JPH0380784A - Encoder for television signal - Google Patents

Abstract

PURPOSE:To improve prediction efficiency and to prevent an S/N lowered by performing the thinning of a digital television signal at every second picture element and with field offset, and performing prediction encoding by using the optimum predicted value. CONSTITUTION:The digital television signal is converted to data with high correlation by eliminating a high frequency component from the signal with a three-dimensional filter 1. Image data is thinned after compressing the data to the half at a sub sampling circuit 2, and applying the field offset. As prediction data, the optimum one out of the prediction data P1, P2, and P3 in an in-field prediction circuit 8, an inter-frame prediction circuit 9, and an adder 10 is selected at an adaptive selection circuit 11. Then, mean prediction is used which takes average between in-field prediction and interframe prediction instead of the inter-field prediction. Therefore, such processing hits prediction more accurately than the inter-field prediction depending on the quality of an image, which heightens the prediction efficiency and to improve the S/N as a result.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a television signal encoding device using predictive encoding.

[Summary of the Invention] The present invention provides a sub-sampling circuit that thins out a digital television signal every other pixel and by field offset, and an average prediction that takes the average value of intra-field prediction and inter-frame prediction. , and a predictive encoding circuit that performs predictive encoding using the optimal predicted value of these predictions, visual resolution is maintained and inter-field prediction is intermediate between intra-field prediction and inter-frame prediction. Focusing on the fact that it has the same characteristics, we use average prediction, which takes the average value of intra-field prediction and inter-frame prediction, instead of inter-field prediction, together with intra-field prediction and inter-frame prediction, which improves prediction efficiency and does not reduce the signal-to-noise ratio. .

[Prior art] Predictive coding, orthogonal transformation,
Although vector quantization is available, predictive coding is often used in fields where a relatively high bit rate and high image quality are required. When applying this predictive encoding, the predictive encoding is generally performed after pixels are thinned out by subsampling. The above-mentioned subsampling thins out the image data every other pixel, and this thinning is performed with an offset of one pixel between the first field and the second field. This field offset subsampling is visually effective because the resolution in the horizontal and vertical directions does not drop, especially when the image is stationary.

[Problems to be Solved by the Invention] However, in field offset subsampling, the pixel position of one field before is not directly to the side or directly above, but is diagonally, so the correlation between fields is poor. Therefore, when an adaptive prediction method including inter-field prediction is configured, the prediction efficiency does not increase and the S/N ratio deteriorates.

SUMMARY OF THE INVENTION Accordingly, an object of the present invention is to provide a television signal encoding apparatus that employs field offset subsampling to maintain visual resolution and improve prediction efficiency, so that the signal-to-noise ratio does not deteriorate.

[Means for Solving the Problems] A television signal encoding device according to the present invention for achieving the above object encodes a digital television signal every other pixel,
In addition, a sub-sampling circuit thins out the first and second fields by offsetting one pixel, and the correlation between the output television signal of this sub-sampling circuit is used to perform intra-field prediction, inter-frame prediction, and average prediction of both. The system is equipped with a predictive encoding circuit that selects the optimal prediction among these three predictions, and performs predictive encoding using the optimal predicted value.

[Effect] Since the subsampling circuit thins out image data by field offset, the visual resolution is relatively unchanged, and instead of using inter-field prediction, which has poor prediction efficiency on average, it uses intra-field prediction and inter-frame prediction. Instead of inter-field prediction, which has intermediate properties, average prediction, which takes the average of intra-field prediction and inter-frame prediction, is used together with intra-field prediction and inter-frame prediction, which improves overall prediction efficiency.

[Example] Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 shows a circuit block diagram of a television signal encoding device. In FIG. 1, an NTSC television signal is converted into an 8-bit digital signal, and this digital television signal is input to a three-dimensional filter l. The three-dimensional filter l cuts the high frequency components of the television signal and limits the pass band. By cutting high frequency components, the correlation of the signals is increased and aliasing caused by the sub-sampling circuit 2 described below is suppressed. The output of the three-dimensional filter l is input to the sub-sampling circuit 2.

The sub-sampling circuit 2 thins out the television signal every other pixel, and offsets by one pixel between the first field and the second field. That is, the pixel arrangement before and after subsampling changes as shown in Figure 2(a) to Figure 2(b), and the band characteristics before and after subsampling change as shown in Figure 3(a) to Figure 3(b). Because of this, the resolution in the horizontal direction (g) and vertical direction (d) does not drop much. The television signal thinned out by this subsampling circuit 2 is sent to a predictive encoding circuit 3.
guided by.

The predictive encoding circuit 3 predictively encodes a television signal using an appropriate prediction function, and is configured as follows. That is, the subtracter 4 receives the television signal output from the subsampling circuit 2 and the optimal prediction data output from the adaptive selection circuit 11, and then inputs the image data (actual value) of the television signal.
The optimum prediction data is subtracted from the prediction error data and the prediction error data is output to the quantizer 5. The quantizer 45 converts the 8-bit prediction error data into 5-bit data whose bit rate is lower than the human input value, and outputs the data to the decoder 6. The decoder 6 converts the 5-bit data back into 8-bit data, and this prediction error data is led to the adder 7. Optimum prediction data from the adaptive selection circuit 11 is also led to the adder 7, and by adding the prediction error data and the optimal prediction data, it is returned to substantially actual image data. The output of this adder 7 is led to an intra-field prediction circuit 8 and an inter-frame prediction circuit 9, respectively. The intra-field prediction circuit 8 calculates prediction data P1 using pixel data immediately next to and directly above the nearest pixel.

Specifically, when z-1 is expressed by Z-transformation as a single operator, the following equation is calculated: P, = ±.z-1+upper.Z-H (H: horizontal line) 2. The interframe prediction circuit 9 calculates prediction data P for pixels within a certain area from the motion vector of pixel data of the previous frame, and when expressed by Z transformation, P
t-Z-F-M (p, field.

M: motion compensation) is calculated. The outputs of the intra-field prediction circuit 8 and the inter-frame prediction circuit 9 are each input to an adder 10, and the average sum is added by the adder IO.

That is, this adder IO is configured as an average prediction means that calculates prediction data P3 by calculating d(P++Pt). Further, the output PI of the intra-field prediction circuit 8, the output P of the inter-frame prediction circuit 9, and the output P3 of the adder 10 are each led to an adaptive selection circuit 11.

The adaptive selection circuit 11 selects the prediction that is optimal for the inner pixel of the three-way prediction, that is, the prediction with the smallest prediction residual, and outputs prediction data using this prediction function. That is, this predictive encoding circuit 3 is configured so that the quantizer 5 is placed in a field pack loop so that the quantization noise generated in the quantizer 5 is not accumulated in the decoding device, and the output of the quantizer 5 has a variable length. It is input to the encoding circuit 12.

The variable-length encoding circuit 12 performs variable-length encoding in which short codes are assigned in the order of the highest ratio by utilizing the fact that the data ratio of predictive encoding exhibits a mountain-like characteristic. This variable length encoded image data is transmitted with an error correction code added thereto.

In the above configuration, the three-dimensional filter 1 removes high frequency components from the digital television signal and converts it into highly correlated data, and the subsampling circuit 2 compresses the data in half. Since the sub-sampling circuit 2 thins out the image data by field offset, the visual resolution is maintained relatively without deterioration.

Further, the image data thinned out by the subsampling circuit 2 is led to a subtracter 4, where prediction residual data is obtained by subtracting the prediction data from the image data, and this prediction residual data is sent to a quantizer 5. The data is then predictively encoded. The prediction data is the prediction data pl, p2 . The optimal one among P3 is selected by the adaptive selection circuit 11. Instead of using inter-field prediction, which has poor prediction efficiency on average, we focused on the fact that inter-field prediction has an intermediate property between intra-field prediction and inter-frame prediction, and instead of using intra-field prediction and inter-frame prediction. Because it uses average prediction, which takes the average of the predictions, it is more accurate than inter-field prediction depending on the nature of the image.

Therefore, intra-field prediction, inter-frame prediction, and average prediction increase overall prediction efficiency, resulting in SN
The ratio will also improve.

FIG. 4 shows a circuit block diagram of the decoding apparatus for the encoded television signal. In FIG. 4, the decoding device is configured to operate in substantially the opposite manner as the encoding device. That is, the variable length decoding circuit 20 uses 5 pixels per pixel.
The data is converted into bit predictive residual data, and this data is decoded into 8-bit image data by the predictive decoding circuit 2I.

The predictive decoding circuit 21 includes a decoder 22, an adder 23, an intra-field prediction circuit 24, an inter-frame prediction circuit 25, an adder 26, and an adaptive selection circuit 27.
It is composed of parts with the same characteristics as that of Part B. The output of the predictive decoding circuit 21 is led to a data interpolation circuit 28, where, for example, pixel data identical to adjacent data is inserted into the thinned out pixel points, and then passed through a three-dimensional filter 29 and output.

[Effects of the Invention] As described above, according to the present invention, there is provided a sub-sampling circuit that thins out a Dinotal television signal every other pixel and by field offset, and an intra-field prediction, an inter-frame prediction, and an average value of both. and the average prediction that takes
Since it is equipped with a predictive encoding circuit that performs predictive encoding using the optimal predicted value among these predictions, visual resolution is maintained by field offset subsampling, and prediction becomes worse due to field offset subsampling. Instead of inter-field prediction, we include average prediction that takes the average value of intra-field prediction and inter-frame prediction, increasing overall prediction efficiency.
This has the effect of not reducing the N ratio.

[Brief explanation of drawings]

1 to 4 show embodiments of the present invention, FIG. 1 is a circuit block diagram of a television signal encoding device, and FIG. 2(a)
is a diagram showing the pixel arrangement before subsampling, Figure 2 (b
) is a diagram showing the pixel arrangement after subsampling, and Figure 3 (
Figure 3(a) is a diagram showing the band characteristics before subsampling, Figure 3(b) is a diagram showing the band characteristics after subsampling, and Figure 4 shows the band characteristics before subsampling.
The figure is a circuit block diagram of a television signal decoding device. 2... Subsampling circuit, 3... Prediction code north circuit. External name】51 field-0-0-10-10--()-Q-12 field-
0-0-0-0-0--○--Name 2 field--〇--1-〇--τ--0-42 Field Affirmation b) Sub-bun all linkage f) National policy diagram 3 showing 0 companies in the imperial area (
b)

Claims (1)

[Claims] (1) Digital television signal every other pixel, and
In the first and second fields, intra-field prediction, inter-frame prediction, and average prediction of both are performed using a sub-sampling circuit that thins out by offsetting one pixel, and the correlation between the output TV signal of this sub-sampling circuit. , and a predictive encoding circuit that selects the optimal prediction among these three predictions and performs predictive encoding using the optimal predicted value.