JP2002204459A - Moving picture coding method and device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a moving picture coder that can prevent information in a pre-designated spatial area like a character area from being deteriorated. SOLUTION: In the image coder that an orthogonal transform device 16 applies orthogonal transform to an input moving picture signal 4 or a prediction error signal outputted from a sbutractor 13 in the unit of blocks to generate a transform coefficient, a quantizer 17 quantizes the transform coefficient, a variable length coder 25 codes the quantized transform coefficient and outputs a coded bit stream 7, a coding control circuit 30 receives an area designation signal 4 denoting a specific spatial area designated for the input moving picture signal 4, a spatial area discrimination section 31 discriminates a specific block including the spatial area denoted by the area designation signal 5 among the blocks of the input moving picture signal and a quantization control section 34 stepwise updates the quantization coefficient until the quantization coefficient reaches a prescribed threshold value or below when the quantization coefficient corresponding to the specific block is greater than the prescribed threshold value.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a moving picture coding method and apparatus suitable for a moving picture including a character area requiring sharpness.

[0002]

2. Description of the Related Art In a moving picture coding system represented by the MPEG system, orthogonal transformations such as motion compensation inter-frame prediction for removing temporal redundancy, discrete cosine transform (DCT) for reducing spatial redundancy, and the like. A moving image signal is compressed with high efficiency by combining quantization and variable length coding.

For example, in the MPEG system, intra-frame encoding (intra-encoding) and inter-frame encoding (inter-encoding) are prepared as encoding modes. In the intra-frame encoding mode, an input video signal, an inter-frame encoding, In the mode, a prediction error signal which is an error of a prediction signal with respect to an input moving image signal is subjected to discrete cosine transform in block units. DCT coefficients generated by the discrete cosine transform are quantized, and the quantized DCT coefficients are subjected to variable length coding. The variable length coded quantized DCT coefficients;
Similarly, the quantized coefficients, the mode information of the coding, and the information of the motion vector, which have been subjected to the variable length coding, are output as a coded bit string. The quantized DCT coefficient is further dequantized by the same quantization coefficient as that used at the time of quantization, and further subjected to an inverse discrete cosine transform, thereby being written into a frame memory as a locally decoded signal.

In the inter-frame coding mode, a motion vector to a rectangular area having the highest correlation with the local decoded signal of the previous frame stored in the frame memory in macroblock units is obtained. According to this motion vector, the pixel value (prediction signal) of the rectangular area specified by the motion vector of the local decoded signal is subtracted from each pixel value of the encoding target block of the input video signal, and a prediction error signal is generated. .

In the moving picture coding apparatus as described above,
Since the quantization of the DCT coefficient is performed by dividing the DCT coefficient by a value obtained by multiplying the quantization coefficient by a predetermined constant, the larger the quantization coefficient is, the coarser the quantization is, and the higher-frequency components having low energy tend to be lost. Become. The quantization coefficient is controlled so that the coded bit string generated so far approaches a predetermined code amount. That is, when the generated code amount is larger than a predetermined value, the quantization is made coarse to reduce the code amount of the coded bit string, and when the generated code amount is smaller than the predetermined value, the quantization is made fine. Then, control is performed so as to increase the generated code amount of the coded bit string.

[0006]

In the conventional moving picture coding apparatus described above, the quantization coefficient is controlled so that the generated code amount approaches a predetermined value. For example, character information is included in the input moving picture signal. Even in such a case, a region including a character may be coarsely quantized. For this reason, especially when the bit rate of the coded bit string is low, it becomes impossible to determine the character information or it becomes difficult to determine the character information.

Further, there is a method in which a still region such as a character is extracted from an input moving image signal, and the quantization of the still region is made finer than the quantization of other regions. Is that pixel information in areas other than characters is mixed around the block due to motion compensation,
There is a problem in that the amount of generated codes increases due to the fine quantization, and the overall image quality deteriorates.

[0008] The present invention solves the above problems,
It is an object of the present invention to provide a moving picture coding method and apparatus capable of preventing deterioration of information in a predetermined space area such as a character area.

[0009]

SUMMARY OF THE INVENTION In order to solve the above-mentioned problems, the present invention orthogonally transforms an input video signal or a prediction error signal of an input video signal with respect to a prediction signal in block units to generate a transform coefficient. In a moving picture coding method of performing variable length coding after quantizing the transform coefficient based on a predetermined quantization coefficient, in a block such as a character area where sharpness is required in each block of the input moving picture signal, When the quantization coefficient corresponding to the specific block including the designated space area is larger than a predetermined threshold, the quantization coefficient is updated stepwise until the quantization coefficient becomes equal to or smaller than the threshold. The update of the quantization coefficient corresponding to the specific block is performed, for example, by decreasing the quantization coefficient by a predetermined amount at predetermined time intervals.

As described above, in the present invention, a character area or the like in a moving image where sharpness is required is specified in advance, and the quantization coefficient in the specified area is independent of the quantization coefficients in other areas. By updating the quantization coefficient to be equal to or smaller than the threshold when the quantization coefficient is larger than the predetermined threshold, quality deterioration of character information or the like is reduced.

Further, in the present invention, a motion vector corresponding to a specific block whose quantization coefficient is updated in the inter-frame coding mode is forcibly set to a zero vector, and a motion vector whose quantization coefficient is not updated in the inter-frame coding mode is specified. A motion vector corresponding to a block is forcibly set to a zero vector, and a prediction error signal corresponding to the specific block is forcibly set to zero, that is, a specific block whose quantization coefficient is not updated is a non-encoding target block. It is desirable to do. By doing so, the amount of code generated in encoding the character area can be effectively reduced, and good image quality can be maintained in areas other than that area.

[0012]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 shows a configuration of a system using a video encoding device according to an embodiment of the present invention. When the moving image signal 1 and the character information generated from the character information generating unit 2 are input to the synthesizing unit 3, the character information is superimposed on the moving image signal 1, and the moving image signal 4 on which the character information is superimposed is a moving image. The image is encoded by the image encoding device 6.

The synthesizing unit 2 further outputs an area designation signal 5 indicating an area (character area) on which character information on the image is superimposed. The area specifying signal 5 is generated, for example, when the user specifies a character area on the image of the moving image signal 1 displayed on the image monitor. This area designation signal 5
Is input to the video encoding device 6. In the moving image encoding device 6, the input moving image signal 4 in which the character information is synthesized is compression-encoded by, for example, the MPEG method, and the encoded bit sequence 7 is output. The encoded bit sequence 7 is sent to a transmission system or storage system (not shown).

Here, in the moving picture coding apparatus 6,
An operation similar to, for example, the conventional MPEG method is performed on an encoding target block including a normal moving image or a still region in which sharpness is not required in the input moving image signal 1. In addition, when encoding a coding target block including a character area where sharpness is particularly required, the moving image encoding device 6 converts the quantized coefficient if the quantized coefficient is larger than a predetermined threshold. The operation of updating in a stepwise manner until the threshold value is reached is performed according to the area designation signal 5. By this operation,
It is possible to prevent image quality degradation in a character area where sharpness is required.

FIG. 2 shows an example of a character area 9 on the moving image display screen 8 where sharpness is required. For example, a telop area such as superimposed subtitles of a movie or characters of a news corresponds to this. The display position of the character area 9 is generally constant. Even if it is a character, for example, a character such as a signboard appearing as a part of a background or the like in an image is not particularly required to be sharp,
Here, it is not included in the character area 9 where sharpness is required. Since the position of the telop area is generally determined, the area specifying signal 5 described above can be automatically generated.

Next, a detailed configuration of the moving picture coding apparatus 6 in FIG. 1 will be described with reference to FIG. The moving picture coding device 6 is a coding device that combines motion compensation, orthogonal transform such as DCT, and variable length coding as represented by the MPEG method in the present embodiment. It has an inner coding mode and an inter-frame coding mode.

In FIG. 3, an input moving image signal 4 input in units of frames is written in a frame memory 11,
Under the control of the read control unit 12, the image data is divided into, for example, 16 × 16 pixel rectangular areas called macroblocks and read. In the input video signal 4, a macroblock to be currently encoded is referred to as an encoding target block. Hereinafter, the configuration of the video encoding device 6 and the normal case where the encoding target block of the input video signal 1 does not include the designated space area (designated area) indicated by the area designation signal 5 will be described. The basic operation will be described.

First, in the intra-frame encoding mode, the input of the switch 14 is switched to the input video signal side and the input of the switch 15 is switched to the output side of the switch 14 by the encoding control unit 30. Therefore, the signal of the encoding target block read from the frame memory 11 is sequentially transmitted to the orthogonal transformer 1 through the switches 14 and 15.
6 is directly input to a frequency domain transform coefficient (or orthogonal transform (for example, DCT)).
CT coefficients).

The transform coefficients generated by the orthogonal transformer 16 are input to the quantizer 17 and are quantized according to the quantized coefficients specified by the encoding control circuit 30. The quantized transform coefficient is input to a variable length encoder 25 using a Huffman code or the like as a code word, and is changed by assigning a code word predetermined in a code word table according to the value of the transform coefficient. Long encoded. The code words of the quantized transform coefficients are quantized coefficients or intra-frame encoded /
It is output as a coded bit string 7 together with mode information indicating a coding mode of inter-frame coding and a codeword indicating motion vector information.

The quantized transform coefficient output from the quantizer 17 is also input to the inverse quantizer 18 where the quantized
Is inversely quantized by the same quantization coefficient as. The transform coefficients after the inverse quantization are further converted into pixel values in the spatial domain by the inverse orthogonal transformer 19. At this time, the encoding control circuit 3
The input of the switch 21 is switched to the output side of the inverse orthogonal transformer 19 by 0, and the pixel value after the inverse orthogonal transform is written to the frame memory 22 via the switch 21 as a local decoded signal.

On the other hand, in the inter-frame encoding mode, a motion vector to a rectangular area having the highest correlation with the local decoded signal of the previous frame stored in the frame memory 22 is obtained by the motion estimator 23 for each macroblock. According to this motion vector, a pixel value of a rectangular area specified by the motion vector of the local decoded signal is generated as a prediction signal, and the prediction signal is subtracted from each pixel value of the encoding target block of the input moving image signal. , A prediction error signal is generated.

In this inter-frame encoding mode, the input of the switch 14 is switched to the output side of the subtractor 13 by the encoding control circuit 30, and the prediction error signal is transmitted through the switches 14 and 15 to the orthogonal transformer. 16 is input. Hereinafter, similarly to the case of the intra-frame encoding mode, the orthogonal coefficients output from the orthogonal transformer 16 are input to the variable-length encoder 25 via the quantizer 17 and become an encoded bit sequence 7.

Further, the output of the quantizer 17 is input to an adder 20 via an inverse quantizer 18 and an inverse orthogonal transformer 19. In the inter-frame encoding mode, the encoding control circuit 3
Since the input of the switch 21 is switched to the output side of the adder 20 by 0, the pixel value in the spatial domain output from the inverse orthogonal transformer 19 is calculated by the adder 20 according to the motion vector of the local decoded signal of the previous frame. The signal is added to the image signal of the rectangular area at the designated position, and the output of the adder 20 is written to the frame memory 22 as a local decoded signal.

Next, the encoding control circuit 30 will be described. The coding control circuit 30 includes a spatial area determination unit 31, a motion prediction control unit 32, a prediction error control unit 33, and a quantization control unit 3.
4 and a motion vector control unit 35.

The area specifying signal 5 input from the synthesizing section 3 of FIG. 1 to the encoding control circuit 30 is input to the spatial area determining section 31. The spatial area determination unit 31
Every time the signal of the encoding target block of the input moving image signal 4 is read from the frame memory 11, it is determined whether or not the encoding target block includes the designated area, that is, the designated spatial area indicated by the area designation signal 5. It discriminates and outputs an area determination signal when the area includes the specified area. This region determination signal is supplied to the motion prediction control unit 32 and the prediction error control unit 3
3. It is supplied to the quantization control unit 34 and the motion vector control unit 35.

When the block to be coded includes the specified area, the motion prediction controller 32 does not search for the motion vector, and the motion predictor 23 does not search for the motion vector from the frame memory 22 before the motion vector “0” is specified. It reads the local decoded signal of the frame and controls the motion estimator 23 so as to output it to the subtractor 13 and the adder 20.

When the block to be coded includes the designated area, the quantization control unit 34 determines the quantization coefficient output by itself when the signal of the block to be coded is read from the frame memory 11. (Comparison step). If the quantization coefficient is larger than the threshold value in this comparison step, the quantization coefficient is set in a stepwise manner at predetermined time intervals until the quantization coefficient becomes equal to or smaller than the threshold value, for example, for each of a plurality of predetermined frames of the input video signal 4. Is updated (update step).

That is, when the quantization coefficient is greater than the threshold value, the quantization control unit 34 determines that, when the same coding target block is encoded, when the quantization coefficient has passed a predetermined number of frames from the previously updated frame, A value obtained by subtracting a predetermined constant value from the current quantization coefficient is output as an updated new quantization coefficient. When the quantization control unit 34 updates the quantization coefficient, it outputs a quantization coefficient update signal to the prediction error control unit 33.

The prediction error control unit 33 receives the quantization coefficient update signal from the quantization control unit 34 while the signal of the current block including the designated area is being read from the frame memory 11. The input of the switch 15 is switched to the output side of the switch 14, and when the quantization coefficient update signal is not input, the input of the switch 14 is switched to the “0” side.

The motion vector control unit 35 sets the input of the switch 24 to "0" when the block to be coded includes the designated area, and switches the input of the switch 2 when the block does not include the designated area.
4 is switched to the motion vector output side of the motion estimator 23. The output of the switch 24 is supplied to the variable length encoder 25 as a motion vector input.

Next, the operation when the signal of the encoding target block including the designated area is read from the frame memory 11 will be described. (1) In the intra-frame encoding mode, when the signal of the encoding target block including the designated area is input, the input of the switch 14 is switched to the input video signal side by the encoding control circuit 30. Therefore, the input moving image signal 4
Are directly input to the orthogonal transformer 16 and are converted into frequency domain transform coefficients. The transform coefficient is input to the quantizer 17 and is quantized according to the above-described quantization coefficient when the current block includes the designated area. The quantized transform coefficients are variable-length coded by the variable-length encoder 25 together with the quantized coefficients and mode information, and output as a coded bit sequence 7.

The transform coefficient after quantization is also input to the inverse quantizer 18 and inversely quantized according to the same quantized coefficient as the quantizer 17. The coefficients after the inverse quantization are further converted into pixel values in the spatial domain by the inverse orthogonal transformer 19. At this time, the input of the switch 21 is switched to the output side of the inverse orthogonal transformer 19 by the encoding control circuit 30, and the pixel value obtained by the inverse orthogonal transform is transmitted to the frame memory 22 via the switch 21 and the local decoded signal Is written as

(2) In the inter-frame coding mode,
When the signal of the encoding target block including the designated area is input and the quantization coefficient corresponding to the encoding target block is not updated, the prediction error control unit 33 sets the input of the switch 15 to the “0” side. , And the outputs of the orthogonal transformer 16 and the quantizer 17 both become zero. Accordingly, the outputs of the inverse quantizer 18 and the inverse orthogonal transformer 19 are also zero, so that the same value as the local decoded signal of the previous frame is written in the frame memory 22. Further, at this time, the input of the switch 24 is switched to "0" by the motion vector control unit 35, and the variable length encoder 2
The motion vector input to 5 is forced to be a zero vector.

As described above, when the quantization coefficient corresponding to the encoding target block including the designated area is not updated in the inter-frame encoding mode, the output of the quantizer 19 is zero (zero coefficient) and the motion vector is also zero. The variable length encoder 25 assigns the shortest code word to the non-coding block that is the coding target block because the vector becomes a vector and the coding target block is a non-coding (NOT_CODED) block. That is, when the quantization coefficient smaller than the encoding target block that does not include the designated area is used, the shortest code word is output, so that an increase in the code amount is suppressed.

(3) In the inter-frame coding mode,
When the signal of the encoding target block including the designated area is input and the quantization coefficient corresponding to the encoding target block is updated, the motion prediction control unit 32 controls the motion estimator 23. As a result, the local decoded signal of the previous frame of the motion vector “0” is read out from the frame memory 22 by the motion predictor 23 as a predicted signal, and the predicted signal is subtracted from the input video signal 4 by the subtractor 13.

The prediction error signal output from the subtractor 13 is orthogonally transformed by the orthogonal transformer 16 via the switches 14 and 15, and the generated transform coefficients are quantized by the quantization control unit 34. It is quantized by the quantizer 17 according to the coefficient. The quantized transform coefficients are variable-length coded by the variable-length coder 25 and output as a coded bit sequence 7. The output of the quantizer 17 is further supplied to an adder 2 via an inverse quantizer 18 and an inverse orthogonal transformer 19.
Input to 0.

In the inter-frame coding mode, the input of the switch 21 is switched to the output side of the adder 20 by the coding control circuit 30, so that the output of the inverse orthogonal transformer 19 in the adder 20 is The signal is added to the image signal of the rectangular area at the position specified by the motion vector of the local decoded signal, and the output of the adder 20 is written to the frame memory 22 as the local decoded signal.

As described above, in the present embodiment, when encoding a signal of an encoding target block including a designated area such as a character area composed of a complex character requiring sharpness, the quantization coefficient is set to character information or the like. If the quantization coefficient is larger than a predetermined threshold value that can be displayed easily, it is updated stepwise at predetermined time intervals until the quantization coefficient reaches the threshold value.

Here, when a large code amount is generated in the encoding target block not including the designated region, the quantization coefficient is increased to suppress the generated code amount even in the encoding target block including the designated region. And may be coarsely quantized. However, according to the present embodiment, the quantization coefficient is gradually reduced in the designated area as described above. Therefore, even for an image including a character having a complicated shape or the like, a code having high visibility can be obtained while suppressing the generated code amount. Can be performed.

The above-described moving picture coding processing according to the present invention can be executed by software using a computer such as a personal computer, a workstation or a portable device. Therefore, according to the present invention, it is possible to provide the following program or a computer-readable recording medium storing the program.

(1) A program that has an intra-frame encoding mode and an inter-frame encoding mode, and is a program for causing a computer to execute a moving image encoding process for compressing and encoding an input moving image signal input in frame units. In the intra-frame encoding mode, the prediction error signal of the input video signal corresponding to a prediction signal obtained by performing motion prediction using a motion vector in the inter-frame encoding mode, A process of generating a transform coefficient by performing an orthogonal transform in block units, a process of quantizing the transform coefficient based on a predetermined quantized coefficient, a process of performing variable-length encoding of the quantized transform coefficient, and A quantization coefficient corresponding to a specific block including a pre-specified spatial area among the blocks of the image signal is compared with a predetermined threshold, and the quantization is performed. A program for causing a computer to execute a quantization coefficient update process of gradually updating the quantization coefficient until the number becomes equal to or less than the threshold when the number is larger than the threshold, or a computer-readable recording storing the program Medium.

(2) In (1), further, forcibly setting a motion vector corresponding to a specific block in which the quantization coefficient is updated in the inter-frame coding mode to a zero vector, and And forcing the motion vector corresponding to the specific block in which the quantization coefficient is not updated in the mode to a zero vector and forcing the prediction error signal corresponding to the specific block to zero. Or a computer-readable recording medium storing the program.

[0043]

As described above, according to the present invention, when encoding a moving image including a character or other complicated area requiring sharpness, fine quantization is performed while suppressing the code amount. Thus, it is possible to prevent image quality deterioration in these areas.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a schematic configuration of a system including a video encoding device according to an embodiment of the present invention.

FIG. 2 is an explanatory diagram of a character area in which sharpness is required in the embodiment;

FIG. 3 is a block diagram showing a configuration of a video encoding device according to an embodiment of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 11 ... Frame memory 12 ... Readout control circuit 13 ... Subtractor 14, 15, 21, 24 ... Switch 16 ... Orthogonal transformer 17 ... Quantizer 18 ... Inverse quantizer 19 ... Inverse orthogonalizer 20 ... Adder 22 ... Frame Memory 23 ... Motion predictor 25 ... Variable length encoder 30 ... Coding control circuit 31 ... Spatial area determination unit 32 ... Motion prediction control unit 33 ... Prediction error control unit 34 ... Quantization control unit 35 ... Motion vector control unit

Claims (6)

[Claims] An orthogonal transform is performed on each of input video signals or a prediction error signal of an input video signal with respect to a prediction signal in block units to generate a transform coefficient, and the transform coefficient is quantized based on a predetermined quantization coefficient. In the moving picture coding method of performing variable length coding after performing, when a quantization coefficient corresponding to a specific block including a pre-specified spatial region among the blocks of the input moving picture signal is larger than a predetermined threshold value, A moving image encoding method, wherein the quantization coefficient is updated stepwise until the quantization coefficient becomes equal to or less than the threshold value. 2. A moving picture coding method having an intra-frame coding mode and an inter-frame coding mode, and compressing and coding an input moving picture signal inputted in a frame unit. In the inter-frame coding mode, the input video signal input in step 2 is orthogonally transformed in block units with respect to the prediction error signal of the input video signal with respect to the prediction signal obtained by performing motion prediction using a motion vector to generate a transform coefficient. Performing the transforming step, quantizing the transform coefficient based on a predetermined quantized coefficient, performing a variable-length coding on the quantized transform coefficient, and specifying a predetermined one of the blocks of the input video signal. Comparing the quantized coefficient corresponding to the specific block including the spatial region with the predetermined threshold value, and when the quantized coefficient is larger than the threshold value, A quantization coefficient updating step of updating the child coefficient in a stepwise manner until the coefficient becomes equal to or smaller than the threshold value. 3. The moving picture coding method according to claim 1, wherein the predetermined spatial area is a character area requiring sharpness. 4. The quantization coefficient updating step, wherein the quantization coefficient is decreased by a predetermined amount at predetermined time intervals when the quantization coefficient corresponding to the specific block is larger than the threshold value. 3. The moving picture coding method according to claim 2, wherein 5. A step of forcibly setting a motion vector corresponding to a specific block in which the quantization coefficient has been updated in the inter-frame coding mode to a zero vector; 3. The moving picture coding method according to claim 2, further comprising: forcibly setting a motion vector corresponding to a specific block which is not updated to a zero vector, and forcibly setting a prediction error signal corresponding to the specific block to zero. . 6. An input moving image signal or a prediction error signal of an input moving image signal with respect to a prediction signal is orthogonally transformed in block units to generate a transform coefficient, and the transform coefficient is quantized based on a predetermined quantization coefficient. A moving image coding apparatus that performs variable-length coding after performing the processing, wherein: a means for generating an area designation signal indicating a specific spatial area designated for the input moving image signal; and among the blocks of the input moving image signal, Means for updating the quantization coefficient stepwise until the quantization coefficient corresponding to the specific block including the spatial area indicated by the area designation signal is larger than a predetermined threshold value, A moving picture coding apparatus characterized by the above-mentioned.