JPS612483A - Encoding transmission device of picture information - Google Patents

Abstract

PURPOSE:To improve quantizing accuracy and to attain highly accurate transmission by quantizing the difference of an encoder on the basis of quantizing characteristics corresponding to a forecasted value. CONSTITUTION:A picture element signal memory 21, a forecasting unit 22, an encoder 23, and a restoration value former 24 are formed in a transmitter. Similarly to the conventional method, a forecasted value for a real value X of an objective picture element is calculated by the forecasting unit 22. The forecasted value and the real value X are inputted to the encoder 23 and encoded with the corresponding relation shown in A and B. The forecaseted value and the real value X are simultaneously inputted also to the restoration value former 24 and the restoration value is formed with the corresponding relation shown in the Fig. C. The restoration value is stored in the memory 21 to be used for the forecasting of the succeeding picture elements. Thus, the quantizing characteristics of the difference can be changed by the forecased value.

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field of the Invention] The present invention relates to an encoding and transmitting device for image information.

[Prior art]

Diff is a conventionally known device of this type.
erential Pulse Code Module
ation (D P CM) method. This DPC
An example of the basic configuration of the M method is shown in FIG. In the figure,
12 is a transmitting side pixel signal memory, and 12 is a predictor that outputs a predicted value of the pixel of interest based on known sample values. When the surrounding pixels are A, B, C, and D, the predicted value is calculated using the following calculation method.

5? = −(A+D) ...(1) However, A, D
indicates the values of the pixels A and D above.

The above equation is a commonly used equation in calculating a predicted value, but of course it may be changed to 裳-A for simplicity. 13 is the input signal X (true value) and the predictor 12
14 is a quantizer/encoder, and the quantization characteristic of this quantizer/encoder 14 is as shown in Fig. 3. Often used. In this example, the number of levels is 16 (0-15). When the prediction difference signal (X-2) is subjected to 3-bit constant length encoding, it is common to provide the quantization characteristics shown in FIG. 3 as described above. Examples of code words are also shown in FIG. 3. In your example, a 3-bit fixed-length codeword is considered to be suitable for online transmission, but if a buffer memory or the like is provided, non-length encoding is also possible.

Furthermore, the representative value shown in FIG. 3 is used as a representative value at the time of decoding on the receiving side. Create a signal value that will be restored with
It is written to 1 and used for predicting the following pixels.

Next, the receiving side of the conventional system will be explained. 17 is a decoder that outputs a representative value of the difference value <x-2> from the received code word;
19 is a receiving side pixel signal memory, 16 is a transmitting side predictor 1
The predictor 18 performs the same operation as 2, and the adder 18 sums the predicted value and the representative difference value and outputs a restored signal.The value output from the adder 18 is stored in the pixel signal memory 19. It is input and used for restoring pixel signals later.

In addition, in the figure, 5 is a transmission path.

This conventional method has the following drawbacks. That is, in the conventional method, since quantization is performed only by the difference, some code words are wasted depending on the value of the predicted value. For example, when the values of pixels A and D are both O, the input value is 2.
becomes father=0 according to the above Tl1 formula. Therefore, the value of the difference (X-5?) that is the output of the subtracter 13 will never be negative, no matter what the value of the pixel X is. That is, the difference <X-S
Code word 010 whose representative values of <> are -1, -2, -6,
100.111 cannot be transmitted, and as you can see from the predicted value 0 column in Figure 4, in reality,
This means that it is quantized to the actual value.

Similarly, when father=15, the opposite code word 001,011°101.11 when (X-father) is positive
1 is wasted, and in effect, X is quantized into four values. FIG. 4 shows restored values for combinations of predicted values and true values using such a conventional method. Here, for those whose restoration value is less than 0, O
In addition, if the number is 15 or more, it is set as 15.

As described above, in the conventional method, quantization is performed by focusing only on the difference between the pixel signal of interest and the predicted value, so depending on the value of the predicted value, the actual quantization accuracy may be coarse, making highly accurate transmission impossible. It had the disadvantage of being unburnable.

[Summary of the invention]

This invention was made to eliminate the drawbacks of the conventional ones, and improves quantization accuracy by performing differential quantization in an encoder with quantization characteristics according to predicted values. It is an object of the present invention to provide an encoding and transmitting device for image information that can perform transmission with higher precision.

[Embodiments of the invention]

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

FIG. 5 is a configuration diagram of an encoding and transmitting apparatus according to an embodiment of the present invention, in which 21 is a transmitting side pixel signal memory;
22 is a predictor, each of which is the same as in the conventional device. 23 is the predicted value (4 bits) and the true value of the pixel of interest (
An example of the contents is shown in FIG. 6. here,
Since we are considering the 3-bit fixed-length codeword mentioned above, 0 to 7
Figure 7 shows the correspondence between the numbers and codewords. Note that the correspondence between such numbers and code words can also be changed depending on each predicted value.

Furthermore, as a code word, it is of course possible to introduce a non-fixed length code word.

Reference numeral 24 denotes a restored value creator, which creates a restored value on the receiving side by combining the predicted value from the predictor 22 and the true value X of the pixel of interest. It can be easily constructed using a 4-bit, 28x4 continuous memory.

Another configuration method of the restored value generator 24 is to input the code word (3 bits) which is the output of the encoder 23 and the predicted value (4 bits) of the pixel of interest which is the output of the predictor 22. In this case, since there are input cuff bits and 4 output bits, it is configured with a 27×4 continuous memory. FIG. 8 is a correspondence table of restored pixel signals for combinations of predicted values and true values of the pixel of interest,
Here, the contents of the memory dedicated for successive use when the restoration value generator 24 is configured using the former configuration method are shown. and the pixel signal memory 21. Predictor 22, encoder 23
．． and a restoration value generator 24 are provided in the transmitter.

Next, the receiver side will be explained. 26 is a predictor having the same configuration as the predictor 22 provided in the transmitter, and 27 is this predictor 2.
This decoder outputs a restored value that is quantized using the predicted value from 6 and the transmitted code word, and this decoder 27
is easily constructed with input cuff bits and 4 output bits, that is, a 22×4 read-only memory. Furthermore, this is exactly the same when considering the latter method of the two methods of configuring the restoration value generator 24 described above.

FIG. 9 shows the contents of the memory when the decoder 27 is configured with a read-only memory, including predicted values and code word numbers (
Restoration values (O-15) are shown for combinations O-7). Further, 28 is a pixel signal memory that stores restored pixel signals output from the decoder 27.

Next, the operation will be explained.

First, as in the conventional case, for the pixel of interest X, its surrounding pixels A
, D, the predicted value father is calculated by the predictor 22.

Then, this predicted value father and the true value X of the pixel of interest are sent to the encoder 23.
The data is input into the code word and encoded using the correspondence shown in FIGS. 6 and 7, and a 3-bit code word is output. At the same time, the predicted value 2 and the true value X of the pixel of interest are also input to the restoration value generator 24.
In this step, restoration values are created with the correspondence relationship shown in FIG. The restored value is then stored in the pixel signal memory 21 and used for the following pixel prediction.

On the receiving side, the reference pixel is output from the pixel signal memory 28 to the predictor 26, and the predictor 26 calculates a predicted value in the same manner as in the prior art. Then, this predicted value and the transmitted code word are input to the decoder 27, and the decoder 2
7, the pixel signal is restored with the correspondence shown in FIG. Further, this restored pixel signal is stored in the pixel signal memory 28.

In the device of this embodiment, the quantization characteristic of the difference is changed depending on the predicted value, and wasteful code word assignment as in the prior art does not occur for any predicted value. Therefore, as can be seen by comparing the reconstructed signal of this embodiment shown in FIG. 8 with the conventional signal shown in FIG. 4, the improvement in accuracy on the receiving side in this embodiment is extremely large. .

Furthermore, the device configuration is not so complicated as can be easily seen by comparing FIG. 1 and FIG. 5.

In the above embodiment, an example of 16-level, constant-length 3-bit encoding was used, but it goes without saying that the number of levels is arbitrary and the assignment of code words is not necessarily of constant length. .

〔Effect of the invention〕

As described above, according to the present invention, in the image information encoding and transmitting apparatus, the quantization characteristic in the encoder is made to be a characteristic according to the predicted value, and quantization and encoding are performed using this characteristic. , it is possible to improve the accuracy of quantization, and there is no needless code word assignment for any predicted value as in the conventional method, which has the effect of enabling highly accurate image transmission.

[Brief explanation of drawings]

Fig. 1 is a configuration diagram of a conventional image information encoding and transmitting device, Fig. 2 is a pixel arrangement diagram, Fig. 3 is a diagram showing representative values of difference values and corresponding code words in the conventional method, and Fig. 4 is a diagram showing the restored value for the combination of predicted value and true value in the conventional method,
FIG. 5 is a block diagram of an image information encoding and transmitting device according to an embodiment of the present invention, FIG. 6 is a diagram showing code word numbers for combinations of predicted values and true values in the device, and FIG. FIG. 8 is a diagram showing the correspondence between code word numbers and code words. FIG. 8 is a diagram showing restored signal values for combinations of predicted values and true values in this device. FIG. 9 is a diagram showing correspondences between predicted values and code word numbers in this device. FIG. 3 is a diagram showing restored signal values. 21... Transmission side pixel signal memory, 22... Transmission side predictor, 23... Encoder, 24... Restoration value creator, 2
6... Receiving side predictor, 27... Restorer, 28...
Receiving side pixel signal memory.

Claims (1)

[Claims] (1) A transmitting side predictor that outputs a predicted value of a pixel of interest by referring to known pixel values stored in a transmitting side pixel signal memory;
A restoration value generator that obtains a restoration value that is quantized on the receiving side based on the predicted value and the true value of the pixel of interest and supplies this to the pixel signal memory on the transmission side; A transmitter includes an encoder that performs quantization with a quantization characteristic according to the predicted value based on the true value and outputs a predetermined code word, and a known pixel signal stored in the receiving side pixel signal memory. A receiving side predictor outputs a predicted value of the pixel of interest by referring to the pixel value, and obtains a quantized restoration value based on this predicted value and the transmitted code word, and transmits this to the receiving side pixel. 1. An apparatus for encoding and transmitting image information, comprising: a receiver having a decoder for providing a signal to a memory.