JPS62299187A - Video encoding processing method - Google Patents

Abstract

PURPOSE:To solve the unnaturalness of movement due to a deframing picture by dividing a frame into one to execute the usual encoding and the other to encode a dynamic vector only, and executing the encoding. CONSTITUTION:A frame extracting part 100 extracts a skipping frame, for example, #1F, #4F, #7F and #10F, and is held at frame memories 3a and 3b. On the other hand, frames, which come to be the object of the time lapse, for example, #2F, #3F, #5F and #6f,...are respectively held at frame memories 4a and 4b. Under this condition, the #1F and the #2F are compared, the #2V and the #4F are compared and concerning the part with an action in the #2F, an action vector #2V is generated. In the same way, the #1F and the #3F are compared, the #3F and the #4F are compared and concerning the part with the action is the #3F, an action vector #3V is generated. A transmission line encoding part 6 outputs onto the transmission line in the sequence of the #4F, #2V, #3V, #7F, #5V, #6V, #10F....

Description

[Detailed Description of the Invention] 3. Detailed Description of the Invention [) Already Required] The present invention solves the problem of unnatural motion caused by frame-dropped images in a video encoding processing method that performs low bit rate encoding of video images. to solve.

In this system, encoding is performed separately into frames in which normal encoding is performed and frames in which only motion patterns 1 to 1 are encoded, thereby preventing deterioration in transmission quality.

[Industrial application field]

The present invention relates to a video encoding processing method, especially regarding frames subject to so-called frame dropping.

The present invention relates to a video encoding processing method that transmits motion vectors.

[Conventional technology]

Conventional methods for this type of technology include a method in which the receiving side directly substitutes the frame that was transmitted before the frame, or a method in which the frame is dropped using frames before and after the frame, on the receiving side. A method of dropping the frame rate ['1] is known.

[Problem that the invention seeks to solve]

In the case of the former method, the movement is unnatural. Further, in the case of the latter method, although the unnaturalness of the movement is eliminated to some extent, there are still problems that need to be improved regarding the image quality of the reproduced image. Furthermore, when the number of frame-dropped frames is unavoidable, there is a problem in that the motion in the reproduced image deviates extremely from the actual motion. In particular, there has been a problem in that image quality deterioration is noticeable with respect to the outline of a moving object relative to the background and the background that appears or disappears.

[Means for solving problems]

The present invention solves the above-mentioned problems, and FIG. 1 shows a basic configuration diagram of the present invention. The reference numeral 100 in the figure is a frame extracting unit, which includes discrete frames, for example #1 frame (#I
F), #4 frame (#4F), #7 frame (#7F
),... are extracted.

3a is a frame memory that holds #1 frame (#IF), 3b is a frame memory that holds #4 frame (#4F), and 4a is a frame memory that holds #2 frame (62F), which is the frame drop frame. , 4b is a frame memory that also holds #3 frame (#3F), 5 is a motion compensation processing section, and 6 is a transmission line encoding section.

The motion compensation processing unit 5 stores the contents (#) of the frame memory 3a.
IF) and the contents of the frame memory 3b (#4F), a motion vector (#2■) corresponding to the contents of the frame memory 4a (t$2F) is generated. Similarly,
With reference to the contents (#IF) and (14F), a motion vector (#3V) corresponding to the contents (#3F) of the frame memory 4b is generated.

[Effect]

The frame extraction unit 100 extracts discrete frames.

For example, #IF, #4F, #7F, and #10F are extracted. Then, it is held in frame memories 3a and 3b. That is, at least two consecutive frames (#IF and #4F, 94F and #7F,
) are made to exist at the same time.

On the other hand, the frame that is subject to frame dropping, for example #
2F, #3F, #5F, #6F, . . . are held in frame memories 4a and 4b, respectively.

Then, under this condition, compare #IF and #2F,
Further, by comparing #2F and #4F, a motion vector #2■ is generated for a moving portion in #2F. Similarly, #IF and #3F are compared, and #3F and #3F are compared.
4F, a motion vector #3■ is generated for a moving portion in #3F.

The transmission path encoding unit 6 performs #4F as described below.

#2V, #3V, #7F, #5V, #6V, #10F・
It outputs on the transmission path in the order of...

〔Example〕

FIG. 2 shows the configuration of an embodiment of the transmitting section according to the present invention.

3 is an explanatory diagram of the operation of the transmitter, FIG. 4 is an explanatory diagram illustrating the concept of detecting a motion vector, FIG. 5 is an example configuration of the receiver, and FIG. 6 is an explanatory diagram of the operation of the receiver, FIG. 7 shows the configuration of the motion compensation processing section, and FIGS. 8 and 9 each show the internal configuration of the motion compensation processing section.

Figure 2 shows the configuration of the transmitter, and the reference numeral 100 in Figure 1 shows the configuration of the transmitter.
．． 3a, 3b, 4a, 4b, 5. -6 corresponds to Fig. 1, 1 is an intermittent frame coding unit (CODI),
2 is a decoding unit (DECI).

7 is a buffer memory, 5W-Al, 5W-Bl.

5W-C1 and 5W-DI represent changeover switches, respectively.

Regarding the discrete frames #IF, #4F..., the encoding unit l performs normal image encoding (for example, Δ
encoding) to produce encoded output data. This data is led to switch 5W-Dl,
It is also returned to a decoded output by the decoder 2. Then, regarding the decoded output, the frame memory 3a
Frame #IF is held above and frame memory 3b
Frame #4F is held above.

On the other hand, frames #2F and #3F that are subject to frame dropping
are held in frame memories 4a and 4b, respectively. Then, as described with reference to FIG. 1, motion vector #2■ corresponding to frame #2F and motion vector #3■ corresponding to frame #3F are generated in the motion compensation processing unit 5, and the switch 5W- Guided to D1. Then, the transmission path encoding unit 6. The signal is sent out through the buffer memory 7 onto the transmission path.

The operation during this time is clarified in FIG. The numbers in the frames in the figure correspond to frame numbers. Third
As you can see from the figure, frame #4F.

Regarding #7F, the encoded output data is immediately inputted to the transmission line encoder 6, so to speak. On the other hand, the motion vector #2■ for frame #2F is stored in the frame memory 3a.
and 3b, and input to the transmission path encoding unit 6 at that point. The same applies to motion vector #3■ for frame #3F and motion vector #5■ for frame #5F to 9.

For example, when extracting the motion vector in frame #2F, as shown in FIG.
Divide the frame into a plurality of blocks, and consider one block as the processing target block B at the relevant point in time, (ii)
Determine "reference area 1" as a range in which the block B on the (n+1)th frame can be considered to have existed on the nth frame + (iii) The block B on the (n+1)th frame (n+3) Determine "reference area 2" as a range that may exist on the frame, (iv) Compare each block R and block B in the reference area 1, and refer to the above Each block R and block B in the ridge area 2 are compared, and (V) the block R that is most similar to block B is determined, and the information indicating the position of the block R is a vector for block B. It is considered information. The same processing as for the processing target block B is performed on all blocks on the (n+1)th frame, and the vector information generated for each is put together to form a motion vector # (n+1) corresponding to the (n+1)th frame. It is considered to be V.

FIG. 5 shows the configuration of the receiving section, where 8 is a buffer memory;
9 is a transmission path decoding unit, 10 is a motion compensation processing unit, 11 is a decoding unit, 12a and 12b are frame memories, respectively;
5W-A2, 5W-B2 and 5W-C2 represent switches, respectively. As can be seen from the operation of the receiving section shown in FIG. 6, frames #4F, #7F, . is output.

Further, frames #5F and #6F are reproduced while referring to the contents of the frame memories 12a and 12b.

FIG. 7 shows the configuration of the motion compensation processing section shown in FIG. 1 or 2, in which reference numeral 13 represents a difference value accumulation circuit, 14 represents a comparison circuit stage, and 15 represents a comparison circuit.

As shown in FIG. 2, the difference value accumulating circuit 13 compares, for example, block R in "reference area 1" and control block B (see FIG. 4), and outputs one of the output signals. That is, the difference between the pixel Ri on the block R and the pixel Bi on the block B is calculated, and the accumulated result for all pixels n in the block is output as one of the output signals (at this time, the result corresponding to the signal (The position information of block R is obtained.) Among the results of taking the differences Σ1Bi-R4l between all the blocks R and the control block B in "reference area 1" and "reference area 2", the one having the smallest value is extracted in the comparator circuit stage 14. The extracted thing expresses the first order matter. That is, block B moves from the position of block R on the (i) n-th frame to the (i)
n+1) has come to the position of the current block B on the frame, or (ii) has moved from the position of the current block B on the (n+1)th frame to the position of the block R on the (n+3)th frame. It depends on what you did. It expresses that.

FIG. 8 shows the configuration of the difference value accumulation circuit shown in FIG. Reference numerals 13-1 and 13-2 in the figure are reference data generating units, respectively, and output the above-mentioned pixels Ri, respectively. 13-3
．． 13-4 are subtracters, respectively.

13-5, 13-6 are absolute value generating circuits 513-7, respectively.
13-8 are adders, 13-9, 13-10 are flip-flops, and 13-11, 13-12 are AND circuits.

The outputs of flip-flops 13-9 and 13-10 respectively represent ΣlB1-R11 as described above. and “reference area 1” and “reference area 2”
Suppose there are m blocks within.

m cumulative difference values are obtained.

FIG. 9 shows the configuration of the comparison circuit. Reference numeral 15-1 represents a comparator, and 15-2 represents a selector. Comparator 15
-1 is supplied with two cumulative difference values and selects the smaller one. Then, the selector 15-2 selects the smaller accumulated difference value.

8) Extract the identification information as to which block R corresponds to Sadai Nao.

〔Effect of the invention〕

As described above, according to the second aspect of the present invention, motion vectors for frames in which frames have been dropped are transmitted.

For this reason, unnaturalness in moving parts can be largely eliminated compared to conventional methods that perform interpolation.

In addition, the information ■ when normal encoding is performed for l frames is P bits/screen, the amount of information when transmitting a motion vector is V bits/screen, and the number of frames when normal coding is performed per unit time is n9 + the number of frames in which motion vectors are transmitted per unit time is nV + In addition, it is assumed that m frames are transmitted per unit time using the frame dropping method in the conventional case.1 If the maximum transmission is performed in both the case of the present invention and the conventional case. Then.

P-n, +V・nv = P-m ■ np+□・nv = m Therefore, since V/P is less than 1.

H,+n,>m, and the number of frames that can be transmitted per unit time is larger than in the case of conventional frame dropping.

Furthermore, since motion vectors are transmitted, there is little deterioration in the quality of one image.

[Brief explanation of drawings]

FIG. 1 is a diagram showing the principle configuration of the present invention, FIG. 2 is a configuration diagram of an embodiment of the transmitting section related to the present invention, and FIG. 3 is an explanatory diagram of the operation of the transmitting section.
FIG. 4 is an explanatory diagram explaining the concept of detecting a motion vector, FIG. 5 is an example configuration of the receiving section, FIG. 6 is an explanatory diagram of the operation of the receiving section, and FIG. 7 is the configuration of the motion compensation processing section. FIG. 8 and FIG. 9 each show the internal configuration of the motion compensation processing section. In the figure, 100 is a frame extraction unit, and 1 is an encoding unit. 3.4 is a frame memory, and 5 is a motion compensation processing unit (
(transmission), 6 represents a transmission path encoding unit.

Claims (1)

[Claims] Regarding an input image composed of a plurality of frames,
A video encoding processing method that performs band compression by performing image encoding on discrete frames and transmits a video image, comprising: a discrete frame extraction unit (100) for extracting the discrete frames; frame memory (3a, 3
b), frame memories (4a, 4b) that hold the respective frames, and frame memories (3a, 4b) that hold the respective frames, and frame memories (3a, 4b) that hold the respective frames located between the above-mentioned discontinuous frames; , 3b), and a frame memory (4a or 4b) that holds frames located between the above-mentioned discrete frames.
), the motion compensation processing unit (5) generates a motion vector corresponding to a frame located between the discrete frames, and the discrete frame obtained by the discrete frame extraction unit (100) is A video encoding processing method characterized in that the motion vector is encoded and transmitted, and the motion vector is encoded and transmitted.