JPH0410788A - Method for controlling coding variable of image signal - Google Patents

Abstract

PURPOSE:To reduce the variation of a coding variable due to the characteristics of an image by using the displacement of an image whose forecasting image is to be generated from a reference image to control the coding variable of an image signal. CONSTITUTION:A reference image is set up in each frame interval or field interval of an image signal and a forecasting image for an image to be formed between the reference images is generated based upon the reference images. In this case, the coding variable of an image signal is controlled by using the displacement of the image whose forecasting image is to be generated from the reference image. Namely, the forecasting image of each inter-data is generated by a movement compensation forecasting device 4 based upon intra-data. Consequently, the coding variable can be changed in accordance with a sudden change in the image or the change of scenes and the variation of the coding variable due to the characteristics of the image can be reduced.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to, for example, a method for encoding a color video signal, and particularly to a code amount control method during interframe encoding for prediction errors of a video signal.

[Conventional technology]

When recording image data on a recording medium such as a magnetic disk, the data is compressed for efficient recording. For this compression, for example, the image data is divided into blocks each having NXN (or NXM) pixels, and each block is orthogonally transformed. The orthogonally transformed data is further quantized at a predetermined quantization step and then zero-run length encoded or Huffman encoded. Compressing data in this way effectively compresses the data, but
The amount of code varies depending on the image.

Therefore, conventionally, control has been performed to keep the code amount constant as follows.

The first method is to calculate the actual quantized data in a predetermined adoption step, and then adjust the number of quantization steps according to the calculation result so that the amount of data becomes the desired value. This method is to change it and redo the quantization.

The second method focuses on the fact that the coefficients of the data after orthogonal transformation have a predetermined relationship with the code amount, and calculates the sum of squares of the coefficients for each block. Each block is divided into, for example, four classes according to its size, and blocks with large amounts of data are allocated more bits, and blocks with smaller classes are allocated fewer bits. It is something.

[Problem to be solved by the invention]

However, the first method described in "-" requires repeating the process of actually calculating the code amount of quantized data at least twice, making it difficult to perform high-speed processing for moving images. .

In addition, the second method described above not only requires performing orthogonal transformation processing, but also requires adding information indicating the class, which increases the amount of code and requires complicated processing. become.

Furthermore, although methods 1 and 2 allow for improved accuracy by using past information, such methods do not allow for changes in image characteristics that are constant or gradual. This is effective in some cases, but has the opposite effect in cases where the image changes rapidly or when there is a scene change.

SUMMARY OF THE INVENTION An object of the present invention is to provide an image signal code amount control method that can reduce code amount fluctuations when an image changes rapidly or when a scene changes.

[Means to solve the problem]

The image signal code amount control method of the present invention sets a reference image for each of a plurality of frames of an image signal or for each of a plurality of field intervals, and generates a predicted image based on the reference image for images between the reference images. The encoding method is characterized in that the amount of code of an image signal is controlled using the displacement of an image for which a predicted image is to be generated with respect to a reference image.

[Effect]

In the image signal code amount control method of the present invention described above, the code amount of the image signal is controlled using the displacement of the image for which a predicted image should not be generated with respect to the reference image, so that it is possible to control the code amount of the image signal in response to sudden changes in images or scene changes. The amount of code can be changed by changing the amount of code, and fluctuations in the amount of code due to image characteristics can be reduced.

〔Example〕

FIG. 1 shows an example of the configuration of an apparatus used to implement the image signal code amount control method of the present invention. Input image data, which is a moving image signal, is stored in the frame memory 2. In the input image data, reference images are set at intervals of multiple frames. A reference image (hereinafter referred to as an in-liner) is an image encoded within a frame, and is supplied directly from the frame memory 2 to the encoder predictor 4. Image between two reference images of human image data (hereinafter referred to as inter)
is supplied from the frame memory 2 to the motion compensator 1tll unit 4. FIG. 2 shows an example of the configuration of in1-ra and inter. In the motion compensation predictor 4, a predicted image is generated for each inter based on in1-1.

(At this time, the base Y (g image or the decoded image of its encoded data) immediately before or after the inter can be used as the in-I-ra. To be more specific, the inter and the in-I-ra are , both 8x8 pixels or i6
The image is divided into blocks of X 1 G pixels, the displacement of the inter block with respect to the intra block (hereinafter referred to as motion vector) is detected, and the intra block is displaced by the detected motion vector to generate a predicted image. . The motion compensator side fence 4 also outputs the motion vector I-le for each block.

As shown in FIG. 3, the difference MV (motion vector I) absolute value sum calculation circuit 6 is obtained by adding the difference in the magnitude of motion vectors between adjacent blocks in the horizontal direction by 1 norm. The horizontal total and the vertical total, which is obtained by adding one frame of the difference in the size of motion background between vertically adjacent blocks, are calculated, and the screen aspect ratio is added to these two totals. Weighting based on 1-ratio J9 and resolution outputs the sum of the values. (However, in the case of normal images such as NTSC, weighting is not necessary.) Using the difference in the size of motion vectors between adjacent blocks, for example, panning where the entire screen is moving in a fixed direction It is possible to reliably detect sudden changes in images and scene changes in the state.

As shown in FIG. 2, the difference MV absolute value sum calculation circuit 6 further uses the vectors 1, V2 and 3 of the inter 2 and 3 for the intra, and calculates the difference motion vector MVI between adjacent frames, MV2 and MV2 are obtained by the following calculations, and the sum of the magnitudes (absolute values) of these differential motion vectors is output: MV1=VI MV2=V2-VI MV3=V3-V2.

The scene change circuit 8 is a difference MV absolute value sum calculation circuit 6
When the output of is larger than a preset value, a scene change signal is output to CPU10. Upon receiving the scene change signal, CPU10 sets the target code amount value of the image to be changed to a lower value by, for example, 20%, and outputs this set value to the code amount predictor 14 via the control data value transmission circuit ]-2. do. As a result, even if there is a sudden change in the image such as a scene change, the fluctuation in the amount of code can be reduced and overflow can be avoided.

The differential image data generator 15 outputs intra-to-inter differential data to the code amount predictor 4 based on the motion vector output from the motion compensation predictor 4 . The code amount predictor 14 predicts the input intra and inter image data, and uses the band limit value and quantization necessary to make this data amount the target code amount value sent from the control data value transmitting circuit 12. Filter circuit 1 for each step
6 and output to a quantizer 20.

The filter circuit 16 filters the input image according to the band limit value given from the code amount predictor 14. Further, the quantizer 20 quantizes the data that has been orthogonally transformed in the DCT circuit 18 according to a quantization step. The encoder 22 compresses and outputs the output data of the quantizer 20 by zero-rank encoding or Huffman encoding, and also feeds back the amount of code actually encoded to the CPU10.

Note that in the first embodiment, the reference image is set at every interval of a plurality of frames, but the reference image may be set at every interval of a plurality of fields. In this case, the frame memory shown in FIG. 11 may be replaced with a field memory.

〔Effect of the invention〕

As is clear from the above description, according to the image signal code amount control method of the present invention, the code amount can be changed in response to sudden changes in images or scene changes, and the code amount can vary depending on image characteristics. can be made smaller.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an example of the configuration of a device used to implement the image signal code amount control method of the present invention. FIG. 2 is an explanatory diagram showing the relationship between intra and inter. is an explanatory diagram showing a horizontal direction differential motion vector and a vertical direction differential motion vector between adjacent blocks of an image. 4... Motion compensation predictor, 6... Difference MV absolute value sum calculation circuit, 8... Scene change detection circuit, 10... CP
U, 12... Control data value transmission circuit, 14... Code amount predictor, 15... Differential image data generator. Patent applicant: Victor Japan Co., Ltd. Agent: Patent attorney: Yuichi Yoshi Inagi

Claims (1)

[Scope of Claims] In an image signal encoding method, a reference image is set for each plurality of frame intervals or each plurality of field intervals of an image signal, and a predicted image is generated based on the reference image for images between the reference images, comprising the steps of: An image signal code amount control method, characterized in that the code amount of the image signal is controlled using a displacement of an image for which a predicted image is to be generated with respect to the reference image.