JPH02248181A - Picture coding system and coder - Google Patents

Abstract

PURPOSE:To improve the subjective evaluation of the entire pattern by arranging more coded bit number to a part in a pattern whose quality is desired to be improved and then applying the coding processing to the part. CONSTITUTION:After a picture frame inputted from an input deice 101 is fetched by a frame memory 102, the frame is split into blocks by a block split circuit 103. The information representing at which block in one pattern an existing block corresponds is outputted from the circuit 103 and fed to a high picture quality request position setting circuit 104. When the circuit 104 makes dial and set a weighted part, the weighted value is decided by a weighting circuit 105. A block signal outputted from the circuit 103 is also fed to an information quantity calculation circuit 106, weighted via a multiplier circuit 107 and coded via an information quantity storage circuit 108 and a bit distribution circuit 109. Thus, the picture quality of a remarkable part is improved and the subjecting evaluation of the entire pattern is improved.

Description

[Detailed Description of the Invention] [Object of the Invention] (Industrial Application Field) The present invention relates to an image encoding method and encoding device related to a compression encoding method for still images or moving images of videophones, electronic cameras, etc. be.

(Prior art) When compressing and encoding still images or moving images from videophones, electronic cameras, etc., there is a demand to encode one screen as efficiently as possible within a predetermined number of encoding bits. be.

In such cases, the problem is how to allocate the predetermined number of encoding bits within the screen, but when it comes to image encoding, it is important to improve the S/N ratio (signal-to-noise ratio) and to conduct subjective evaluations. The most important issue is how to improve it. At this time, if you know which part of the screen the person actually viewing (evaluating) the image focuses on, you can allocate a larger number of encoding bits to that part and perform the encoding process (if the SN ratio is Even if the performance does not improve, the subjective evaluation for that person will improve.

For example, when you take a snapshot of a person standing in a certain landscape, the part that attracts the most attention is the person, and the image quality of this part greatly contributes to the overall subjective evaluation.

However, when there is a mixture of coarse and fine parts in one screen, the fine parts of the picture or the amount of fineness corresponding to the frequency band to which human visual perception is sensitive. A method of performing encoding processing by allocating a large number of encoding bits to each portion has been considered. Therefore, in the case of a picture where the entire screen can be evaluated equally,
Although it is expected that subjective evaluation will improve overall, for images where a person is standing in a certain landscape as mentioned above,
Particularly when the background has a detailed pattern, a large number of encoding bits will be taken up for the background part, and the number of encoding bits allocated to the human part, where image quality is most important, will decrease. Therefore, despite adaptive processing according to the detail of the picture, the subjective evaluation did not improve.
In some cases, this method has the disadvantage of causing deterioration of subjective evaluation.

(Problem to be solved by the invention) As described above, in the past, adaptive processing that placed importance on subjective evaluation was not performed sufficiently, and the image quality of the parts that were originally attracting attention within one screen did not improve, resulting in There was a problem that the overall subjective evaluation did not improve.

Therefore, the present invention solves these problems and focuses on the part of the screen that attracts the most visual attention, or the part on which the person viewing the processed image wants to place the highest importance on image quality, or the part that performs the encoding process. By distinguishing that part from others in advance and allocating a larger number of encoding bits to the part for which humans most want to improve the image quality, the image quality of that part is improved and the subjective evaluation of the entire screen is performed. The purpose of this study is to provide an image encoding method that improves image quality.

[Structure of the invention]

(Means for Solving the Problems) The present invention determines in advance a portion of one screen for which the image quality should be improved, weights that portion to distinguish it from other portions, and assigns more encoding bits to that portion. distribute the numbers. During actual encoding, the encoding process for each part is finished using the pre-allocated number of encoding bits, so even though the entire screen uses the same amount of code as before, subjective evaluation of the parts where the image quality should be improved is necessary. This method also improves the overall subjective evaluation.

Note that the position, size, or shape of the part you want to improve image quality can be set for fixed settings (fixed type: For example, in the case of photographs, important parts are often located in the center of the screen, so (distinguishing the center area from the rest and allocating a large number of encoding bits), or setting it by selecting from presets (
A selection type) or a type in which the position and shape can be freely changed and set (free type) can be considered.

In addition, regarding the weighting process for the part where you want to improve the image quality, the weighting value can be fixed type, two selection type, or free type as mentioned above, but in particular, the resolution of the weighted part and the resolution of the other parts are It is also possible to use techniques that adaptively adjust the weighting values at their boundaries so that they are connected.

(Function) In this way, a portion of one screen for which the image quality is desired to be improved is determined in advance, and by weighting that portion, it is distinguished from other portions, and a larger number of encoding bits is allocated to that portion. In actual encoding, by finishing the encoding process for that part using the allocated number of encoding bits, the subjective evaluation of the part where the image quality should be improved is improved, and ultimately the subjective evaluation of the entire screen is also improved. It becomes possible to do so.

(Example) Hereinafter, the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram of a transmitting side and a receiving side of an embodiment of the present invention. An image input from an input device 101 is captured into a frame memory 102 and then divided into blocks by a block division circuit 103.

On the other hand, the block division circuit 103 outputs information indicating the position of the current block within one screen, and sends it to the high-quality required position setting circuit 104. This is where you set the part of one screen for which you want to have the best image quality (the part that is weighted when determining the number of encoding bits for that part), and its position, shape, weighting value, etc. can be set from outside. It is possible to input

For example, as shown by the shaded areas in Fig. 2 (a) to (c),
It is possible to change the size of the weighted part or change the position of the weighted part by external operation, as shown in FIG. 2(d). Furthermore, the weighting value may be a continuous function as shown in (a) in FIG. It may also be a step-like function as shown in . It is also possible to apply steep weight to only the shaded area as shown in (c), but in this case, the continuity of resolution between the shaded area and other areas will be lost, and block distortion will occur there. Therefore, the methods (a) and (b) are more desirable.

Ideally, the method for setting the weighted portion would be to be able to freely set its position, shape 2 weighting value, etc., but the type may be limited due to constraints such as hardware scale and encoding processing time. In that case, in addition to the example shown in Fig. 2, for example, the human shape is assumed in advance as shown in Figs. It is also conceivable to prepare a system for multiple users, (C) for multiple users, etc., and set the weighted portion by selecting from these. Furthermore, the present invention is not limited to humans, but may be a method in which a wide range of general figures are prepared and their positions and sizes can be changed, as shown in FIGS. 5(a) to 5(c), for example. Furthermore, as shown in (d), it is also possible to determine the center position C of the part to be weighted, and then draw a circle with a radius from there and set the inside of the circle as the weighted part. Alternatively, as shown in Figure 6, use two horizontal and vertical straight lines (both can be moved freely) and set the weighting area in the diagonally shaded area surrounded by these four straight lines. Another possible method is to do so.

In this case, all four lines do not need to be straight lines, and the moving directions of the lines do not need to be only horizontal or vertical. Incidentally, settings for the high image quality required position setting circuit 104 can be selected or set using a dial.

For example, in the case of Fig. 4, (a), (b), and (c) are input into the frame memory in advance, and one of them is selected from the outside, and in the case of Fig. 6, two each are input. Move each dial in a straight line and use the input device 101 on the monitor 100.
It also matches images sent in real time.

As described above, the setting of the weighted part in the high image quality required position setting circuit 104 is performed using the dial, and the weighting value is determined in the weighting circuit 105 using address information corresponding to this setting.

The block signal output from 103 is sent to the information amount calculation circuit 1
06, and here the amount of information in that block (the amount that indicates the fineness of the pattern, expressed as the variance, standard deviation, or output value after applying a band pass filter) is calculated. The weighting value calculated and determined in step 105 is multiplied by the above information amount in an arithmetic circuit 107 (weighting is applied), and the final information amount is determined and sent to the information storage circuit 108. Here, the weighted information amount Bn of each block (n = 1, 2, . . . NUN is the total number of blocks in one screen) is stored for -screens, and the total information amount A
(=B□182+...BN) is calculated.

Thereafter, the input image stored in 102 is again divided into blocks 103, and each block is sent to the encoding circuit 11.
Sent to 0. On the other hand, since the weighting information amount for each block has already been stored in the information storage circuit 108, the information amount Bi of the block currently transmitted to 110 is calculated from here.
The ratio Bi/A between the total information amount A and the total information amount A is output and sent to the bit allocation circuit 109.

Here, for the preset total number of encoded bits for one screen X (which can be input externally), xj =
is calculated, and xl is the number of encoded bits for that block.
110 is sent. 110 encodes the block signal sent from 103 within the number of encoding bits of this xi, or even if it exceeds the number of encoding bits, the excess is within a predetermined range. is processed and sent to the transmission path 111 (through this processing, the total number of encoded bits for one screen as a whole is within the predetermined number of bits, or even if the total number of encoded bits exceeds the total number of encoded bits, the excess amount is (enables encoding processing that falls within the specified range).

On the other hand, on the receiving side, the signal sent from 111 is decoded by a decoding circuit 112, and a reproduced image is displayed on a monitor 113. In other words, on the receiving side, decoding is performed through very simple processing.

FIG. 7 is a block diagram showing a portion of the transmitting side (portion 150 in FIG. 1) of another embodiment of the present invention.

The signal sent from the information amount calculation circuit 106 is stored in the information amount storage circuit 301 for each block, and - the ratio of the information amount of each block to the total information amount within the screen (108 in the embodiment of FIG. ) is sent to the bit allocation determining circuit 302. Here, the number of encoded bits for each block is determined as in the process at step 109, but weighting has not yet been performed at this stage.

On the other hand, based on the signal sent from the weighting circuit 105, the bit shift setting circuit 303 shifts the number of encoded bits between each block (a part of the number of encoded bits of the unweighted block is transferred to the number of encoded bits of the block with weighting). Determine the ratio (to the number of encoded bits). For example, as shown in FIG. 8, when weighting is performed on five blocks near the center, the shift of the number of encoded bits as shown in the figure is determined depending on the magnitude of the weighting values.

In other words, in addition to the number of encoding bits determined in step 302, 1 to 10 bits are added to the weighted part.On the other hand, the total number of encoding bits for the entire screen must be kept constant, so the weighting is The unweighted surrounding blocks (for example, as shown in the figure, if there are 25 blocks in one block line, the remaining 20 unweighted blocks will be the remaining 20 blocks) - subtract one rule bit at a time. This instruction is sent to the redistribution circuit 304.

In step 304, upon receiving this instruction, addition and subtraction processing of the number of bits is performed on the number of encoded bits of each block.

As a result, the output state (a
) (output information: number of encoded bits for each block) ゛“
a91alfi l ai□l ai21 ”'a 工、
, a1□, the output state from 304 (output information:
The final coding bit number distribution after weighting of each block is (b)"'a9-Laio-L all-L'
h2-L"'a16-1. It becomes like a□7-1,
As a result, weighting was performed.

Note that the number of encoding bits determined in step 304 is determined by the encoding circuit 1.
10, and undergoes encoding processing similar to that described in the embodiment of FIG.

〔Effect of the invention〕

As described above, according to the present invention, - weighting is applied to a portion of the screen where the image quality is desired to be improved, particularly a portion that is easily noticeable to the viewer of the reproduced image, to distinguish it from other portions;
Allocate a larger number of encoding bits to that part. During actual encoding, by finishing the encoding process of that part with the allocated number of encoding bits, the image quality of the reproduced image of the part whose image quality is desired to be improved is improved, and the subjective evaluation of that part is improved. This ultimately makes it possible to improve the subjective evaluation of the entire screen.

[Brief explanation of drawings]

Fig. 1 is a block diagram showing an embodiment of the present invention, Fig. 2 and Figs. FIG. 7 is a block diagram showing another embodiment of the present invention, and FIG. 8 is a diagram explaining the weighting process. 107... is an arithmetic circuit, 111... is a transmission line. 150... Encoding bit allocation determination part for each block. 201...Frame (or field). 202 Boundary between parts to be weighted and parts not to be weighted. 207...Center of weighted part. 208...Radius of the weighted part (circle) from the center. 211... Weighting value. 220-223: A movable straight line for determining a weighted portion, 224: A weighted portion. Agent Patent Attorney Noriyuki Chika Yudo Hikaru Matsuyama

Claims (7)

[Claims] (1) After allocating encoded bits to each part in one screen using part or all of the image information in one screen, the number of encoded bits allocated to each part, or the number of encoded bits An image encoding method that encodes each part so that even if the number exceeds the number, the excess is within a predetermined range, and the image information represents the fineness of the picture on the screen. It is possible to weight the evaluation value or the evaluation value representing the fineness of the picture for each location on the screen and use the weighted evaluation value, or simply to weight each location and use the weighted value as is. Featured image encoding method. (2) An evaluation value representing the fineness of a picture in the screen is used as the image information according to claim 1, and after determining the number of encoding bits for each part in the screen based on the evaluation value, An image code characterized in that the assigned number of encoding bits is weighted for each location in the screen, and the number of encoding bits after weighting is set as the predetermined number of encoding bits for each part. method. (3) It has a function that allows the location of weighting or the weight value when weighting is to be set freely, or the location and value can be set from one or more types of pre-prepared types. 3. The image encoding system according to claim 1, further comprising a function that can be set by selecting. (4) It has a function that allows you to freely set the shape of the part to be weighted, or it has a function that allows you to set the shape by selecting from one or more types of patterns prepared in advance. Claim 1 characterized by
and the image encoding method described in 2. (5) The image encoding method according to claim 1 or 2, further comprising a processing function that allows the resolution of the portion to be weighted and the resolution of the other portion to be smoothly connected. (6) As an evaluation value that represents the fineness of the picture on the screen, the variance, standard deviation, the difference between the maximum and minimum pixel values in each part, or the band pass filter on the pixel values in each part Claim 1 characterized in that the output value after multiplication is used.
and the image encoding method described in 2. (7) In an image encoding device that encodes and transmits an input image or stores it in a memory, there is a means for specifying an arbitrary part of the image during encoding, and a means for specifying an arbitrary part of the image, An image encoding device comprising: means for controlling to allocate a large number of bits; and means for encoding the image according to the bit allocation by this means.