JPH06105300A - Picture encoder - Google Patents

Abstract

PURPOSE:To shorten an information processing period of time, and to reduce an information amount at the time of hierarchically encoding a picture with a large amount of information. CONSTITUTION:An input picture is divided into plural bands by a band dividing part 101, and a picture encoding is operated on each band. At that time, the number of the band division is equalized with the number of provided memory parts 102-108, provided moving vector detecting parts 109-115, and provided encoding/decoding parts 117-123, and they are operated in parallel, so that the processing period of time can be shortened. And also, a moving vector control part 116 limits the number of moving vectors so that only the moving vectors of one part of the bands can be used instead of using the moving vectors of the entire bands, so that the information amount can be reduced.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image coding apparatus used for video conference, high-definition signal transmission and the like.

[0002]

2. Description of the Related Art An image encoding device is a device for compressing and communicating image signals, and in the process of compression, techniques such as inter-frame difference, motion compensation, orthogonal transform, quantization and encoding are generally used. Is used. Also, as a processing method,
There are a method in which the original signal is image-encoded as it is without being band-divided, and a method in which the original signal is divided into frequency bands and image-encoded for each band (called a hierarchical coding method). Here, the apparatus based on the hierarchical coding system will be described below with reference to FIGS.
Will be explained.

FIG. 3 is a block diagram showing the structure of an image coding apparatus according to the conventional hierarchical coding system, and FIG. 4 is a block diagram for explaining the band division processing. In FIG.
Reference numeral 301 denotes a band division unit, which divides an input signal into four bands by applying a one-dimensional frequency division filter in the horizontal direction and the vertical direction as shown in FIG. Divide into 7 bands by applying a vertical filter. 302 is a memory unit composed of a current frame memory and a previous frame memory, and 303 is
A motion vector detection unit for detecting a motion vector from the data of the current frame memory and the data of the previous frame memory, 30
Reference numeral 4 calculates an interframe difference signal, performs orthogonal transformation, quantization, coding, etc., and outputs it to the line, and inversely quantizes and inversely orthogonally transforms the quantized data. It is a coding / decoding unit for writing to.

Next, the operation of the conventional image coding apparatus will be described. The input image is divided into seven bands by the band dividing unit 301 as shown in FIG. 4, and stored in the current frame memory in the memory unit 302, respectively. The motion vector detection unit 303 detects a motion vector for each band using the current frame image data and the previous frame image data.
One motion vector is assigned to each 16 × 16 block or 8 × 8 block. Next, the encoding / decoding unit 304
Calculates an inter-frame difference signal using this motion vector, performs orthogonal transformation, quantization, coding, etc., and outputs it to the line. Further, the quantized data is inversely quantized and inversely orthogonally transformed, and written in the previous frame memory of the memory unit 302.
These encoding / decoding processes are performed for each 16 × 16 block or 8 × 8 block.

[0005]

However, in the above-mentioned conventional image coding apparatus, when handling an image having an extremely large amount of information such as high-definition, it takes a long time for the coding process and all motions in all bands. Transmitting a vector has a problem that a large amount of information is required and a broadband line is required.

[0006]

SUMMARY OF THE INVENTION The present invention provides an excellent image coding apparatus that solves such a conventional problem. The first means of the present invention is a memory section and a motion vector detecting section. The number of encoding / decoding units is the same as the number of band divisions, and these are operated in parallel. The motion vector to be transmitted to the line is such that only the motion vector in a part of the band is transmitted.

The second means of the present invention is to detect a motion vector from an image before band division, calculate a motion vector of each band after band division from this motion vector, and transmit it. .

[0008]

According to the first means of the present invention, the memory section, the motion vector detecting section, and the code decoding section are provided in the same number as the number of band divisions, and these are operated in parallel. It can be done at high speed. Further, since the motion vector to be transmitted is limited to a part of the band, it is possible to reduce the amount of information.

Further, the second means of the present invention detects the motion vector from the image before band division and calculates the motion vector after band division from this, so that the amount of information can be reduced.

[0010]

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is a block diagram showing the structure of an image coding apparatus according to a first embodiment of the present invention, and FIG. 4 is a block diagram for explaining band division processing. In FIG. 1, reference numeral 101 denotes a band division unit, which divides an input signal into four bands by applying a one-dimensional frequency division filter in the horizontal direction and the vertical direction as shown in FIG. Then, it is divided into 7 bands by applying horizontal and vertical filters. Reference numerals 102 to 108 denote seven memories that operate independently of each other, each of which includes a memory unit A to a memory unit G, and each memory unit has a current frame memory and a previous frame memory. Reference numerals 109 to 115 denote motion vector detection units that operate independently, and include seven motion vector detection units A to G. 1
Reference numeral 16 is a motion vector control unit that calculates the motion vector detected by the motion vector detection units 109 to 115 and calculates a new motion vector. Reference numerals 117 to 123 are code decoding units that operate independently. It is composed of 7 units from A to a coding / decoding unit G, calculates an inter-frame difference signal from the current frame and the previous frame, performs orthogonal transformation, quantization, coding, etc., and outputs it to the line, Inverse quantization and inverse orthogonal transform are performed, and the result is written in the previous frame memory of the memory units 102 to 108.

Next, the operation of the first embodiment of the present invention will be described. The input image signal is divided into seven bands by the band dividing unit 101 and written into the current frame memories of the memory units A to G, respectively. The motion vector detection units 109 to 115 detect the motion vectors MVa to MVg from the current frame data of the memory units 102 to 108 and the previously input previous frame data, and send them to the motion vector control unit 116.

The motion vector control unit 116 uses the motion vectors detected by the motion vector detection units 109 to 115 as
The arithmetic processing as shown in the schemes 1 to 7 of (Table 1) is performed and sent to the code decoding units 117 to 123.

[0013]

[Table 1]

In Table 1, method 1 sends all motion vectors as they are, as usual.

In method 2, only MVa is used, and motion compensation is performed only in band A. In this case, the transmission information amount of the motion vector is 1/16 of the method 1.

Method 3 uses only MVa to perform motion compensation for band A, band B, band C, and band D. In this case, the transmission information amount of the motion vector is 1/16 of the method 1.

Method 4 uses only MVa and performs motion compensation for band A, band B, band C, band D, band E, band F, and band G. However, the motion vector in the band E, the band F, and the band G is twice the MVa, and the block corresponds to 1: 4. In this case, the transmission information amount of the motion vector is the method 1
1 / 16th of that.

Method 5 is MVa, MVb, MVc, MVd
Are used to perform motion compensation for band A, band B, band C, and band D, respectively. In this case, the transmission information amount of the motion vector is 1/4 of the method 1.

Method 6 is MVa, MVb, MVc, MVd
Are used to perform motion compensation of band A, band B, band C, and band D on the average of these four motion vectors. In this case, the transmission information amount of the motion vector is 1/16 of the method 1.

Method 7 is MVa, MVb, MVc, MV
Using d, MVe, MVf, and MVg, the average of the seven motion vectors is band A, band B, band C, band D, and
Motion compensation of the band E, the band F, and the band G is performed. In this case, the transmission information amount of the motion vector is 1/16 of the method 1.

One motion vector is detected for each 16 × 16 block or 8 × 8 block.

Using the motion vectors as described above, each of the coding / decoding sections 117 to 123 calculates an inter-frame difference signal, performs orthogonal transformation, quantization, coding, etc., and outputs it to the line. In addition, the quantized data is inversely quantized and inversely orthogonally transformed, and is written in the previous frame memory of the memory units 102 to 108. These coding / decoding processes are performed for each 16 × 16 block or 8 × 8 block.

FIG. 2 is a block diagram showing the arrangement of an image coding apparatus according to the second embodiment of the present invention. In FIG. 2, 201 is a memory A for storing the image of the current frame, 2
Reference numeral 02 is a memory B for storing the image of the previous frame, 203 is a band division unit for dividing the image signal of the current frame into seven bands as shown in FIG. 4, and 204 is the image of the current frame of the memory A201 and the previous frame of the memory B202. , A memory C that stores a band-divided image for each band, 206 a differential signal between the current frame and the previous frame, and orthogonal transformation, quantum Output to the line by performing encoding, encoding, etc., and dequantized and inverse orthogonal transformed the quantized data, and written in the memory for the previous frame of the memory C205,
It is also a code decoding unit that synthesizes each band and writes it in the memory B202.

Next, the operation of the second embodiment of the present invention will be described. Now, assuming that the information of the previous frame is stored in the memory B202, the input signal is input to the memory A201 as the current frame image. Motion vector detection unit 20
4 detects the motion vector MV from the image of the previous frame and the image of the current frame, and sends it to the encoding / decoding unit 206. The band dividing unit 203 extracts the image of the current frame from the memory A201, divides it into seven bands as shown in FIG.
205 to the current frame memory.

The coding / decoding unit 206 determines the motion vector MV sent from the motion vector detection unit 204 as the motion vector of each band as in the system shown in (Table 2).

[0026]

[Table 2]

In Table 2, the method 1 is MVa to MV.
In d, the motion vector MV obtained from the information before division is divided by 4 to be an integer, and in MVe to MVg, the motion vector MV obtained from the information before division is divided by 2 to be an integer.

In the method 2, MVa to MVd are the motion vectors MV obtained from the information before division and are divided by 4 to obtain integers,
e to MVg force the motion vector to zero.

In the method 3, MVa is the motion vector MV obtained from the information before division, divided by 4 to be an integer, and MVb to MV.
g forces the motion vector to zero.

The motion vector MV obtained before the division is 1
If 6 × 16 blocks are required, MVa,
MVb, MVc, MVd correspond to 4 × 4 blocks,
MVe, MVf, and MVg correspond to 8 × 8 blocks.

In these cases, the information is reduced to about 1/7 as compared with the case where the motion vectors of all the blocks in each band are output after the band division.

The coding / decoding unit 206 calculates an interframe difference signal using the above motion vector, performs orthogonal transformation, quantization, coding, etc., and outputs it. Further, the quantized data is inversely quantized and inversely orthogonally transformed, and is written in the previous frame memory of the memory C, and each band is synthesized and written in the memory B202. These encoding / decoding processes are 16
It is performed for each block of × 16 or 8 × 8.

[0033]

As described above, according to the first embodiment of the present invention, the same number of memory sections, motion vector detecting sections, and code decoding sections as the number of band divisions are provided so that the code decoding time can be reduced. Shortened and provided with a motion vector control unit,
By limiting the amount of motion vector information to be transmitted,
The amount of information can be reduced.

Further, according to the second embodiment of the present invention, a motion vector detecting section for detecting a motion vector from the information before band division is provided, and the motion vector of each band is limited by using this motion vector. Therefore, the amount of information can be reduced.

[Brief description of drawings]

FIG. 1 is a configuration block diagram of an image encoding device according to a first embodiment of the present invention.

FIG. 2 is a configuration block diagram of an image encoding device according to a second embodiment of the present invention.

FIG. 3 is a configuration block diagram of a conventional image encoding device.

FIG. 4 is a diagram for explaining the operation of a band division unit.

[Explanation of symbols]

 101 band division unit 102-108 memory unit 109-115 motion vector detection unit 116 motion vector control unit 117-123 code decoding unit 201 memory A 202 memory B 203 band division unit 204 motion vector detection unit 205 memory C 206 code decoding Unit 301 band division unit 302 memory unit 303 motion vector detection unit 304 code decoding unit

Claims (7)

[Claims] 1. A band division unit that divides an input image signal into a plurality of frames having different bands through a filter, and a plurality of independent units that store images of the plurality of divided current frames and band-divided previous frames. A memory unit, a plurality of independent motion vector detection units for detecting motion vectors based on the images of the current frame and the previous frame corresponding to each memory unit, and a motion vector of each motion vector detection unit. A motion vector control unit for calculating a motion compensation method for each divided band, and using the motion vector calculated by the motion vector control unit, calculates an inter-frame difference signal between the previous frame and the current frame, On the other hand, orthogonal transformation, quantization, encoding, etc. are performed and output sequentially, and the quantized data is inversely quantized and inverse orthogonally transformed, and the previous frame memory of the memory unit An image encoding device including a plurality of independent encoding / decoding units for writing data in the memory. 2. The image coding apparatus according to claim 1, wherein the plurality of independent memory units and the plurality of independent motion vector detection units have the same number of band divisions, respectively. 3. The image coding apparatus according to claim 1, wherein the plurality of independent memory units and the plurality of independent coding / decoding units have the same number of band divisions, respectively. 4. The motion vector control unit controls to calculate a motion vector of another division unit from one motion vector of a band-divided frame. Image encoding device. 5. The motion vector control unit is controlled so as to calculate a motion vector of another division unit from a motion vector of a plurality of band-divided frames. Image encoding device. 6. The motion vector control unit is controlled so as to calculate the motion vector of another division unit by averaging the motion vectors of a plurality of band-divided frames. 3. The image coding device according to 3. 7. A first memory unit for respectively storing an input image and a previous frame image, a motion vector detection unit for detecting a motion vector from the previous frame image and the input image in the memory unit, the input image and the previous image. A band division unit that divides the frame image into a plurality of frames having different bands through a filter, a second memory unit that stores each of these band-divided images, and each band using the detected motion vector. Motion vector is calculated, and this motion vector is output, and using this, the inter-frame difference signal between the previous frame and the current frame is obtained, orthogonal transformation, quantization, coding, etc. are performed, and the quantized data Is quantized and inversely orthogonally transformed and written in the previous frame memory of the second memory section, and each band is synthesized to generate the front frame of the first memory section. An image encoding device, comprising an encoding / decoding unit for writing to a frame memory.