JPH04287486A - In-frame coding system - Google Patents

Abstract

PURPOSE:To prevent a picture change at in-frame coding from being perceived with respect to the coding system of a moving picture. CONSTITUTION:A coding section 1 implements in-frame coding and preceding in-frame coding information is stored to a lst memory 2. A preceding original picture is stored in a background memory 3 and an error detection section 4 detects an error between a preceding original picture and a present original picture. A movement compensation section 5 obtains a motion vector from the movement caused in the present original picture and the information of magnitude of a moving vector from the preceding in-frame coding till the present in-frame coding or valid/invalid information is stored in a 2nd memory 6. A discrimination section 7 discriminates that any of motion vectors in the 2nd memory 6 is valid or not zero or invalid and a detected error exceeds a prescribed threshold level and when the discrimination is implemented, the present in-frame coding result is selected and when no discrimination is implemented, the preceding coding information is selected.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a video encoding system, and more particularly to an intraframe encoding system when there are a plurality of frames to be intraframe encoded.

[0002] In a high-efficiency video encoding system, intra-frame encoding is performed for encoding video images for storage, and inter-frame encoding is performed for encoding video images for communication.

[0003] In such intra-frame encoding for storage video encoding, etc., when there are multiple frames to be intra-frame encoded, each time intra-frame encoding is performed, a change in image is detected. It is requested that this is not done.

0004

DESCRIPTION OF THE PRIOR ART FIG. 7 shows a conventional encoding circuit, in which 101 is a counter, 102 is a selection section, and 103
104 is a quantization unit (Q) that performs quantization, 105 is a motion compensation unit that performs motion compensation, 106 is a variable delay unit, and 107 is a frame memory (107) that stores images for one frame. FM), 108 is a VLC unit that performs variable length encoding, 109 is an inverse quantization unit (*Q) that performs inverse quantization, and 110 is an inverse DCT that performs inverse discrete cosine transformation.
section (*DCT), 111 is a subtraction section, 112 is an addition section, 1
13 and 114 are switches.

[0005] The selection section 102 selects the original picture or the subtraction section 111.
Select the difference from. When an original image is selected, the information of the original image for one frame is subjected to discrete cosine transform in the DCT unit 103 to compress the amount of information, quantized in the quantization unit 104, and variable length encoded in the VLC unit 108. Intraframe encoding is thereby performed to produce a coded output.

On the other hand, in the selection section 102, the subtraction section 111
In the state where the difference from is selected, the input original image is subjected to motion compensation in the motion compensation unit 105, and a motion vector is extracted. The extracted motion vector is input to the variable delay section 106, and is also input to the VLC section 108.
is input. On the other hand, the previous frame screen output from the frame memory 107 is delayed in the variable delay section 106 according to the motion vector from the motion compensation section 105 to generate a predicted value signal. In the subtraction unit 111,
A difference (prediction error) is generated by subtracting this predicted value signal from the original image. The difference signal is sent to the DCT section 10
3, and is quantized in quantization section 104 to produce a quantized difference (prediction error), which is input to VLC section 108. The VLC unit 108 performs interframe coding by variable-length coding the quantized difference and the motion vector to generate a coded output. The quantized difference from the quantization unit 104 is dequantized in the dequantization unit 109, and
An inverse discrete cosine transform is performed in a T section 110, and a predicted value signal is added in an adder 112 via a switch 114 to generate a reproduced signal. This reproduced signal is stored in the frame memory 107, thereby generating the signal of the screen of the previous frame described above. Note that the DCT section 103, the quantization section 104, and the inverse quantization section 1
09. Regarding the inverse DCT unit 110, other encoding methods (
For example, vector quantization) may be performed.

[0007] The selection unit 102 compares the original image with the difference from the subtraction unit 111 and selects the one that can be estimated to have less information, and when the amount of information in the original image is large, the intra-frame code is selected. If there is a large amount of difference information, interframe coding is performed. In addition, when the intra-frame encoding (Y/N) signal that instructs intra-frame encoding is Y, which instructs intra-frame encoding, intra-frame encoding is forcibly performed; otherwise, when it is N, the intra-frame encoding is Performs intercoding. Furthermore, it has a counter 101 whose counter value changes every frame, and performs intraframe encoding every fixed counter value. When performing interframe encoding, the selection unit 102 controls the switch 113 to input the difference output of the subtraction unit 111 to the DCT unit 103, and controls the switch 114 to input the predicted value signal to the addition unit 112. .

[0008] In this way, the information generated by interframe coding is transmitted in real time and used for the purpose of communication for moving image transmission. On the other hand, information generated by performing intraframe encoding is used as an initial value when performing interframe encoding, and is also used for the purpose of storage to enable random access or playback from the middle. In a high-efficiency coding method for moving images, a sequence is adopted in which the above-described communication coding is performed for each frame, and storage coding is performed, for example, once every several frames.

[0009]

[Problem to be Solved by the Invention] In a high-efficiency encoding system for moving images, when encoding moving images for storage at a rate of once every several frames, for example, if the camera is fixed,
When inputting an image with little movement, even though parts such as the background have not moved since the previous intra-frame encoding, the next time intra-frame encoding is performed, lighting flickers, Due to noise in the camera's image pickup tube, differences in quantizers, etc., an encoding result that is different from the previous intraframe encoding may occur. in this case,
For example, due to differences in background brightness, etc., each time intra-frame encoding is performed, it is sensed that the image has been switched, and this is evaluated as a deterioration in image quality.

[0010] The present invention is an attempt to solve the problems of the prior art as described above.
A frame that prevents the perception of image switching due to changes in stationary parts such as the background when intra-frame encoding, which encodes all image information in one frame, is performed over multiple frames. The purpose is to provide an inner encoding method.

[0011]

[Means for Solving the Problems] The present invention provides an encoding unit that performs intra-frame encoding, a first memory that stores the previous encoding information in the encoding unit, and a first memory that stores the previous encoding information in the encoding unit. A background memory that stores the original image at the time, an error detection unit that performs processing for each block to detect errors between the original image stored in the background memory and the original image at the time of intra-frame encoding, and an error detection unit that detects errors that occurred in the current original image. A motion compensation unit that detects motion and obtains a motion vector, and a second unit that stores information about the size of the motion vector or whether the motion vector is valid or invalid from the previous intraframe encoding to the current intraframe encoding. A process is performed for each block to determine whether any of the motion vectors in the memory and the second memory is valid or not 0, and if not, the error detected by the error detection unit exceeds a predetermined threshold. When this determination is made, the intraframe encoding result in the encoding section is selected as the encoding information of the block in the frame, and when this determination is not made, the first This method is characterized in that the encoded information stored in the memory is selected as the encoded information for the block in the frame.

0012

[Operation] FIG. 1 shows the basic structure of the present invention. The encoding unit 1 performs intraframe encoding for the entire one screen. Previously encoded information in the encoder 1 is stored in the first memory 2. The background memory 3 stores the original image from the previous encoding in the encoding unit 1, and the error detection unit 4 calculates the error between the original image stored in the background memory 3 and the original image from the current intra-frame encoding. Perform detection processing for each block. On the other hand, the motion compensation unit 5
, the motion that occurred in the current original image is detected to obtain a motion vector, and the size of the motion vector or the motion vector from the previous intra-frame encoding to the current intra-frame encoding is stored in the second memory 6. Stores valid/invalid information. Then, in the determination unit 7, the second memory 6
Performs processing for each block to determine whether any of the motion vectors stored in the motion vector is valid or not 0, and if not, the error detected by the error detection unit 4 exceeds a predetermined threshold; In the selection unit 8, when this determination is made, the encoding unit 1 selects the result of newly performing intra-frame encoding using the current frame as the encoding information of the block in question in the frame;
If this determination is not made, the encoded information stored in the first memory 2 is selected as the encoded information of the block in the frame, thereby generating an encoded output.

In this case, the error detection unit 4 performs a process of detecting an error between the original image stored in the background memory 3 and the original image at the time of intra-frame encoding for each macroblock consisting of a plurality of blocks, and makes a judgment. In the unit 7, a process of determining whether any of the motion vectors in the second memory 6 is valid or not 0, and if not, whether the error detected in the error detection unit 4 exceeds a predetermined threshold It is also possible to perform this for each macroblock.

[0014] The error detection unit 4 can detect errors by calculating the average value of absolute errors between the original image stored in the background memory 3 and the original image during the current intra-frame encoding. However, error detection can also be performed by other methods.

[0015]

[Embodiment] FIG. 2 shows the configuration of an embodiment of the present invention, in which the same parts as in FIG. 123 is an error detection unit that detects the error between the output of the background memory 121 and the original image; 124 is a valid/invalid state of motion vectors and pixel blocks from the previous intra-frame encoded frame to the next intra-frame encoded frame; 125 is a determination unit that determines whether to use the previous intraframe encoded information or the current intraframe encoded information, and 126 and 127 are switches.

In FIG. 2, the parts indicated by numbers 101 to 114 are the same as the conventional example shown in FIG.
Since the operation is similar, the explanation will be omitted. The memory 121 stores the original image of the frame subjected to intra-frame encoding, and the memory 122 stores the encoding information at that time, and stores it until the next intra-frame encoding. Further, the memory 124 stores information on whether motion vectors and pixel blocks are valid or invalid between the frame on which intraframe coding was previously performed and the frame on which intraframe coding was performed next.

The error detection unit 123 compares the original image of the previous intra-frame encoding in the background memory 121 with the original image of the current frame to detect an error. Based on this error value, motion vector information (motion compensation information) in the memory 124, and image block validity/invalidity information, the determination unit 125 determines whether to use the current intraframe encoding information or not.
Alternatively, it is determined whether to use the previous intraframe encoding information. Switches 126 and 127 select the output of the quantizer 104 when the current intraframe encoding information should be used, and select the output of the quantizer 104 when the previous intraframe encoding information should be used, according to the determination result of the determination unit 125. selects the output of memory 122. Such selection is performed for each pixel block making up the frame.

In the embodiment shown in FIG. 2, during intra-frame encoding, for parts that do not move, such as background parts, when the difference between the original image of the previous frame and the original image of the current frame is small, the difference between the previous frame and the original image of the current frame is When the motion vector with respect to the frame is small, or when the block in question in the previous frame is invalid, variable-length encoding is performed using the previous encoding information to generate an output, and the previous encoding information is also used. Since the playback signal is created by using the image data, there is no change in the image information for portions where there is no movement. Therefore, the switching of images will not be detected.

FIG. 3 is a flowchart showing an example of the operation of the error detection section and determination section. Let F1(X,Y) be the original image data that was last intra-frame encoded, and F2(X, Y) be the original image data that is currently being intra-frame encoded.
, Y). Here, X and Y are the vertical and horizontal sizes (number of pixels) of the image. In addition, the threshold value for determination in the determination unit is TH, and the minimum unit pixel number for processing is b(pel
) (step 301). The minimum unit for selecting encoded data is usually a block, but it may also be a macroblock that is a collection of several blocks. In FIG. 3, the determination is made using a block of b=B×B (pixels) as the minimum unit of processing. For example, CCITT H.
In H.261, b=8×8 is specified.

First, the error is calculated (step 302A).
). Step 302A indicates that the error is calculated by calculating and accumulating the average value of the absolute value errors. Next, the motion vector and valid/invalid information are checked (step 303). VX1(u,v),VY
1(u,v),...,VXN(u,v),VYN(u,v
) is the value of the motion vector from the previous intra-frame encoding frame to the current intra-frame encoding frame, and u and v are the numbers of vectors in the X and Y directions. .

Next, a determination is made (step 3).
04). Vector VX1(
i, j), VY1 (i, j), ..., VXN (i, j),
Check whether VYN (i, j) is 0 or not and whether the vector is valid or invalid. If any of the vectors is not 0 or is valid, step 3
If not, proceed to step 305. Note that a valid vector here refers to a case where the motion vector is too large to be detected, and an invalid vector refers to a case where the motion vector is too small to be detected. .

[0022] In step 305, the error obtained in step 302A is compared with the threshold value TH, and if the error is less than the threshold value, the information of the previous intra-frame encoded frame is used (step 306). , otherwise, proceed to step 307. In step 306, the encoding information of the previous intra-frame encoded frame is used, and in step 307, new intra-frame encoding is performed using only the current frame. Such processing is i=1~u,j=1
~v is done for all blocks (step 3
10).

FIG. 4 shows another method of calculating the error in the operation example of FIG. 3, and shows several other methods that can be used in place of step 302A. (a) calculates the squared error (step 302B
) shows a method, (b) shows a method of calculating the value at the point where the absolute value of the error is the maximum value as an error (step 302C), and (c) shows a method of calculating the number of errors exceeding a certain threshold value. The method of calculating the error (step 302D) is shown.

FIG. 5 is a flowchart showing another example of the operation of the error detection section and determination section. In FIG. 5, steps 401 to 405 are almost the same as steps 301 to 305 in the operation example of FIG. 3, but in the case of FIG. The difference is that it is made into a macroblock that is a collection of several macroblocks. In the operation example shown in FIG. 5, determination is made with the unit of processing as B×B×A (pixel). For example, CCITT H. 2
In 61, B=8, A=4 (Y signal only) is specified. Such processing is performed for all macroblocks of i=1 to u and j=1 to v (step 410).
). In each step, S and T represent each block in the X direction and Y direction in the macroblock, and S=0 to
N×B-1, T=0 to N×B-1.

In steps 404 and 405, it is shown that for each block, if at least one vector is non-zero or valid, the process proceeds to step 407; otherwise, the process proceeds to step 406. In step 406, the intraframe encoding information of the previous frame is used, and in step 407, new encoding is performed using only the current frame.

FIG. 6 shows another method of calculating the error in the operation example of FIG. 5, and shows several other methods that can be used in place of step 402A. (a) calculates the squared error (step 402B)
) shows a method, (b) shows a method of calculating the value at the point where the absolute value of the error is the maximum value as an error (step 402C), and (c) shows a method of calculating the number of errors exceeding a certain threshold value. The method of calculating the error (step 402D) is shown.

[0027]

[Effects of the Invention] As explained above, according to the present invention, 1
When intraframe encoding that encodes all image information in a frame is performed over multiple frames, if even one motion vector is not 0 or is valid,
Otherwise, if the error with the previous image exceeds a certain threshold, new encoding is performed using only the current frame; otherwise, the previous intra-frame encoding is performed. Since the same encoding information is used again as the encoding information for that block, it is possible to avoid detecting image switching due to changes that occur in stationary parts such as the background each time intra-frame encoding is performed. is prevented, and deterioration of image quality is reduced.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing the basic configuration of the present invention.

FIG. 2 is a diagram showing the configuration of an embodiment of the present invention.

FIG. 3 is a flowchart illustrating an example of the operation of an error detection section and a determination section.

FIG. 4 is a diagram showing another method of calculating the error in the operation example of FIG. 3, in which (a) shows a method of calculating a squared error;
(b) shows a method of calculating the value at the location where the error is the maximum value as an error, and (c) shows a method of calculating the number of errors exceeding a certain threshold value to obtain the error.

FIG. 5 is a flowchart showing another example of the operation of the error detection section and the determination section.

6 is a diagram illustrating another method of error calculation in the operation example of FIG. 5, in which (a) is a method of calculating a squared error;
(b) shows a method of calculating the value at the location where the error is the maximum value as an error, and (c) shows a method of calculating the number of errors exceeding a certain threshold value to obtain the error.

FIG. 7 is a diagram showing a conventional encoding circuit.

[Explanation of symbols]

1 Encoding section 2 First memory 3 Background memory 4 Error detection section 5 Motion compensation section 6 Second memory 7 Judgment section 8 Selection section

Claims (3)

[Claims] Claim 1: An encoding unit (1
), a first memory (2) that stores the previous encoded information in the encoder (1), and a background memory (3) that stores the original image that was encoded last time in the encoder (1). ), an error detection unit (4) that performs processing for detecting errors between the original image stored in the background memory (3) and the original image at the time of intra-frame encoding this time, for each block; Motion compensation unit (which detects motion and obtains a motion vector)
5), a second memory (6) that stores information on the size of the motion vector or the validity/invalidity of the motion vector from the time of the previous intraframe encoding to the time of the current intraframe encoding; a determination unit (7) that performs processing for each block to determine whether any of the motion vectors in the memory (6) is valid or not 0, and if not, the detected error exceeds a predetermined threshold; ), when the determination is made, the intraframe encoding result in the encoding unit (1) is selected as the encoding information of the block of the frame, and when the determination is not made, the intraframe encoding result is selected as the encoding information of the block of the frame. An intra-frame encoding method characterized in that encoded information stored in a memory (2) of a frame is selected as encoded information of a corresponding block of a corresponding frame. 2. The error detection unit (4) performs a process of detecting an error between the original image stored in the background memory (3) and the original image during the current intra-frame encoding using a macroblock consisting of a plurality of blocks. The determination unit (7)
A process of determining whether any of the motion vectors in the second memory (6) is valid or not 0 and, if not, that the detected error exceeds a predetermined threshold, is carried out for each macroblock. 2. The intraframe encoding method according to claim 1, wherein the intraframe encoding method is performed in the following manner. 3. Error detection in the error detection unit (4) calculates an average value of absolute value errors between the original image stored in the background memory (3) and the original image during the current intra-frame encoding. 3. The intraframe encoding method according to claim 1, wherein the intraframe encoding method is performed by: