JPS641997B2 - - Google Patents

Description

[Detailed Description of the Invention] This invention relates to a black and white television signal (hereinafter referred to as a black and white television signal).
This invention relates to a predictive encoding device for television signals that compresses data for digital transmission of television signals (abbreviated as TV signals) at a low bit rate.

A conventional device of this type is shown in FIG. In the figure, 1 is an A/D converter that digitizes an analog monochrome TV signal, 2 is a DPCM encoder that predictively encodes the output of the A/D converter 1,
3 is a subtracter that calculates a prediction error; 4 is a quantizer that assigns the prediction error to a predetermined quantization level;
5 is an adder that adds the DPCM signal of the quantizer 4 and a predicted signal; 6 is a predictor that forms a predicted signal; 7 is a DPCM encoder that performs the inverse operation of the DPCM encoder 2;
A DPCM decoder for decoding, 8 is a D/A converter.

Next, the operation will be explained. analog black and white tv
Signal 101 at a predetermined sampling frequency _fS ,
As shown in Figure 2, samples are extracted in a square grid pattern on the screen and digitized, resulting in a digital signal series S.
(i, j) (i= 1 , 2, ... m and j = 1, 2, ...
n). In this case, the number of sampling points in one field is m×n pixels. This A/D
Black and white TV signal S digitized by converter 1
If (i, j) is transmitted by PCM, the data rate is too high, so the DPCM encoder 2 performs data compression processing using the correlation between neighboring pixels.
Here, 102 is a black and white TV signal S (i, j), 103
is the predicted signal P^(i, j), and 104 is the DPCM signal ε^
(i, j), 105, the reproduced TV signal S^(i, j), the prediction coefficients a ₁ and a ₂ , and the quantization noise q _N generated by the quantizer 4 to reduce the number of quantization levels. Then ε^(i,j)=S(i,j)−P^(i,j)+q _N S^(i,j)=ε^(i,j)+P^(i,j) P^( i, j)=a ₁・S^(i−1, j)=a ₂・S^(i−
DPCM signal 1 with the number of quantization levels reduced by +a ₂・S^(i, j−1)
04 can be formed. That is, the black and white TV signal 10
If the prediction error signal from the prediction signal 103 predicted from the neighboring pixel signals is quantized by taking advantage of the fact that 2 has a strong correlation between neighboring pixels, a DPCM signal 104 with a reduced number of quantization levels can be obtained, resulting in a low bit rate. Transmission becomes possible.

On the receiving side, the DPCM decoder 7 performs DPCM decoding on the transmitted DPCM signal 104 to decode the reproduced TV signal 105. That is, S^ (i, j) = ε^ (i, j) + P^ (i, j) P^ (i, j) = a ₁・S^ (i−1, j) + a ₂・S^ (i , j-1) to decode the reproduced TV signal 105. This is D/
The A converter 8 converts it into an analog signal and reproduces an analog black and white TV signal 101'.

In this case, the data compression ratio R is 10 for a black and white TV signal.
2 is K bits/sample, and the DPCM signal 104 is l bit/sample, then R = l/K, and the average distortion (squared error power) in one field is
D becomes D=1/m×n _n 〓 ⁱ⁼¹ _o 〓 ^j=1 {S(i, j)−S^(i, j)} ² .

Since the conventional predictive coding device is configured as described above, it is possible to
It requires DPCM encoding and decoding processing, requiring high-speed arithmetic elements, and despite this, it has the disadvantage that the image quality of the reproduced TV signal is poor considering the data compression rate.

Furthermore, as described in Japanese Patent Application Laid-Open No. 128081/1981, a device has been considered that reduces the amount of transmitted data by half and interpolates missing pixels using adjacent pixels. Since the control signal for interpolation is transmitted to the receiver (receiving side), the amount of information increases, and the disadvantage is that sufficient data compression and high-speed signal processing cannot be achieved.

This invention was made in order to eliminate the drawbacks of the conventional ones as described above, and after performing phase inversion sampling for each line of a black and white TV signal, hexagonal prediction is performed.
By DPCM encoding and transmission, and after DPCM decoding on the receiving side, a selective interpolation calculator interpolates and reproduces the monochrome TV signal, resulting in a high data compression rate, halving the data compression processing speed, and high quality image transmission. It is an object of the present invention to provide a predictive coding apparatus for television signals that realizes the following.

An embodiment of the present invention will be described below with reference to the drawings. In Fig. 3, 1 is an A/D converter that digitizes an analog black-and-white TV signal, and 9 is an A/D converter that converts the digitized black-and-white TV signal into a digital image by inverting the phase of the sample pulse for each horizontal scanning line. A phase inversion sampler 3 subsamples and extracts pixels on a square lattice, and a phase inversion sampler 3 calculates the difference between a subsample black and white TV signal (hereinafter referred to as subsample TV signal) 106 which is the output of the phase inversion sampler 9 and a prediction signal described later. 4 is a quantizer that assigns the output of the subtracter 3 to a predetermined quantization level; 5 is a hexagonal prediction which is the output of the quantizer 4;
The adder 10 which adds the DPCM signal and the predicted signal is a hexagonal predictor which forms the predicted signal from the reproduced sub-sampled TV signal reproduced by the adder 5.

Further, the subtracter 3, quantizer 4, adder 5, and hexagonal predictor 10 constitute a hexagonal predictive DPCM encoder for transmitting the hexagonal predictive DPCM signal 108.

In addition, in FIG. 4, 5 is an adder that adds the transmitted hexagonal predicted DPCM signal and the predicted signal, and 1
0 is the reproduced subsample which is the output of the adder 5
These are hexagonal predictors that form predictive signals from TV signals, and these constitute a hexagonal predictive DPCM decoder that decodes the hexagonal predictive DPCM signal 108 into a reproduced sub-sampled TV signal 109. 11 is a line buffer memory that stores the reproduced sub-sample TV signal sequence for two lines; 12 is a delay element that delays the reproduced sub-sample TV signal by one sub-sample; 13 is the adder 5, the line buffer memory 11 and The four reproduced sub-sampled TV signals output simultaneously by the delay element 12 (i.e.
Interpolation for the missing pixel from signals for the four vertically and horizontally adjacent pixels of the missing pixel due to subsampling in an orthorhombic grid on the screen
a selective interpolation calculator for forming a TV signal; 20 is a reproduction subsample of the line buffer memory 11;
TV signal output and interpolation by the selective interpolation calculator 13
An interpolation switch 8 alternately selects the TV signal output to form an interpolated reproduced TV signal, and 8 is a D/A converter that converts the output of the interpolation switch 14 into an analog signal.

Further, FIG. 6 shows an embodiment of the selective interpolation calculator 13 in the receiving side device of FIG. 4. In the figure, 1
4 is a horizontal interpolation calculator that calculates the horizontal average value and horizontal difference absolute value between two pixels adjacent in the horizontal direction of the reproduced sub-sampled TV signal; 15 is a horizontal interpolation calculator 2 adjacent in the vertical direction of the reproduced sub-sampled TV signal; 3 is a subtracter, 5 is an adder, 16 is a divider that halves the input signal, 17 is absolute value conversion 18 is a comparator that compares the absolute value of the horizontal difference and the absolute value of the vertical difference and detects the direction in which the change is smaller; 19 is the change in the horizontal or vertical direction determined by the comparator 18; This is a select switch that selects the average value of the direction component with the smaller (difference absolute value) and outputs either the horizontal average value or the vertical average value as an interpolated TV signal for missing pixels.

Next, the operation of the television signal predictive coding apparatus according to the present invention will be explained. Analog black and white TV signal 101 at first sampling frequency f _S
sample, A/D convert, and digital black and white.
When the TV signal 102 is extracted, x
(i, j) (i=1, 2,...m, and j=1, 2,
...n) is obtained. Furthermore, among x(i, j),
When this is sampled by inverting the phase of the sample pulse of the second sampling frequency f _S /2 line by line, the sub-sample shown by the white circle in Figure 5 is obtained.
A TV signal x(i,j) (i odd and j odd or i even and j even) is obtained. The pixels indicated by black circles here need to be interpolated and reproduced on the receiving side. In the above, the number of sampled pixels in one field is reduced to 1/2 of that in the first sampling stage.

Next, the hexagonal predictive DPCM encoder shown in FIG. 3 performs data compression on the sub-sampled TV signal 106 obtained through the phase inversion sampler 9 by the following calculation. In this case, as is clear from FIG. 5, the prediction points suitable for forming the prediction signal 107 for the sub-sampled TV signal x(i, j) are x(i-2, j), x( They are three pixels in the hexagonal direction of i-1, j-1) and x(i+1, j-1). Therefore, 106 is the sub-sampled TV signal x
(i, j), 107 as predicted signal y^(i, j), 108
is the hexagonal predicted DPCM signal D^(i, j), prediction coefficient a ₁ ,
Play a ₂ , a ₃ , 109 sub-sampled TV signal x^
(i, j) and quantization noise q _N , then D^ (i, j) = x (i, j) - y^ (i, j) + q _N x^ (i, j) = D^ (i , j)+y^(i, j) y^(i, j)=a ₁・x^(i−2, j) +a ₂・x^(i−1, j−1) +a ₃・x^( i+1, j-1). Here subsample the number of quantization levels
Assuming that the TV signal 106 is 2K bits/sample and the hexagonal predicted DPCM signal 108 is 1 bit/sample, the data compression rate (compression ratio R) per sample is R=1/2K.

Next, the TV signal decoding process on the receiving side shown in FIG. 4 will be explained. On the receiving side, an adder 5 and a hexagonal predictor 10 perform hexagonal prediction on the transmitted hexagonal prediction DPCM signal 108.
In the DPCM _{decoding loop}・x^(i-1, j-1) +a ₃・x^(i+1, j-1)
Decode 9. Furthermore, this reproduced subsample signal 1
09 is a line buffer memory 11 for two lines.
and one subsample's worth of delay element 12, the reproduced subsample TV signal 109 is input to The reproduced sub-sample TV signal x^(i-1, j-1) 110, which corresponds to the two lines before the top,
x^(i+1,j-1)111 and x^(i,j-2)
112 signals are output simultaneously.

Four reproduced sub-sampled TV signals 1 including the input of the line buffer memory 11 obtained in this way
Signals 09, 110, 111 and 112 are sent to the selective interpolation calculator 13 and used to interpolate and reproduce the signal x^(i, j-1) that is missing due to the sub-sampling. In addition, the reproduced subsample TV signal 111
is simultaneously sent as is to the interpolation switch 20, and is alternately selected with the interpolated TV signal 113 formed by the selective interpolation calculator 13, and the interpolated reproduced TV signal x^(i, j-1)(i =1, 2, 3,
...m and j=1, 2, ...n). The interpolated reproduced TV signal 114 reproduced at the rate of this apparent sampling frequency _fs is D/A converted to obtain a reproduced analog TV signal 115.

Next, the operation of the selective interpolation calculator 13 in the TV signal reproduction process will be explained with reference to FIG. Here, the horizontal interpolation calculator 14 and the vertical interpolation calculator 15 calculate a horizontal average value 116 and a vertical average value 118 between adjacent pixels that sandwich the missing pixel. That is, the following signals are calculated from the reproduced sub-sampled TV signals 110 and 111 horizontally adjacent to the missing pixel x^(i, j-1) and the reproduced sub-sampled TV signals 109 and 112 vertically adjacent.

Horizontal average value 116 x^ _h (i, j-1) = 1/2{x
^(i-1, j-1)+x^(i+1, j-1)} Vertical average value 118 x^ _v (i, j-1) = 1/2{x
^(i, j-2) + x^(i, j))} Horizontal difference absolute value 117 z^ _h (i, j-1) = |x^
(i-1, j-1)-x^ (i+1, j-1) | Vertical difference absolute value 119 z^ _v (i, j-1) = |x^
(i, j-2)-x^(i, j)| These are calculated by the adder 5, the divider 16, the subtracter 3, and the absolute value converter 17. Furthermore, using the comparator 18 and the select switch 19, interpolation
The TV signal 113 is calculated as follows.

x^ (i, j-1) = x^ _h (i, j-1)...z^ _h (i, j-
1) z^ _v (i, j-1) x^ _v (i, j-1)...z^ _h (i, j-1)>z^ _v (i, j
-1) In other words, when there is little change in the horizontally adjacent TV signal, the horizontally average value is 116, and the vertically adjacent TV signal is
When there is little change in the signal, the vertical average value 118 is adopted as the interpolated TV signal 113. This eliminates jaggedness at changing points in the horizontal or vertical direction of the reproduced image caused by subsampling in an orthorhombic grid pattern on the screen.

As described above, according to the present invention, even if high-efficiency encoding with a data compression ratio R = 1/2 + l/K is performed by line-by-line phase inversion sampling and hexagonal predictive DPCM encoding, the selective interpolation By adopting this method, a predictive coding device for monochrome TV signals with little image deterioration can be constructed.

As described above, the present invention is configured to employ line-by-line phase inversion sampling, hexagonal predictive DPCM, and selective interpolation filters, so it can obtain a high data compression effect for black-and-white TV signals, and the signal processing is quick and the image quality deteriorates. Therefore, it is possible to realize a predictive coding apparatus for television signals with a small amount of noise.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a conventional predictive encoding device for black-and-white TV signals, FIG. 2 is an explanatory diagram showing the arrangement of sampled pixels on the screen in the predictive encoding device, and FIG. 3 is an illustration of one of the present invention. Black and white TV according to the example
FIG. 4 is a block diagram of a signal predictive coding transmitter, FIG. 4 is a block diagram of a black and white TV signal predictive coding receiver according to an embodiment of the present invention, and FIG. 5 is a monochrome TV signal predictive coding device according to the present invention. FIG. 6 is an explanatory diagram showing the arrangement of sample extraction pixels and interpolation pixels in FIG. In the figure, 1...A/D converter, 8...D/A converter,
9... Phase inversion sampler, 10... Hexagonal predictor, 11
... Line buffer memory, 12 ... Delay element, 13
...Selective interpolation calculator, 14...Horizontal interpolation calculator, 1
5...Vertical interpolation calculator, 16...Divider, 17...Absolute value converter, 18...Comparator, 19...Select switch, 20...Interpolation switch, 106...Subsample monochrome television signal, 108...Hexagonal prediction
DPCM signal, 109... Reproduction subsample television signal, 113... Interpolated television signal, 1
14...Interpolated reproduction television signal, 116...Horizontal average value, 118...Vertical average value, 117...
Horizontal difference absolute value, 119... Vertical difference absolute value. Note that the same or corresponding parts in the figures are indicated by the same reference numerals.

Claims (1)

[Claims] 1. A sub-sampled black and white television signal is extracted by sampling the black and white television signal in an orthorhombic grid pattern on the television screen by inverting the phase of the sampling pulse for each horizontal scanning line. a phase inversion sampler that performs the sub-sampled black-and-white television signal; and a hexagonal hexagon whose prediction reference points are two pixels one line before the diagonally upper right and upper left of a predetermined pixel of the subsampled black and white television signal, and the previous pixel on the same line. Hexagonal prediction using a predictor
a DPCM encoder; a hexagonal predictive DPCM decoder that decodes a reproduced subsample television signal from a hexagonal predictive DPCM signal that is DPCM-encoded and transmitted by the hexagonal predictive DPCM encoder; and the reproduced subsample. Regarding the missing pixels at the midpoints of the orthorhombic lattice on the screen of the television signal,
Calculate the horizontal average value and vertical average value between each adjacent pixel sandwiching the missing pixel, and compare the horizontal difference absolute value and the vertical difference absolute value to determine the horizontal average value of the smaller directional component. or a selective interpolation calculator that interpolates the vertical average value as an interpolated television signal for the missing pixel, and alternately extracts the reproduced sub-sampled television signal and the interpolated television signal to create an apparently
A predictive coding device for a television signal, comprising: an interpolation switch that generates an interpolated reproduction television signal at double the rate;