JPH0411481A - Picture encoding device - Google Patents

Abstract

PURPOSE:To broadly improve picture quality by executing encoding and decoding to respective subblocks divided from a point with the point to move for the arbitrary number of the picture elements of up and down and right and left as a start point within the range of subblock size. CONSTITUTION:For encoding, a quantized video signal is inputted to a memory 1 and is inputted to an adaptive subblock dividing circuit 2 next. An address to indicate a moving quantity from a picture data start point within a subblock size range is generated every one flame in an address generating circuit 3 and is transmitted as additional information. A subblock diving start point is moved based on the address from the circuit 3 in the circuit 2, picture data are divided into subblocks from the point, ADCT encoding indicated in circuits 4 to 14 is executed and it is transmitted. For decoding, the data that decoding is finished in an ADCT decoding circuit indicated in 15 to 23 are inputted to an inverse adaptive subblock dividing circuit 24 and a decoded picture is reproduced from the subblock that decoding is finished by the transmitted information of the dividing start point and the rule of copying executed at encoding.

Description

[Detailed description of the invention] (a) Industrial application field The present invention is applicable to image coding and transmission devices, image databases, VT
This relates to an image encoding method used in R.

(b) Conventional technology A typical example of dividing an image into subblocks and encoding them is ADC (IEEE Transact).
ions c.

mmunjctlons, vol, com-25,
No, 11. November 1977) is well known. FIG. 2 shows a block diagram of this method. (1
) The image data that has entered the memory is divided into blocks by the sub-block division circuit (25) with a specified bronch size of 0 from the data start point, and then (4) to (14) AD for each block in the circuit of
CT encoding is performed and the data is transmitted. The encoding is (1
The circuits 5) to (23) perform ADCT decoding, and the image divided into blocks by the inverse division circuit (26) is
Restore the image by storing the data in order from the starting point. It is known that this encoding can achieve a very high compression rate, but it is also known that increasing the compression rate causes the problem of noticeable block distortion.

(c) Problems to be solved by the invention When using an encoding method that divides an image into sub-blocks and encodes them, when the compression rate is increased, distortion can be seen between the divided blocks. If this is applied to an image spanning several frames, block distortion will occur at the same position in each frame, making the block distortion very noticeable and possibly causing problems in terms of image quality.

(d) Means for solving the problem When dividing an image into sub-blocks and performing encoding, the sub-block division start point Ep from the left end of the original image for each image,
q] (p, q are T O = <p-Q + < n), and the first feature is to divide the image into sub-blocks of each nun from this point.

(E) Effect According to the present invention, since the starting point of sub-block division is changed, it is possible to make block distortion less noticeable.

(-\)Example An embodiment of the present invention will be described with reference to FIG.

First, in encoding, the quantized video signal is memo) (1)
and then input to the adaptive subblock division circuit (2). Further, the address generation circuit (3) randomly generates an address indicating the amount of movement from the image data starting point within the sub-block size range for each frame, and transmits this information as additional information.

The adaptive subblock division circuit (2) moves the subblock division start point based on the address input from the address generator, and divides the image data into subblocks from that point.

The situation is shown in FIG. 3. The area indicated by the solid line is the area where the quantized image exists. The dotted line is the division boundary line of the sub-pro/ri that is about to be divided. The sub-pro/re division start point moves from the image data start point S to K according to the amount of movement input from the address generator, and from that point the image is divided into sub-blocks as shown by dotted lines. At this time, pixels that are not included in the sub-block due to the amount of movement of the sub-block division start point are copied in association with the open area (shaded area) of the sub-block.

The image data divided into sub-blocks is divided into (4) to (1).
A DCT encoding shown by the circuit of =1) is performed and transmitted.

At the time of decoding, the data that has been subjected to ADCT decoding, decoding, and decoding shown in (15) to (23) is input to the inverse adaptive subblock dividing circuit (24). Based on the sub-block division start point information obtained and the copying rules used during encoding, the decoding
The decoded image is reproduced from the completed sub-block.

The reason why the range of changing the starting point is limited to the sub-block size range is because the degree of block distortion improvement does not change even if the range is expanded beyond that range.

FIGS. 4 and 5 are flowcharts showing operations during encoding and decoding, respectively. During encoding, the adaptive subblock division circuit 2 first selects the subblock division start point (P) from the address generation circuit 3.

Obtain data for Q) (4]). And this (P.

Q) Based on the data, organize the image data so that the image data starts from (P, Q). Wow, when the image data is M x N, the data d (i.

j: (0≦i<P, o≦j　<Q) as e(K.

f), (X d〈K≦AI+P, N<J!≦N+Qj
t Nicoby (42).

Then, starting from (P, Q), the image data is divided into specified sub-processors 43) and encoded (44).

At the time of decoding, the sent sub-block is decoded.
1), get the starting point (P, Q) of the subblock (52)
. Then, so that the image data starts from (o, o),
To organize data.

In this embodiment, the subblock division start point is determined at the time of encoding and that information is transmitted as invalid information, but the adaptive subblock division circuit and the inverse adaptive subblock division circuit use the same data indicating the subblock division point. In other words, if the starting point is determined in advance, it is possible to eliminate the transmission of additional information.

Furthermore, it is clear that the present invention is applicable not only to ADCT but also to general coding that performs block division.

It is a decoding flowchart for explanation.

2... Sub-block division circuit, 3... Address generation circuit, 24... Reverse sub-block division circuit.

Claims (3)

[Claims] (1) In an encoding device that divides an M×N (M and N are natural numbers) two-dimensional video image into subblocks and encodes them, a point is moved by an arbitrary number of pixels vertically and horizontally within the subblock size range. A video signal encoding device characterized in that the video signal encoding device performs encoding and decoding on each sub-block divided from an arbitrary point of the starting point. (2) The video signal encoding device according to claim 1, characterized in that the point at which an arbitrary pixel has been moved up, down, right, left, etc. is randomly determined for each pixel, and the data of the starting point is transmitted as additional information. . (3) The video signal encoding device according to claim 1, characterized in that the point that is moved by an arbitrary pixel in the vertical and horizontal directions is a point that has been determined in advance during encoding and decoding.