JPS59143485A - Encoding method of ntsc composite color signal - Google Patents

Abstract

PURPOSE:To convert a forecasting order based upon the statistical quantity of a picture and to improve data compressibility by executing encoding for forecasting order conversion only when the difference between a standard reference level and a reference picture element selected from a picture element is included in a set up level range. CONSTITUTION:The level of a picture element transmitted immediately before a transmitted remarkable picture element is compensated and operated by the level of a previously transmitted picture element to obtain a standard reference level. A combining picture element is selected from already transmitted picture elements as a reference picture element. When the level difference between the reference picture element level and the standard reference level is always included in a previously set up level range, only the level difference of each reference picture element from the standard reference level is referred. The order by which the level difference between the remarkable picture element and the standard reference level apears is forecasted, the level difference between a really formed remarkable picture element and the standard reference level is converted into the forecasted order and the forecasted order is encoded. If the level difference exceeds the set up range, different encoding is performed to improve the data compressibility.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a redundancy suppression coding method that can significantly reduce data errors required for transmission or storage of NTSC merged color signals.

Conventionally, D P CM (Di
ffer-e'Xtial P CM ) method.

Now let us consider the case where an NTSC decoupled composite signal is sampled at three times the color subcarrier frequency (fsc) using this method.

FIG. 1 shows the pixel arrangement in this case, and in the figure,
XO is the pixel of interest that is about to be transmitted,
The solid line indicates that the pixels X] to X6 above this have already been transmitted. Further, arrows M and 5 indicate the main scanning direction and the sub-scanning direction, respectively. Figure 2 is NTSC
' Shows the phantom relationship between the inter-pixel distance of the signal and the correlation coefficient. As can be seen from the figure, the correlation is high at pixel positions where three pixels touch.

Therefore, in the DPCM encoding method, the pixel XI three pixels before the current π+i element Xo to be transmitted in FIG. 1 is considered as the predicted value, and the difference value between the two (this is called the prediction error) E ==Xo −Xt is calculated, and this difference E is encoded and transmitted. Naturally, the shape of the distribution of the difference values E shows a sieve shape with 0 as the apex, and therefore data can be compressed by assigning short code words to the difference values in the vicinity of O. can.

However, the above-mentioned PCM encoding method cannot be said to sufficiently take into account the detailed statistical properties of an image, and has the disadvantage that encoding efficiency may decrease in some cases.

For example, if the correlation in the sub-scanning direction is larger than the correlation in the main-scanning direction, the prediction error value will be smaller if the pixels x2 to x4 in FIG. 1, which have the same color phase relationship as the pixel of interest Xo, are referred to. It is thought that efficiency will improve. In addition, pixel X5, which is located closest to the pixel of interest XO but has a different color phase and therefore has a small correlation, can also increase its correlation with the pixel of interest XO by performing color phase correction calculations using other pixels. Therefore, it is considered that efficiency can be improved by also referring to this pixel Xs.

The first invention of the present application was made based on the above-mentioned idea with the aim of eliminating the drawbacks of the conventional ones and realizing more efficient encoding. In this method, we focus on the fact that the distribution width of the level of the reference pixel selected in consideration of the color phase is often within a relatively small level range. The value obtained by correcting the color phase component using the level of the pixel obtained is set as the standard reference level, and a certain combination of pixels is selected as the reference pixel from among the pixels that have already been transmitted, and the level of the reference pixel and the standard reference level is determined. When all the differences are within a preset level range, the level difference between the pixel of interest and the standard reference level is determined in advance based on the image statistics by referring only to the level difference of each reference pixel from the standard reference level. By converting and encoding into the created appearance prediction order, and performing other encoding different from the above when it is not within the above level range,
It is an object of the present invention to provide a method for encoding an NTSC composite color signal that can reduce prediction errors and improve data compression rate. .

Next, the second invention of the present application is that the image number has parts where the level changes greatly, such as the contour part, and parts where the level changes little, such as the flat part, and the statistical properties of the two parts are different. Therefore, by distinguishing between the two and performing encoding appropriate to the statistical properties of each, the data compression rate is further increased. Prepare multiple combinations in advance, determine the optimal prediction ranking for each prepared combination, and determine which of the prepared combinations the reference pixel belongs to based on the level difference from the standard reference level of the actually generated reference pixel. It is an object of the present invention to provide a method for encoding an N T S C composite color signal, which can further improve the compression rate by determining the combinations to which they belong and performing predictive transform encoding as described above.

An embodiment of the second invention of the present application will be described below with reference to the drawings.

First, from among the pixels in FIG. 1, select the pixel X5 immediately before the pixel of interest XO that is about to be transmitted, and
A color phase correction calculation is performed using the pixels X1 and X6 as shown in equation (1), and this XS is determined as the standard reference level.

Xs = Xs + (Xt - XS) [
11 The standard reference level X5 as a result of this calculation is the pixel of interest X.

There will be a strong correlation with Next, as a reference pixel candidate, a pixel XI that has the same phase relationship as the pixel of interest Xo
, X2, XS and x4. Then, the differences between these reference pixels X1 to X4 and the standard reference level X5 are calculated, and are designated as d1 to d4. That is, ds = Xl - XS d2 == X2 - XS da = XS - In the determined mode, prediction rank encoding is performed.

Now, this process will be explained in detail below. First, the total level is 6 bits, that is, 0 to 63, and six combinations of level width ranges and reference pixels are prepared as shown in the table below. The determination of which of these six modes it belongs to is made as follows. That is, first, the level difference d
x, d2. When ds and d4 are all greater than or equal to -2 and less than or equal to +1, mode 1 is determined. Otherwise, the level difference d1. d2. and d4 are all greater than or equal to -4 and less than or equal to +3,
If this condition is satisfied, mode 2 is determined. If not, it is determined whether the level differences d2 and d4 are greater than or equal to -8 and less than or equal to +7, and mode 3
Decide whether or not. If it is not mode 3, it is determined to which of modes 4 to 6 it belongs by determining whether or not the level differences d2 and d4 are within the range of level widths shown in the table. Then, prediction rank conversion is performed in the mode determined by the above procedure, and encoding is performed. For example, to explain the case of mode 2, in each state determined by the level differences d+, dz, and 'd4 (the total number of states is 83-512), the level difference dO is calculated based on the statistics of the image. 7. Predict the rank to be used, and when the level difference dO that actually appears is the i-th predicted rank, convert this level difference dO to this predicted rank value i, and encode this predicted +1 [rank]. . At this time, if you assign a short code word to a word with a low predicted rank and a long code to a word with a high predicted rank, you can actually control the probability that a word with a low predicted rank will appear. Therefore, data compression encoding becomes possible.

Also, the predicted j ranking ranges from 1 to 64, but since the probability that a large predicted ranking will appear is low, it is not necessary to strictly rank the predicted rankings, and usually the fleet corresponding to the level width of the mode to which it belongs is Even if variable-length encoding is performed by ranking the level difference do within the range, and a fixed-length code is used when the level difference do is within that range, the coding efficiency will not be improved. Most of the deterioration is considered to be hot. Furthermore, when modes 4 to 6 are selected, since the levels of the leading pixels differ greatly, it is considered difficult to accurately predict the level difference do. Therefore, even if prediction is performed focusing only on the upper few bits of the level difference, the encoding efficiency does not decrease, so prediction can be simplified. For example, in the case of mode 5, the level differences d2 and d4 are -32 to +3, respectively.
1, and this can be expressed with 6 bits, but by focusing only on the top 4 bits, we determine the state (total number of states is 162 = 256), and calculate the level difference dO for each state. The order of the upper 4 bits of the data is predicted and encoded. The remaining lower bits of the level difference do can be transmitted as they are without being encoded as additional bits. By adopting such a method, the number of states to be identified can be reduced, and the number of storage units required for encoding and decoding can be reduced.

The above has described an embodiment of the second invention of the present application, or as an embodiment of the first invention of the present application, in the above embodiment, for example, the reference pixel is X2
, X4, the level range of the level difference between the reference pixel and the reference reference level is -8 to +7 (corresponding to mode 3), and the level difference d2. It is conceivable that when all d4 fall within this level range, transcoding is performed on the predicted 1jji order &', and when not, other conventionally known coding is performed. In this case, the predictive order position conversion coding is based on the level difference dz.

Each state determined by d4 (total number of states is 162 = 2
! ; 6), the order in which the level difference dO appears is predicted in advance based on the image statistics, and when the level difference dO that actually appears is the i-th predicted order,
The level difference dO is converted into a prediction order position, and this prediction order is encoded.

As described above, according to the present invention, the direct f of the pixel of interest is
The n-element CD level of A is corrected using the already transmitted pixel level and set as the standard reference level, and the already transmitted i:TQ
j Only when the level difference between the reference pixel selected from gO and the standard reference pixel is within a preset level range, predictive "IN'N displacement encoding is performed, or the combination of the reference pixel and the level range is performed. Since a plurality of combinations A are used and prediction rank conversion coding is performed for each combination, the optimal prediction rank 1111 conversion based on the statistics of image 1p is performed in rows f and ge, which is similar to the conventional DP CM method. It has the effect of obtaining a higher data pressure rate of 44.4 than the other methods.

Figure 2 is a diagram showing the arrangement of the pixel of interest and the transmitted pixels when sampled at three times the number of pixels, and Figure 2 is a diagram showing the relationship between the inter-pixel distance and the correlation coefficient in the N'rS C composite color signal. .

XO: Pixel of interest, Xl-X6: Pixel that has been transmitted.

Distance = - Procedural amendment (voluntary) +1,4J115M, 771411 To1; Correction 11 Secretary-General 1゜Remarks f'-5 Application No. 58-17699 No. 2 Invention Name of NTS CPH combined color signal encoding method; 3. Representative of the person making the amendment Fragment 111 A: 8 Part 4, Agent 5, Column 6 for detailed explanation of the invention in the specification subject to amendment, Contents of the amendment (1) "Level" on page 9, line 3 of the specification "difference do" is replaced by "level difference d between the pixel of interest Xo and the standard reference level X5. (=
XO-X5) Correct to J.

(2) Correct "within range" in line 6 of page 10 to 1 out of range - 1.

that's all

Claims (1)

[Claims] (A redundancy suppression coding method for a signal obtained by sampling an 11NTSC composite color signal at an integral multiple of the color subcarrier frequency, which is a method for encoding a signal with redundancy reduction, which is the level of a pixel transmitted immediately before a pixel of interest to be transmitted. The value obtained by performing a correction calculation on the level of the already transmitted pixels is set as the standard reference level, and a certain combined pixel from among the already transmitted pixels is selected as the reference rlIi! element 1
-1 When the level difference between the reference pixel and the base thread reference level is within a preset level range, the target pixel is determined by referring only to the level difference of each reference pixel from the standard reference level. A prediction that predicts the order in which a level difference with the above reference reference level will appear, converts the level difference between the pixel of interest that actually occurs and the above reference reference level into a higher ranking, and encodes this predicted order. When rank conversion encoding is performed and all of the above level differences are not within the preset level range, "-" pre-shaping i1E+i (3'
7. A method for encoding an NTSC composite color signal, characterized by performing encoding different from P14 encoding. +21 NTSC composite color signal is converted to an integer multiple of the carrier frequency (redundancy suppression coding method for a unified signal, in which the level of the pixel that was transmitted immediately before the pixel of interest that is about to be transmitted is already Transmitted 1ttii element 1
Correction performance at 7 bells! A, (7 is based on the C reference level, and the level range 9J of the difference between the reference pixel selected from the already transmitted pixels C and the reference pixel and the above standard reference level)
), and for each prepared combination, only the level difference between each reference pixel and the standard reference level is referred to, and the order in which the level difference between the pixel of interest and the standard reference level appears is determined. is predicted, and based on the level difference between the reference pixel that actually occurs and the above standard reference level, it is determined which of the above ■4 combinations it belongs to, and the pixel of interest that actually occurs in the combination to which it belongs and the above standard are determined. A method for encoding NTSC composite RA codes, characterized in that a level difference from a reference level is converted into the prediction rank, and this prediction rank is encoded.