JPH04239271A - Picture data compressor - Google Patents

Abstract

PURPOSE:To improve an encoding efficiency, and to operate the efficient compression of a pseudo halftone reproducing picture, as for a picture data compression which uses the pseudo halftone reproducing picture by a random dither method. CONSTITUTION:This device is equipped with a pseudo halftone processing part 1 which converts a multilevel picture signal into a binary signal by using a pseudo halftone reproducing method, line memory 2 which stores a pseudo halftone binary signal, prediction processing part 4 which defines one of the pseudo halftone binary signals as a noticed picture signal, classifies the noticed picture element by searching both the signal state value of an area adjacent to the noticed picture element and density coefficient value of an area near the noticed picture element, and decides and outputs the prediction signal of the noticed picture element by using a prediction table 5. And also, this device is equipped with an exclusive OR circuit 3 which compares the prediction signal from this prediction processing part with the noticed picture signal from the pseudo halftone processing part, and outputs a prediction error signal '0' indicating that the prediction comes true and a prediction error signal '1' which indicates that the prediction doesn't come true, and an encoding part 7 which encodes the prediction error signal.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention is applied to data compression for facsimiles and the like, and particularly relates to an image data compression apparatus that binarizes a multivalued image such as a photograph using a pseudo halftone reproduction method and then encodes it.

0002

[Prior Art] A systematic dithering method is known as a method for converting a multivalued image signal such as a photograph into a binary signal and reproducing the gradation in a pseudo manner. As an example, a method is known in which prediction is performed using processed signals that have equal or similar thresholds for binarization, and the prediction error signal is encoded. On the other hand, recently, the error diffusion method, the minimum average error method, and the like have been attracting attention as binarization methods with excellent gradation reproducibility.

However, a signal sequence that has been binarized by a random dither method such as the error diffusion method or the minimum average error method described above has been binarized by a systematic dither method that has been commonly used. Since there is no periodicity like a signal sequence, and white "0" and black "1" appear randomly, the prediction method applied to the pseudo-halftone reproduction image binarized by the systematic dither method can be used as is. It cannot be applied to pseudo-halftone reproduced images using the dither method.

0004

As described above, in the past, it has not been possible to sufficiently improve encoding efficiency in image data compression using a pseudo-halftone reproduced image by the random dither method.

[0005] Therefore, the present invention provides an image data compression device that can sufficiently improve encoding efficiency and efficiently compress pseudo-halftone reproduced images in image data compression using pseudo-halftone reproduced images using the random dither method. This is what we are trying to provide.

[0006] The present inventor has previously proposed an image data compression device that can sufficiently improve encoding efficiency while employing a predictive method in image data compression using a pseudo-halftone reproduction image using a random dither method. is provided. (Patent Application No. 176330/1999)

[0007]

[Means for solving the problem] The invention corresponding to claim 1 is:
A pseudo-halftone processing unit that converts a multilevel image signal into a binary signal using a pseudo-halftone reproduction method, a line memory that sequentially stores the pseudo-halftone binary signal from the pseudo-halftone processing unit, and a signal state value. A prediction table in which a prediction signal is set at an address determined by the density coefficient value and one of the pseudo halftone binary signals from the pseudo halftone processing section are used as the target pixel signal, and the target pixel adjacent to or stored in the line memory is Divide a plurality of adjacent pseudo-halftone binary signals that have already been processed into multiple regions, calculate the signal state value of a part of each region and the density coefficient value of the remaining region, and calculate the signal state value and the density coefficient. A prediction processing unit that classifies the pixel of interest based on numerical relationships, determines and outputs a prediction signal of the pixel of interest using a prediction table, and compares the prediction signal from this prediction processing unit with the signal of the pixel of interest to determine whether the prediction is correct. , a prediction error generation section that outputs a prediction error signal indicating non-accuracy of prediction, and an encoding section that encodes the prediction error signal from this prediction error generation section.

[0008] The invention corresponding to claim 2 also includes a pseudo halftone processing unit that converts a multivalued image signal into a binary signal by a pseudo halftone reproduction method, and a pseudo halftone processing unit that converts a multilevel image signal into a binary signal by a pseudo halftone reproduction method,
A line memory that sequentially stores value signals and one of the pseudo-halftone binary signals from the pseudo-halftone processing section as a pixel signal of interest, and a plurality of already processed pixels adjacent to or close to the pixel of interest stored in the line memory. Divide the pseudo-halftone binary signal into multiple regions, calculate the signal state value of a part of each region and the density coefficient value of the remaining region, and select the pixel of interest from the relationship between the signal state value and the density coefficient value. A determination unit that performs classification and determines whether the pixel signal of interest has a signal value with a high probability of appearance based on the classified state, and performs an arithmetic operation based on the signal value determination by this determination unit to perform encoding processing. This system is equipped with an arithmetic encoding section that performs the following steps.

[0009]

[Operation] In the present invention having such a structure, the pseudo halftone binary signals from the pseudo halftone processing section are sequentially stored in the line memory. At this time, one of the pseudo halftone binary signals is set as a pixel signal of interest, and a plurality of already processed pseudo halftone binary signals adjacent to or close to the pixel of interest stored in the line memory are stored in the prediction processing unit. The pixel of interest is divided into regions, the signal state value of a part of each region and the density coefficient value of the remaining region are determined, and the pixel of interest is classified based on the relationship between the signal state value and the density coefficient value. Then, based on the classification of the pixel of interest, a prediction signal of the pixel of interest is determined and output using a prediction table. Then, the prediction error generation unit compares the prediction signal with the pixel signal of interest and outputs a prediction error signal indicating whether the prediction is accurate or not. The prediction error signal obtained in this way is encoded.

Further, in the present invention, the pixel of interest is classified based on the relationship between the signal state value and the density coefficient value, and based on the classified state, it is determined whether the pixel of interest signal has a signal value with a high probability of appearance. Then, arithmetic operations are performed based on this signal value determination to perform encoding processing.

[0011]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings.

As shown in FIG. 1, a multilevel image signal is input to a pseudo halftone processing section 1 that converts it into a binary signal using a pseudo halftone reproduction method such as an error diffusion method or a minimum average error method. It is converted into a pseudo-halftone binary signal.

The pseudo-halftone binary signals from the pseudo-halftone processing section 1 are sequentially stored in a line memory 2 and are input to one input terminal of an exclusive OR circuit 3 as a prediction error generating section. The plurality of pseudo halftone binary signals stored in the line memory 2 are each supplied to a prediction processing section 4.

The prediction processing unit 4 uses one of the pseudo halftone binary signals from the pseudo halftone processing unit 1 as a pixel signal of interest, and uses an existing pixel adjacent to or close to the pixel of interest stored in the line memory 2. Divide a plurality of processed pseudo-halftone binary signals into a plurality of regions, calculate the signal state value of a part of each region and the density coefficient value of the remaining region, and calculate the signal state value and the density coefficient. Classifies the pixel of interest based on numerical relationships,
The prediction table 5 is used to determine and output a prediction signal for the pixel of interest.

Specifically, as shown in FIG. 2, the region division is performed by dividing the pixel of interest * into four adjacent pseudo halftone binary signals 81 , 82 , 83 , 84 that have already been processed.
Eight pseudo-halftone binary signals 91, 92, 93, 9 which have already been processed and which form a region
4 , 95 , 96 , 97 , 98 in the second region
2 is formed.

The prediction table 5, as shown in FIG.
"0" or "1" is assigned to the address determined by the signal state value A=0 to 15 representing the signal state of the first region X1 as 24 and the density coefficient value B=0 to 8 corresponding to the second region X2. A predicted signal is set in advance. The prediction signal output from the prediction processing section 4 is input to the other input terminal of the exclusive OR circuit 3.

The exclusive OR circuit 3 compares the prediction signal from the prediction processing section 4 and the target pixel signal from the pseudo halftone processing section 1, and when both signals match, that is, "1" and " If the prediction error signal is "1" or "0" or "0", a prediction error signal "0" indicating a correct prediction is output, and if the two input signals do not match, that is, "1" or "0" or "0" is output. In the case of "1", a prediction error signal "1" indicating a non-accurate prediction is output. The prediction error signal from the exclusive OR circuit 3 is stored in a prediction error signal memory 6.

The prediction error signal stored in the prediction error signal memory 6 is read into an encoding section 7 and encoded. The encoding unit 7 employs, for example, a run-length encoding method.

In this embodiment having such a configuration, the input multi-valued image signal is converted into a pseudo-halftone binary signal by the pseudo-halftone reproduction method in the pseudo-halftone processing section 1. The pseudo halftone binary signals from the pseudo halftone processing section 1 are sequentially stored in the line memory 2 and also input to one input terminal of the exclusive OR circuit 3.

The prediction processing section 4 uses one of the pseudo halftone binary signals from the pseudo halftone processing section 1 as a pixel of interest signal, and uses the signal of the first region X1 adjacent to the pixel of interest * stored in the line memory 2. The signal state value and the density coefficient value of the adjacent second region
Based on the prediction table 5, a prediction signal of the pixel of interest is determined and output.

That is, the signal state values of the first region X1 are the pseudo halftone binary signals 81, 82, 83, 84.
is regarded as a binary code sequence, and the converted value is 0 for "0, 0, 0, 0" and 5 for "0, 1, 0, 1", and as a result, the signal state value The value of A is 0~
Obtain one of 15 values. Further, the density coefficient value of the second region X2 means each of the pseudo halftone binary signals 91, 92,
When the weighting coefficients of 93, 94, 95, 96, 97, and 98 are "a, b, c, d, e, f, g, h", the density coefficient value B is calculated using the formula, B=91 × a + 92 × b
+93 ×c+94 ×d+95 ×e+96 ×f+
97 × g + 98 × h, and the weighting coefficients “a, b, c, d, e, f, g, h” are changed to “1, 1, 1,
1, 1, 1, 1, 1'', the concentration coefficient value B is 0 to 8.
Get one of the values. In this way, the pixel of interest * is classified. Then, a corresponding prediction signal is determined from the prediction table 5 based on the state signal value A and the density coefficient value B. Note that the setting of the predicted signal in the prediction table 5 is performed by calculating the predicted signal value in advance using a training image or the like.

By performing such prediction processing, the prediction accuracy rate is improved, and as a result, the probability that the exclusive OR circuit 3 outputs a prediction error signal "0" indicating a prediction accuracy is increased. In other words, the length of consecutive 0's in the prediction error signal becomes longer, and when this prediction error signal is encoded using an encoding method such as run-length encoding, the encoding efficiency can be improved, and therefore, sufficient This allows data compression. Next, other embodiments of the present invention will be described with reference to the drawings. Note that the same parts as in the above embodiment are given the same reference numerals and detailed explanations will be omitted.

As shown in FIG.
A plurality of pseudo halftone binary signals stored in the pixel * are supplied to the determination unit 11, and the determination unit 11 sets one of the pseudo halftone binary signals as a pixel signal of interest and selects a first region adjacent to the pixel of interest *. X1 pseudo halftone binary signals 81, 82, 83
, 84 , and the pseudo halftone binary signal 91 , of the second region X2 close to the pixel of interest *.
92 , 93 , 94 , 95 , 96 , 97, 98
Find the density coefficient value from , classify the pixel of interest* based on the relationship between this signal state value and the density coefficient value, and determine whether the pixel of interest signal is a signal value with a high probability of appearance, that is, a dominant symbol, based on the classified state. It is determined whether the signal value has a low probability of appearance, that is, whether it is an inferior symbol. This determination uses state management information from the state management memory 12 that manages the past state of the pixel signal. The determination section 11 supplies the determination result to the arithmetic encoding section 13.

The arithmetic encoding unit 13 performs arithmetic operations to form a code based on the determination result from the determination unit 11, that is, the determined probability of occurrence of the symbol. This is an encoding method that shows a tendency for the encoded signal sequence to become shorter as the probability of appearance of an inferior symbol with a lower value decreases.

[0025] Thus, the signal state value obtained from each pseudo halftone binary signal 81, 82, 83, 84 in the first region X1 and each pseudo halftone 2 in the second region
Value signals 91, 92, 93, 94, 95, 96
, 97 , 98 , by classifying the pixel of interest * based on the relationship of density coefficient values determined by , 97 , 98 , the probability of appearance of an inferior symbol in each classified state tends to be lower, and sufficient data compression can be achieved. In this way, the same effects as in the previous embodiment can be obtained in this embodiment as well.

In the above embodiment, the pixel of interest is classified using the first area adjacent to the pixel of interest and the second area adjacent to the pixel of interest, but the classification is not limited to this. 2, nine pseudo-halftone binary signals 101, 102, 103, 104, 105 are shown by dotted lines.
, 106 , 107 , 108 , 109 is set, and the pixel of interest is determined from the signal state value of the first region and the density coefficient value of the third region without using the second region. Alternatively, the pixel of interest may be classified three-dimensionally using the first, second, and third regions, and in this case, the allocation of signal state values and density coefficient values is limited. It's not a thing. Furthermore, it goes without saying that the method of determining the area is not limited to the above embodiment.

[0027]

As described in detail above, according to the present invention, it is possible to sufficiently improve encoding efficiency in image data compression using a pseudo-halftone reproduction image by the random dither method, and It is possible to provide an image data compression device that can perform efficient compression.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing one embodiment of the present invention.

FIG. 2 is a diagram showing the relationship between a pixel of interest and each area in the same example.

FIG. 3 is a diagram showing the contents of a prediction table in the same embodiment.

FIG. 4 is a block diagram showing another embodiment of the present invention.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1... Pseudo halftone processing section, 2... Line memory, 3... Exclusive OR circuit (prediction error creation section), 4... Prediction processing section, 5...
Prediction table, 7... Encoding section, 11... Judgment section, 13... Arithmetic encoding section.

Claims (2)

[Claims] 1. A pseudo-halftone processing unit that converts a multilevel image signal into a binary signal using a pseudo-halftone reproduction method, and a line memory that sequentially stores the pseudo-halftone binary signal from the pseudo-halftone processing unit. , a prediction table in which a prediction signal is set at an address determined by a signal state value and a density coefficient value, and one of the pseudo halftone binary signals from the pseudo halftone processing section are set as a pixel signal of interest, and stored in the line memory. A plurality of already processed pseudo-halftone binary signals adjacent to or close to the pixel of interest to be processed are divided into a plurality of regions, and the signal state value of a part of each region and the density coefficient value of the remaining region are respectively determined, a prediction processing section that classifies the pixel of interest based on the relationship between the signal state value and the density coefficient value, determines and outputs a prediction signal of the pixel of interest using the prediction table; A prediction error generation unit that compares the pixel signal of interest and outputs a prediction error signal indicating whether the prediction is correct or incorrect, and an encoding unit that encodes the prediction error signal from this prediction error generation unit are provided. An image data compression device characterized by: 2. A pseudo-halftone processing unit that converts a multilevel image signal into a binary signal using a pseudo-halftone reproduction method, and a line memory that sequentially stores the pseudo-halftone binary signal from the pseudo-halftone processing unit. , one of the pseudo halftone binary signals from the pseudo halftone processing unit is taken as a pixel signal of interest, and a plurality of already processed pseudo halftone binary signals adjacent to or close to the pixel of interest stored in the line memory is divided into a plurality of regions, the signal state value of a part of each region and the density coefficient value of the remaining region are determined, the pixel of interest is classified based on the relationship between the signal state value and the density coefficient value, and the classified pixel is classified. a determination unit that determines whether the pixel signal of interest has a signal value with a high probability of appearance based on the state of An image data compression device comprising: