JPH0244819A - Prediction encoder - Google Patents

Abstract

PURPOSE:To set the division of a '0' code in the run length code of a prediction result at minimum by adding the identification code of a predictor to the prediction result, adding the identification code concerned only when the predictor has changed and adding '0' except for said case. CONSTITUTION:When the number of the predictors 101A, 101B... to be integrated is increased, several bits of the identification codes 108 are required. When the identification codes are added in the forms as they are, the prediction result, namely, the '0' code of the run length code is divided, and that an encoding efficiency is deteriorated comes into question. Thus, the identification code concerned 108 is added only when the predictors 101A and 101B... have changed, and '0' is added except for said case. Thus, the division of the '0' code in the run length code can be held at minimum and the encoding efficiency as a whole can be improved.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of Industrial Application The present invention relates to a predictive encoding device for image information that uses an adaptive prediction method.

BACKGROUND OF THE INVENTION Recently, high-efficiency encoding methods for binary images using adaptive prediction methods have come to be used as a core technology for high-speed facsimile in fields such as newspaper transmission, which require high code compression rates. Ta. This adaptive element 2... sub-encoding system is described, for example, in ``Dot Signal Processing Apparatus'' of Japanese Patent Laid-Open No. 56-27565.

Adaptive predictive coding uses multiple pixels surrounding the coded pixel as reference pixels, predicts the value (color) of the coded pixel from the value of this reference pixel (a combination of black and white colors), and calculates the prediction result. This is run-length encoding.

FIG. 3 shows a circuit block diagram of an adaptive predictive coding method using a conventional predictor. Signals read by the original reading device 1 through main scanning and sub-scanning are sampled and quantized by the binarizer 2 and become digital image signals.

This is temporarily stored in memory 3. The 2 to 12 reference pixels read out from the memory 3 are used by the predictor 4 to predict the value of the encoded pixel. The prediction result from the predictor 4 is run-length encoded in the encoder 5.

In the case of a paper surface such as a newspaper page in which text parts and halftone photographs of various mesh numbers are mixed, a plurality of predictors 4 having reference pixels suitable for the properties of each image are used to improve encoding efficiency. Individually, they are 3 /=-.

Prediction processing is performed by adaptively switching in units of a fixed number of pixels,
Techniques for encoding the prediction results are widely used.

Problems to be Solved by the Invention However, when prediction is performed in units of a fixed number of pixels using multiple predictors, it is necessary to add an identification signal indicating which predictor has been selected. As a method of adding, a method is known in which the prediction result is added to a code after run-length encoding. However, in the above-mentioned method, if the effect of adaptively switching the predictor by reducing the unit of a fixed number of pixels is achieved, the code amount of the identification code increases and the coding efficiency decreases.

In view of the above problems, the present invention makes it possible to improve encoding efficiency.

Means for Solving the Problems The present invention provides encoding means for adding and encoding an identification code indicating which predictor has been selected among a plurality of predictors to a prediction result, and only when the predictor changes. The above object is achieved by providing a code adding means for adding a corresponding identification code.

Effect of the Invention With the above configuration, the present invention adds an identification code indicating which predictor has been selected from a plurality of predictors, thereby making the identification code a run-length code as a prediction result.

Here, the run-length encoder uses the continuation II of the prediction result.
It is the total length of the number of OIT codes and one Par1'' code, and the longer the It OI + code continues, the more the number of codes after being encoded (for example, the output of the encoder described in the conventional example). is reduced, and encoding efficiency is improved.

However, as the number of predictors to be incorporated increases, the identification code requires several bits, and if it is added as is, the prediction result, that is, the run-length code II OI + code, will be divided, resulting in a decrease in coding efficiency. There's a problem. To avoid this, the corresponding identification code is added only when the predictor changes, and “O” is added at other times, thereby minimizing the division of the °′0″′ code of the run-length code and improving the overall The aim is to improve the encoding efficiency.

5.-7 Example Hereinafter, a method for adding an identification code to the results of an example of the present invention and its contents will be described with reference to the drawings.

In FIG. 1, an input pixel signal 100 is input to a predictor 101.
A, 101B, 101C, and 101D make predictions and convert them into prediction results, and each prediction result is sent to the minimum prediction error selection circuit 102 and the buffer memo! J103A
, 103B, 103C, and 103D.

The minimum prediction error selection circuit 102 counts the prediction errors of each prediction result and selects a predictor 101 with a small prediction error.
Identification code 1 for selecting prediction results from A to 101 D
08 is output to the data selector 104 and the additional circuit 105. At this time, the identification code 108 is added by referring to the predictor selection addition pattern of the next light. For example, if predictor 1 is selected in pixel unit t-1 and predictor 2 is selected in pixel unit t, the identification code is "11".
is added, and if 6・-7 predictor 1 is selected, "oo" is added as an identification code.
" is added. That is, the code 1" appears in the identification code only when the selection of the predictor changes.

Next, the data selector 104 stores the prediction result selected by the identification code 108 in the buffer memo! J1037
...-2 Read from A to 103D and output to the additional circuit 105. The addition circuit 105 adds an identification code 108 to the selected prediction result, and adds the prediction result 10 in units of the fixed number of pixels.
9 to the encoder 106. The prediction result 109 is encoded by the encoder 106 and output as a code output 107.

An example of the code format of the prediction result 109 in this case is shown in FIG. Here, the prediction results 20OA and 200B indicate the prediction results in units of the constant number of pixels. In the prediction result 200A, the identification code 201°A is 'oo', indicating that the predictor is not switched.

In other words, this indicates that the predictor used for each pixel before the prediction result 200A is used again.

In the prediction result 201A, the identification code 201B is "1"
1'', indicating that the predictor is switched.At this time, if predictor 3 was used in prediction result 20OA, predictor 4 was used in prediction result 201A as shown in the previous article. Indicates that it will be used.

In addition, since four predictors are used in this embodiment, the identification codes are "00", "01", "NO" and "1".
It is expressed by a bit pattern of 1''.It goes without saying that a similar configuration is also possible when the number of predictors is five or more in this embodiment.

Effects of the Invention As described above, the present invention adds the identification code of the predictor to the prediction result, adds the corresponding identification code only when the predictor changes, and adds "o" otherwise. Since division of the tt-on code of the run-length code of the prediction result can be minimized, the overall coding efficiency is greatly improved.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a predictive encoding device according to an embodiment of the present invention, FIG. 2 is a diagram showing prediction results by the same device, and FIG. 3 is a block diagram of a conventional predictive encoding device. 101A-101D...Predictor, 102...Minimum prediction error selection circuit, 103A-103D...Panza memory, 104...Data selector, 105...Additional circuit, 106...Encoder. Diagram

Claims (1)

[Claims] A plurality of predictors that predict the value of the predicted pixel in units of a certain number of pixels from a combination pattern of values of reference pixels existing around the predicted pixel, and an identification code indicating whether the predictor is selected are used in the prediction. A predictive encoding device comprising: encoding means for adding and encoding a result; and code addition means for adding a corresponding identification code only when the predictor changes.