JPS6212214A - Nonlinear quantized mode switching device in variable length coding system - Google Patents

Abstract

PURPOSE:To obtain a device changing smoothly the picture quality according to the storage quantity of a buffer memory while avoiding the large sized device by using the result of comparison between a forecast error of an adjacent picture element and a predetermined threshold value in addition to the storage quantity of the buffer memory so as to control the circuit. CONSTITUTION:A picture signal read from a camera 1 is converted into a digital signal by an analog/digital converter 2 and the difference with a forecast value is calculated by a subtraction circuit 3. Quantization and coding are applied to the forecast error by a nonlinear quantization circuit 4 and one of them is selected by a switch 5. The selected code word is given to a transmission line 7 via a buffer memory 6. A representing value of adjacent picture elements, e.g., just before is retarded by one picture element by a delay circuit 15, an optional conversion is applied by a conversion circuit 16 and the result is compared with the storage quantity in place of a constant input to a comparator circuit 12.

Description

[Detailed Description of the Invention] [Field of Application of the Invention] The present invention relates to a variable length coding method, which is one of high efficiency coding methods for transmitting image signals etc. in a short time, and particularly relates to a variable length coding method, which is one of high efficiency coding methods for transmitting image signals etc. in a short time. The present invention relates to a switching device suitable for improving image quality by facilitating mode switching.

[Background of the invention]

As is well-known, variable-length coding is a method of predicting signals such as images and non-linearly quantizing the prediction error. It takes advantage of the fact that there is a bias in the frequency of occurrence of signals, and uses short code words for frequently occurring signals. , a long codeword is assigned to a low signal.

As a result, compared to fixed-length coding, which assigns codewords of a constant length regardless of frequency, the same nonlinear quantized signal can be transmitted using codewords that are shorter on average, which has the advantage of shortening transmission time. .

It is also known that a buffer memory is required to implement this. This is because although the codeword length is variable, the speed of the transmission path is constant, so it is necessary to match the speed difference between the two.

Now, when the above-mentioned signal is generated according to a predetermined frequency, the input/output speed of the above-mentioned buffer memory is balanced, so the amount of storage increases and the buffer memory does not overflow. However, when a large number of signals assigned long code words occur, the input speed increases and the buffer memory overflows.

To prevent this, normally, when the amount of storage in the buffer memory exceeds a certain threshold, the nonlinear quantization mode is switched and long code words are prohibited. As the types of code words decrease, the types of nonlinear quantized signals that can be transmitted also decrease.
There is a problem that image quality deteriorates. This is called overload.

On the other hand, if a large number of short code words occur, the buffer memory becomes empty, and in this case too, short code words are prohibited when the amount of storage in the buffer memory falls below a certain threshold. However, in this case, since the types of nonlinear quantized signals do not decrease, the image quality does not deteriorate, and there is no problem.

Therefore, deterioration of image quality during overload is a problem. In particular, in the past, the mode was directly switched from the normal mode to the nonlinear quantization mode during overload, which caused variations in image quality within a screen or between screens, causing the deterioration to become even more noticeable.

As a method to prevent this, (1) a method in which the prediction error is multiplied by a constant determined from the amount of storage in the buffer memory described above, thereby gradually reducing the frequency of occurrence of long code words as the amount of storage increases (Research and practical application by the Telephone and Telecommunications Public Corporation). Report Vol. 32, No. 9, pp. 273 (1983)), (2) A method of controlling the frequency of occurrence of long code words by setting multiple types of nonlinear quantization circuits and switching them according to the amount of storage in the buffer memory [ CCITT COM, XV-&D 1
, p p 29) is known.

These methods are desirable in terms of image quality because the image quality gradually deteriorates as the load increases from normal to overload. However, the former requires a multiplication circuit on the transmitting side and an inverse function circuit on the receiving side.

The latter also requires a number of circuits corresponding to the types of nonlinear quantization, and has the problem of increasing the size of the device.

[Purpose of the invention]

An object of the present invention is to provide an apparatus that smoothly changes image quality according to the amount of storage in a buffer memory while preventing the apparatus from increasing in size as described above.

[Summary of the invention]

To achieve the above purpose, due to the statistical nature of the image signal,
This method takes advantage of the fact that the prediction errors of pixels in the vicinity of a pixel with a large prediction error are often also large, and due to the characteristics of visual perception, even if this pixel is quantized by a nonlinear quantization circuit during overload, the deterioration in image quality is not noticeable. do.

That is, the present invention is characterized in that the switching of the nonlinear quantization mode is controlled using the result of comparing the prediction error of neighboring pixels with a predetermined threshold value, in addition to the conventional amount of storage M in the buffer memory. .

[Embodiments of the invention]

An embodiment of the present invention will be described below with reference to FIG. (a) of the figure is an example of switching according to the conventional method, and (b) is one according to the present invention. In the present invention, it can be seen that as the storage amount increases, the probability that the nonlinear quantization mode during overload will be selected increases, and the image quality gradually deteriorates.

FIG. 2 illustrates the configuration and operation of the present invention.

In the figure, the part surrounded by a dashed line is a feature of the present invention,
Others are the same as before. First, the conventional part will be explained.

That is, the image signal read from camera 1 is an analog
The digital converter 2 converts the signal into a digital signal, and the subtraction circuit 3 calculates the difference from the predicted value (prediction error). The prediction error is quantized and encoded by a plurality of nonlinear quantization circuits 4, and one of them is selected by a switch 5. The selected code word is sent to the transmission line 7 via the buffer memory 6.

At the same time, the code word is decoded by a plurality of decoding circuits 8, and representative values are determined for each code word. The representative value is selected by the switch 9 which is switched in synchronization with the switch 5, and the adder circuit 1
0 is added to the predicted value. The addition result is sent to the prediction circuit 11
is used for subsequent predicted value generation.

Next, a conventional nonlinear quantization mode switching method will be explained. In order to align the transmitting and receiving sides, the rli product in the buffer memory 6 is sampled by a sampling circuit 12 at a constant period such as a scanning line or screen, and is transmitted through a switch 13. The sample signal is compared with a constant value by a comparison circuit 14, and the nonlinear quantization mode is switched based on the result.

In contrast, in the present invention, the representative value of the neighboring pixel, for example, the immediately preceding pixel, is delayed by one pixel by the delay circuit 15, and the conversion circuit 15
6, and compares it with the accumulated amount instead of the constant input of the comparison circuit 12. As a result, (b
) is clearly possible.

Note that the operation on the receiving side follows the reverse process, but the principle is the same, so the explanation will be omitted.

〔Effect of the invention〕

As described above, the present invention is highly effective in that it is possible to smoothly switch the nonlinear quantization mode with the addition of a small number of devices, thereby reducing deterioration in image quality during overload.

[Brief explanation of the drawing]

FIG. 1 is an example of switching the nonlinear quantization mode according to the present invention, and FIG. 2 is a block diagram of a circuit for implementing this. 1...Camera, 2...Analog/digital converter. 3... Subtraction circuit, 4... Nonlinear quantization circuit, 5, 9
゜13...Switch, 6...Buffer memory, 7.
...Transmission line, 8...Decoding circuit, 10...Addition circuit,
11... Prediction circuit, 12... Sampling circuit, 14...
- Comparison circuit, 15... Delay circuit, 16... Conversion circuit.

Claims (1)

[Claims] A prediction error signal obtained by subtracting a prediction signal from an image signal, etc. is quantized and encoded using several nonlinear quantization modes,
In a variable length coding method in which smoothing is performed using a buffer memory before transmission, the above-mentioned nonlinear quantization mode is switched using the storage amount of the buffer memory and the prediction error of neighboring pixels. Nonlinear quantization mode switching device.