JPS62234423A - Intra-frame predictive coding device - Google Patents

Abstract

PURPOSE:To produce a signal suited to the intra-frame predictive coding operation by preparing plural quantization characteristics and switching these characteristics via the histories of quantizing results of the points at the same position of the immediately preceding frame to prevent the edge business of the intra-frame prediction CONSTITUTION:The TV signals image picked up by a camera 1 are digitized by an A/D converter 2 with a normal PCM. When a certain point is quantized, the quantizing signals of the points at the same position of the immediately preceding frame are checked and switched to the quantization characteristics so that those checked quantizing signals are easily selected. In this respect, the nonlinear quantizing signal is delayed by a frame memory 10 by an amount equal to a frame. Then a switch 11 is changed over according to said delay output to select the outputs of two quantizing circuits 5-l and 5-2 of different characteristics. In such a way, the quantization is secured to the value equal to the quantizing signal preceding by a frame despite many types of variance as long as the pictures are kept static. Thus the edge business is prevented.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention is an intraframe predictive coding device for digitally converting a television signal, etc., and compressing redundancy therein to reduce transmission speed. The present invention relates to a high-efficiency encoding device.

[Conventional technology]

As is generally well known, the operation of the high-efficiency encoding device is as follows. That is, an image signal obtained by scanning like a television signal is digitized, the next signal is predicted from the signal that has already been transmitted, and its child 8 (!I error is nonlinearly quantized at nonuniform intervals. encoded and transmitted.The receiving side.

The original image signal is restored by the reverse operation.

Here, prediction using only the signal in the same frame as the already transmitted signal is called intra-frame prediction, and prediction using the signal of the immediately previous frame is called inter-frame prediction.

The former is a device of m t+1-, and when a transmission error occurs, the influence can be quickly resolved, and the latter has the advantage of greatly increasing the transmission speed. A method of combining these advantages was described at the 1185 National Conference of the Institute of Electronics and Communication Engineers (1985). That is, this is an attempt to construct a network in which the total sum of the device cost and the transmission cost as described in 1) is minimized by connecting both devices in cascade.

[Problem that the invention seeks to solve]

However, the conventional known device lacks consideration of the following points9: Intra-frame prediction 1111 causes a large prediction ♂ (difference) at the contour of the image. When quantizing such a large prediction error, it is coarsely quantized.For this reason, the image signal fluctuates slightly and the quantized signal fluctuates greatly, so the output signal also fluctuates greatly between frames.This phenomenon is commonly used in edge business. Even if the original image is still, the outline appears to be moving, which is one of the causes of image quality deterioration.

Furthermore, a signal in which this phenomenon occurs is not suitable for interframe predictive coding. Because this encoding is
The correlation between frames is calculated, and areas with small differences are assumed to be stationary and are not transmitted at all, while only areas with large differences are transmitted, resulting in significant compression. However, when the above phenomenon occurs, there is a difference between frames even in stationary parts, so the data must be transmitted, and significant compression becomes impossible.

An object of the present invention is to prevent edge business in intraframe prediction and to generate a signal suitable for interframe predictive coding.

[Means for solving problems]

The present invention achieves the object by preparing at least two types of quantization characteristics and switching the quantization characteristics using a history such as quantization results of points at the same position of the immediately preceding frame 11.

[Effect]

When quantizing a certain point, the quantization signal of the point at the same position in the immediately previous frame is checked, and the quantization characteristic is switched to one that makes it easier to select that quantization signal.

Then, when the image is stationary, even if there is some variation, it is quantized to the same value as the quantized signal of one frame before, thereby preventing edge business.

Also, when one image is moving, the quantization characteristic is switched to a value that tends to be the same as the quantized signal of one frame before, but since it is almost the same as the normal quantization characteristic, the quantization value is also almost the same. .

Therefore, the required quantization value can be obtained whether the image is stationary or moving. Furthermore, since there is no need to determine whether the image is still or not, the apparatus can be simplified.

〔Example〕

An embodiment of the present invention will be described below with reference to FIG. 1. In the figure, the portion surrounded by dotted lines relates to the present invention, and the other portions are the same as conventionally known high-efficiency encoding devices.

The TV double signal imaged by the camera 1 is digitized by the analog/digital converter 2 using ordinary PGM.

The prediction circuit 3 generates a predicted value, and the subtraction circuit 4 calculates the difference (prediction error). The prediction error is nonlinearly quantized in a quantization circuit 5, encoded into a redundantly compressed code signal in an encoding circuit 6, and then sent to a transmission path 7.

At the same time, the above-mentioned nonlinear quantized signal is added to the prediction signal by the addition circuit 8, and is input to the prediction circuit 3, where it is used for subsequent prediction value generation.

Next, the main parts of the present invention will be explained. The above nonlinear quantized signal is delayed by one frame by the frame memory 10, and the switch 11 is changed according to the output thereof to select the outputs of the two quantization circuits 5-1 and 5-2 having different characteristics.

An example of the characteristics of the quantization circuit 5 will be explained with reference to FIG. That is, the quantization characteristic 1 has a wide range in which the quantization signal is 4 or 1:3, and conversely, the quantization characteristic 2 has a wide range in which the quantization signal is 1 or 8.

Therefore, when the quantization signal of the previous frame is 4 or J3, the quantization characteristic] is adopted, and in other cases, the quantization characteristic 2 is adopted. Note that a circuit 12 for determining the output level of the frame memory 11 is provided on the output side of the frame memory 10.

Note that the following modifications are also within the scope of the present invention.

(1) In this embodiment, the quantization characteristics were switched using the quantization signal of one frame before, but the quantization signal of any number of frames before may be used as necessary.

(2) In addition to the normal prediction within the frame 21 for each point, the present invention may be applied to a case where a plurality of pixels are grouped together into a block and a signal that is subjected to orthogonal transformation or the like is quantized.

(3) The encoded signal may be sent to a normal transmission path, or may be stored in any storage device.

(4) The quantization characteristics prepared in advance are also
It is not limited to the street, but may be the opposite street. Alternatively, by using one quantization circuit and adding and inserting information from one frame before between the subtraction circuit 4 and the quantization circuit 5 or after the analog-to-digital conversion circuit 2 to give a bias, it is possible to substantially Another possible method is to use multiple quantization characteristics.

Modification (2) is particularly suitable for a color TV signal in which the color signal is modulated with subcarriers and superimposed on the brightness signal. This will be explained using FIG. That is, since the phase of the subcarriers is reversed every frame as shown in the figure, the phase difference between the signals between frames is small even when the image is stationary. Therefore, it is difficult to apply the above embodiment.

If, for example, four points of the signal in the figure are orthogonally transformed using the four coefficients shown in FIG. 4, the sum of the four points is obtained (a)
The term and the term (c), which is the sum of two pixels in the vertical direction, almost cancel out the phase difference of the subcarriers, so they have a large correlation between frames.

Therefore, it is sufficient to quantize only these two terms with a history between frames.

Note that in this case, edge business is generated by the remaining two terms, so relatively long code words may be assigned to these two terms and finely quantized.

〔Effect of the invention〕

As explained above, according to the present invention, edge business can be prevented, so not only can image quality be improved when only intra-frame prediction is applied, but also prediction of inter-frame prediction when inter-frame prediction is cascaded to this output. It has the effect of improving efficiency.

[Brief explanation of drawings]

FIG. 1 is a block diagram of an embodiment of the present invention, FIG. 2 is an example of the characteristics of the quantization circuit shown in FIG. 1, and FIG. phase,
FIG. 4 shows coefficients for orthogonally transforming the signal in FIG. 3. DESCRIPTION OF SYMBOLS 1...TV camera, 2...Analog/digital converter, 3...Prediction circuit, 4...Subtraction circuit, 5...
Quantization circuit, 6... Encoding circuit, 7... Transmission line, 8
...Addition circuit, 10...Frame memory, 11...
·switch. τ\ Not 1 Figure 100 2 ■

Claims (1)

[Claims] 1. A quantization circuit having at least two types of quantization characteristics, a device for storing a value determined from a quantized signal in a transmitted frame, and means for switching the above-mentioned quantization characteristics according to this value. An intraframe predictive coding device characterized by the following.