JPH02274080A - Image encoding transmission system - Google Patents

Abstract

PURPOSE:To shorten the processing time required for encoding by deriving a difference or exclusive OR of video signals of one frame before and the present frame, allocating a Huffman code, based on the appearance probability of the same information pattern and transmitting encoded information. CONSTITUTION:When a difference or exclusive OR of an image signal of one frame before and an image signal of the present frame is derived by an arithmetic means 5, since there is a correlation between the frames of the image signals, an information pattern to a picture element which scarcely moves is concentrated in the vicinity of '0'. With regard to an information pattern to all picture elements obtained in such a way, a statistical processor 6 derives the number of times of appearance of the same information pattern. By the number of times of appearance, the appearance probability of each information pattern is calculated, and in the descending order of the appearance probability, a Huffman code is allocated to that of a prescribed number determined in advance of the information pattern and encoded information is transmitted. In such a way, the processing time can be shortened by a simple constitution.

Description

[Detailed Description of the Invention] The present invention relates to an image encoding transmission method with a high compression rate for encoding and efficiently transmitting images, and particularly for image transmission used in videophones, video conferences, etc. The present invention relates to an image coding transmission method suitable for.

Conventionally, typical methods for encoding images include orthogonal transform encoding, vector quantization, predictive encoding, and the like. Orthogonal transform encoding methods include Hadamard transform and cosine transform, all of which require a lot of arithmetic processing, vector quantization methods that require repeated searches during encoding, and predictive encoding methods that This method predicts the next pixel from a pixel and encodes the difference between it and the actual pixel, but it requires a considerable amount of calculation processing.

The problem that Akira is trying to solve All of the conventional image encoding methods mentioned above require a huge amount of calculation, which requires a dedicated arithmetic unit, making the configuration complex and expensive, and the processing time required for encoding is long. There was a problem.

An object of the present invention is to provide an image coding transmission system that solves the above problems.

In order to achieve the above-mentioned object, the present invention includes a storage means for storing digital image signals for one frame, and an arithmetic means for calculating the difference or exclusive OR between the image signals of the previous frame and the current frame. , a statistical processing means that performs statistical processing on the information patterns for all pixels from the calculation means and calculates the appearance probability of the same information pattern, and a descending order of the appearance probability (larger The present invention is provided with an encoder that allocates and encodes a predetermined number of the information patterns by Huffman codes in a predetermined number of the information patterns, and a transmission means that transmits the information encoded by the encoder.

Therefore, when calculating the difference or exclusive OR between the image signal of the previous frame and the image signal of the current frame using the calculation means,
Since there is a correlation between frames of image signals, information patterns for pixels with little movement are concentrated near O. The number of occurrences of the same information pattern is calculated for the information patterns for all pixels obtained in this way, and the probability of appearance of each information pattern is calculated from this number of occurrences. Huffman codes are assigned to numbers and encoded information is transmitted.

"Leaving" L± FIG. 1 is a bronc diagram of one embodiment of the present invention.

The present invention will be described in detail below with reference to FIG.

In the figure, 1 is an image input terminal that receives image signals from cameras, VTRs, etc., and 2 is an analog/digital terminal (hereinafter referred to as "A/DJ") that converts the analog image signal led to the image input terminal 1 into a digital image signal. ) converter, 3 is a current frame memory that stores the image signal of the current one frame, 4 is a previous frame memory that stores the image signal of the previous frame, and 5 is the image stored in the current frame memory 3 and the previous frame memory 4. an arithmetic device that calculates the difference between signals;
6 is a statistical processing device that obtains the appearance probability of the information pattern of each pixel calculated by the arithmetic device 5, and derives a predetermined number of the above information patterns in descending order of the appearance probability; 7 is the statistical processing device; An encoder that assigns a Huffman code to a predetermined information pattern with a high probability of appearance derived from the processing device; 8 a memory for storing the information encoded by the encoder together with the information pattern; 9 an encoder for storing the information encoded by the encoder; This is a transmitting device that transmits from an output terminal 10 a Huffman code for an information pattern from the memory 8 that matches the information pattern for each pixel.

Therefore, an analog image signal from a camera VTR or the like led from the image input terminal 1 is converted into, for example, a 6-bit digital image signal by the A/D converter 2, and the image signal for one frame is stored in the current frame memory 3. Ru. The previous frame memory 4 stores the image signal of the previous frame. The calculation device 5 calculates the difference between the image signals stored in the current frame memory 3 and the previous frame memory 4 over all pixels, and derives an information pattern for each pixel. In this embodiment, the arithmetic unit 5 uses the difference between the image signals of both frame memories 3 and 4, but exclusive OR may also be used.

When the arithmetic processing in the arithmetic unit 5 is completed, the current frame memory 3 transfers its contents to the previous frame memory 4 in preparation for inputting the next frame.

For example, a 6-bit information pattern for each pixel that has been subjected to arithmetic processing in the arithmetic unit 5 is led to a statistical processing unit 6, and is
The number of times the same information pattern appears is calculated for the information patterns for all pixels of one frame. Then, the probability of occurrence of the same information pattern is determined from this number of occurrences, the information patterns are rearranged in descending order of probability of occurrence, and the top patterns are adopted. The number of top patterns to be adopted may be determined in advance, and the number of patterns to be adopted may be varied using a switch or the like. For example, if 16 patterns are used, it can be processed using 4 bits.

Here, the information pattern adopted is one obtained by calculating the difference between the current frame and the previous frame by the arithmetic unit 5, and since the image signals of both frames have a correlation, the information pattern is concentrated to some extent near 0. Therefore, by performing such processing, it is possible to reduce the number of bits required compared to the number of bits when inputting an image.

A predetermined number of information patterns with high occurrence probabilities derived from the statistical processing device 6 are led to an encoder 7, Huffman codes are assigned in descending order of appearance probabilities, and the information patterns are stored in a memory 8 together with the corresponding information patterns. For information patterns with a low probability of appearance, a pattern that can be approximated is selected and coded. Then, the transmitting device 9 sequentially scans the information patterns of the image for one frame from the arithmetic device 5, and outputs a Huffman code corresponding to the information pattern that matches the information pattern stored in the memory 8 from the output terminal 10. Send.

Note that the information pattern data to enable the receiving side to compensate for the portions of the image that are not transmitted (that is, the portions that do not change) and the data to inform the code rules of the Huffman coded signal to be sent are prepared in advance. Since the data must be sent to the other party, it is better to make this data smaller than the total number of pixels. Furthermore, if the number of patterns to be adopted is known, there is no need to transmit the Huffman code.

On the receiving side, the first information pattern sent (without the above processing) is stored in advance, and the information pattern corresponding to the Huffman code is extracted from the subsequently sent Huffman code that has undergone the above processing. , On the other hand, for the portion corresponding to the Huffman coded information that is not sent, the stored information patterns may be extracted and decoded.

In the above embodiment, if the input image signal is, for example, 6 bits, and the number of information patterns adopted by the statistical processing device 6 in descending order of appearance probability is 16, the compression ratio of the signal to be transmitted can be set to 273. Also, by employing a Huffman code, the signal can be compressed to about 2/3 (depending on the data), so the signal can be compressed to about 479 in total.

Since the present invention has the above-described configuration, processing can be performed by firmware without special provision of a complicated arithmetic unit, and the processing time can be shortened with a simple configuration. In addition, since the difference between frames or exclusive OR is used, the pixel information pattern is concentrated near 0 or 1,
The number of patterns can be reduced, and image signals can be highly compressed, especially for images with little movement.

[Brief explanation of drawings]

FIG. 1 is a block diagram of an embodiment of the invention. 2--A/D converter, 3--current frame memory. 4.-Previous frame memory, 5-Arithmetic unit. 6-statistical processing device, 7-encoder. 9- Transmitting device. Figure 1

Claims (1)

[Claims] (1) Storage means for converting an analog image signal into a digital image signal and storing one frame worth of digital image signals, and a correspondence between the image signal of the previous frame and the image signal of the current frame stored in the storage means a calculation means for calculating the difference or exclusive OR for each pixel; a statistical processing means for performing statistical processing on the information pattern for all pixels from the calculation means to calculate the probability of appearance for each same information pattern; an encoder that allocates and codes Huffman codes to a predetermined number of the information patterns in descending order of the occurrence probabilities based on the determined appearance probabilities; and a transmission means that transmits the information encoded by the encoder. An image encoding transmission method comprising: