JPH0685580B2 - Two-dimensional code assignment method for motion vector - Google Patents

Description

Description: (a) Technical Field of the Invention The present invention relates to motion-compensated inter-frame predictive coding in moving image band compression transmission, and relates to a two-dimensional code allocation method of motion vectors capable of improving transmission efficiency.

(B) Background of the Technology In the motion-compensated interframe predictive coding method used for moving image band compression transmission, one block image of an input frame,
The block image closest to this block image is 1
Detect from the image of the frame before the frame, move the found block image to the same position as one block image on the input frame (the spatial movement of the block is obtained by a variable delay in time), and predict this As a block image, an error signal with respect to an image block of an input frame is generated, a motion vector composed of a moving distance and a moving direction is obtained, and the error signal and the moving vector are encoded and transmitted. The present invention relates to the coding of motion vectors.

(C) Conventional Technology and Problems Conventional motion vectors are variable length coded individually in the horizontal and vertical directions and transmitted. An example of this variable length coding is shown in FIG. That is, the transmission code is assigned to the motion difference value for each horizontal and vertical direction and encoded. However, this conventional method has a drawback in that the transmission efficiency is poor because a considerable number of transmission bits are required to transmit the motion vector information.

(D) Object of the Invention The object of the present invention is to provide a two-dimensional code assignment system of motion vectors which can improve the transmission efficiency in view of the above-mentioned drawbacks.

(E) Structure of the Invention In order to achieve the above object, the present invention generally has many block images with little motion between successive frames of an actual moving image, so that the motion vector actually has a statistically absolute value. It is concentrated in the vicinity of 0, and by paying attention to this point and shortening the code length of the motion vector having a small absolute value, the average length of the code is shortened to the whole motion vector.

That is, in the motion-compensated inter-frame prediction encoding, when encoding a motion vector given by a horizontal component value and a vertical component value in units of pixel intervals, a predetermined number n ² whose absolute square value includes 0 A variable length code having a shorter code length is assigned to the following plurality of motion vectors as the square value of the absolute value is smaller, and the horizontal vector of the motion vector is set to the motion vector having the square value of the absolute value of n ² +1 or more. This is a two-dimensional code assigning method for motion vectors, characterized in that equal-length codes are assigned to the component values and vertical component values, respectively, and an identification code for distinguishing from the variable-length codes is added.

(F) Embodiments of the Invention Hereinafter, embodiments of the present invention where n ² = 4 (decimal number) will be described with reference to the drawings. FIG. 2 is a two-dimensional layout diagram of the square value of the absolute value of the motion vector according to the embodiment of the present invention, in which the square value of the absolute value in decimal number corresponding to the vertex of each motion vector is entered. FIG. 3 is a transmission code assignment diagram for the squared absolute value according to the embodiment of the present invention, and FIG. 4 is a block diagram of a circuit for converting a motion vector into a transmission code according to the embodiment of the present invention.
The figure is a diagram showing the frequency of occurrence of motion vectors and the coding efficiency.

In FIG. 4, 1 is a ROM and 2 is a multiplexing circuit.

In this embodiment, the motion vector has a pixel interval as a unit,
The component values in the horizontal and vertical directions are given as a binary 4-digit number.

Horizontal vector Vx and vertical vector of motion vector
The square value Vx ² + Vy ² of the absolute value of Vy is arranged two-dimensionally around the reference point (square value 0 of the absolute value) as shown in FIG. 2, and the square value of the absolute value is 1,2,4. Each of the four motion vectors is given a number, and the variable code length of shorter bits is assigned to the transmission code as the square value of the absolute value and the number is closer to the reference point according to the number. To do. In this case, the absolute square value 0
Up to -4, a code obtained by comprehensively converting the absolute value of the motion vector and the direction is transmitted. On the other hand, if the absolute square value is 5 or more, 00
001 is added to the beginning as an identification code, and the binary digits of equal digits of the horizontal and vertical components of the motion vector are transmitted as equal length codes. The two-dimensional arrangement of the motion vectors shown in FIG. 2 and the transmission code shown in FIG.
When the horizontal vector Vx and the vertical vector Vy of the motion vector are input and stored in the M1, the ROM1 detects the square value of the absolute value and sends the transmission code corresponding thereto to the multiplexing circuit 2. In this case, the transmitted transmission code is transmitted for the square value of the absolute value from 0 to 4. Therefore, relative value vector transmission is performed. In case of 5 or more, 00001 from ROM1 is multiplexed circuit 2
To the multiplexing circuit 2, the input horizontal vector V
A transmission code in which 00001 is added to the beginning of an 8-bit code indicating the 4-bit absolute value for each of x and Vy is transmitted. Therefore, absolute value vector transmission is performed. The frequency of occurrence of the square value of the absolute value of the motion vector is said to be 83% at the reference point, 1% at 1 and 2 points respectively, 0.7% at 4 points and 4.3% at the other points due to the nature of the image. . It is the first to transmit motion vectors based on this.
In the case of the present invention, the average code length is 4.05, 1.52 as shown in FIG. 5 when the transmission code of the conventional example shown in the figure is used and when the transmission code of the embodiment of the book of FIG. 3 is used. Will transmit a motion vector with a code length of 37.5% compared to the conventional example.

In the description of the above-mentioned embodiment, when the square value of the absolute value of the motion vector is n ² +1 or more, the identification code is added to the binary code itself of the horizontal and vertical component values of the motion vector. Although an example has been shown, when the horizontal and vertical component values are not given as binary numbers, it is necessary to add a process of binary conversion.

(G) Effect of the Invention As described in detail above, according to the present invention, since the motion vector information can be transmitted with a smaller number of bits by arranging the motion vector two-dimensionally, the transmission efficiency is improved. Together with this, there is an effect of band compression.

[Brief description of drawings]

FIG. 1 is a diagram showing a transmission code for a motion difference value in a conventional example, FIG. 2 is a two-dimensional layout diagram of square values of absolute values of motion vectors according to an embodiment of the present invention, and FIG. FIG. 4 is a block diagram of a circuit for converting a motion vector into a transmission code according to an embodiment of the present invention, and FIG. 5 is a motion vector generation frequency and coding efficiency. FIG. In FIG. 4, 1 is a ROM and 2 is a multiplexing circuit.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Toshitaka Tsuda Toshitaka Tsuda 1015 Kamiodanaka, Nakahara-ku, Kawasaki City, Kanagawa Prefecture, Fujitsu Limited (72) Inventor Hideo Kuroda 1-2356 Takeshi, Yokosuka City, Kanagawa Prefecture Yokosuka Electric Corporation In the Communications Research Laboratory (72) Inventor Naoki Takekawa 1-2356 Take, Yokosuka City, Kanagawa Inside the Yokosuka Telecommunications Research Laboratories, Nippon Telegraph and Telephone Public Corporation (56) Reference JP-A-58-197983 (JP, A) JP-A-53-146526 (JP, A)

Claims (1)

[Claims] 1. In a motion compensation interframe predictive coding system for moving image data, when coding a motion compensation motion vector given by two-dimensional coordinates of a horizontal component value and a vertical component value in units of pixel intervals. , The sum of the square value of the horizontal component value of each motion vector and the square value of the vertical component value is calculated with the motion vector of which the horizontal component value and the vertical component value are 0 as a reference point. So that the length becomes shorter,
A method for allocating a two-dimensional code of a motion vector characterized in that a variable-length code is assigned in advance according to the two-dimensional coordinate of the motion vector and the two-dimensional coordinate of each motion vector is encoded by the assigned variable-length code.