₁ of pixels g, f, g ₁ and f ₁ are delayed by 1 to 12 samples in the shift register 24a'.
and F ₁ is obtained. Terminals 21, 22 and 2
Similarly, the sample values applied to pixels d, c, b, a, d ₁ ,
c ₁ , b ₁ , a ₁ , g ₄ , f ₄ , g ₂ , f ₂ , d ₄ , c ₄ , b ₄ , a ₄ ,
Sample values D, C, B, A of d _{2 ,} c ₂ , b ₂ and a 2 ,
D ₁ , C ₁ , B ₁ , A ₁ , G ₄ , F ₄ , G ₂ , F ₂ , D ₄ , C ₄ ,
B ₄ , A ₄ , D ₂ , C ₂ , B ₂ and A ₂ are obtained. Sample values A, B, C, D, F, G of neighboring pixels of sample value X
are exclusive OR circuits 27a, 27b and 27
c, the sample values A 1 , B 1 , B 1 , sample values Y ₁ , Y ₂ , Y ₃ of the pixels that are separated from the sample value X by the period of the halftone dot, and the sample values A ₁ , B ₁ ,
Corresponds to C ₁ , D ₁ , F ₁ , G ₁ and A ₂ , B ₂ , C ₂ , D ₂ , F ₂ , G ₂ and A ₃ , B ₃ , C ₃ , D ₃ , F ₃ , G ₃ respectively Exclusive OR is performed for each sample value, and the result is output. ROM28a, 28b and 28c
Then, the degree of similarity is calculated according to the result of the exclusive OR and is input to the comparator 29. In the comparator 29,
The results of the ROMs 28a, 28b, and 28c are compared, and a selection signal is output for selecting the sample value Yi of the pixel whose neighboring pixel is most similar to the neighboring pixel of the sample value X. An example of the ROM used here is:
One that directly counts the number of outliers in the output of the exclusive OR circuit, or one that calculates the weighted sum of each output.For example, for the sample values of A, F, and D, it is 1, and for the sample values of B, C, and G, You can think of something like multiplying the weight by 2.

Although the embodiments of the prediction selection circuit of the present invention have been described above, it goes without saying that these do not add any limitations to the present invention, and it goes without saying that other methods can be used as a measure for measuring the similarity of halftone dots. It is.

FIG. 6A shows a first embodiment of the adaptive predictive decoding device of the present invention, which decodes the code encoded by the adaptive predictive coding device of FIG. 2A. In the figure, the code applied to the terminal 31 is the code applied to the decoder 32.
is decoded into a prediction error signal Y. Prediction error signal Y
An exclusive OR circuit 6 performs an exclusive OR with the predicted value Px output from the multiplexer 5 to generate a decoded signal X. Other line memories 2a, 2b ₁ , 2b ₂ , ......, 2b _n , predictor 3a ₁ ,
3a ₂ , ...3a _o , the prediction selection circuit 4 and the multiplexer 5 are the first part of the adaptive predictive encoder of the present invention.
The circuit configuration is exactly the same as that of the embodiment shown in FIG. 2A, and the operation is the same. Moreover, the predictors 3a ₁ , 3a ₂ ,... of the present invention
..., 3a _o , the predictor shown in FIG. 3A is used.

FIG. 6B shows a second embodiment of the adaptive predictive decoding device of the present invention, which decodes the code encoded by the adaptive predictive coding device of FIG. 2B. The difference between the figure and FIG. 6A is that the decoder 32' performs run-length decoding of the prediction error signal using the prediction state signal, and the predictors 3a' ₁ , 3a' ₂ , .
_3a'o is used to generate a prediction error signal and a prediction state signal as shown in FIG. 3B.

As described above, the present invention uses a plurality of predictors, and the predicted value obtained from the predictors can be selected using sample values of the pixel used in the predictor and some of its neighboring pixels, so it is different from the conventional prediction method. It is an adaptive predictive coding device with a large prediction effect compared to the previous model.

[Brief explanation of drawings]

FIG. 1 is a diagram showing an example of reference pixels of a predictor used in the present invention, and FIGS. 2A and 2B are diagrams showing first and second reference pixels of an adaptive predictive coding device according to the present invention.
FIG. 3A and FIG.
FIG. B is a block diagram showing first and second examples of a predictor used in the present invention, and FIGS. 4 and 5 are block diagrams showing first and second examples of a prediction selection circuit used in the present invention. 6A and 6B are block diagrams showing first and second embodiments of the present invention. In the figure, reference numbers 2a, 2b ₁ , 2b ₂ ,...
..., 2b _n-1 and 2b _n are line memories, reference numbers 3a, 3a ₂ , ......, 3a _o , 3a' ₁ , 3a' ₂ , ...
..., 3a′ _o-1 and 3a′ _o are predictors, reference numeral 4
is a prediction selection circuit; reference numerals 5 and 5' are multiplexers; reference numeral 6 is an exclusive OR circuit; reference numerals 7 and 7' are encoders; reference numerals 12a, 1 are
2b, 12c, 12d, 12e, 12f, 12
g, 24a, 24b, 24c, 24d, 24a',
24b', 24c', and 24d' are shift registers;
Reference numerals 13, 13', 25, 28a, 28b and 28c represent ROM, reference numerals 27a, 27b and 27c represent exclusive OR circuits, reference numeral 29 represents a comparator, and reference numeral 32 represents a decoder.

Claims (1)

[Claims] 1 Generate a plurality of predicted values P ₁ , P ₂ , ......, P _o for the sample value X of pixel x on the halftone photograph, select one predicted value Px from the plurality of predicted values,
In an adaptive predictive coding device that decodes a code obtained by an adaptive predictive coding device that codes a prediction error signal Y obtained from the predicted value P _x and the sample value X, the prediction error signal Y is decoded. means for obtaining a sample value X of the decoded pixel x from the prediction error signal Y and the predicted value Px, and a plurality of predicted values P ₁ , P ₂ , . . . for the sample value X of the decoded pixel x. _Po
, a plurality of pixels y _{1 ,} y _{2 ,} . means for generating using sample values of neighboring pixels of y ₂ , ......, y _n ; and the plurality of predicted values P ₁ ,
P ₂ , ......, means for generating a selection signal using some sample values of the pixel used to generate P _o and its neighboring pixels; and means for generating the plurality of predicted values according to the selection signal. An adaptive predictive decoding device comprising means for selecting the predicted value P _x for the sample value X from P ₁ , P ₂ , P _o .