JPH08130735A - Picture data encoding and decoding device - Google Patents

Abstract

PURPOSE: To provide a picture data encoder capable of obtaining visually high quality pictures by making block noise inconspicuous. CONSTITUTION: In this picture data encoder provided with a block division means 1 for dividing a screen into blocks composed of plural picture elements, a DCT processing means 3 for discrete cosine transforming pixel data for the respective blocks, a quantization means 4 for lineraly quantizing a DCT coefficient inside the block obtained by a DCT processing by a step size different for the respective coefficient positions and an encoding means 8 for variable length encoding the quantized data, a block start position selection means 9 for specifying a coordinate position to be the reference of the division of the blocks differently for respective frames is provided. The block noise is moved for the respective frames and becomes inconspicuous.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image data encoding / decoding device for encoding a moving image signal using a band compression technique, and more particularly, it is constructed so as to suppress the appearance of noise in a reproduced image. Is.

[0002]

2. Description of the Related Art Image data often has a large correlation between adjacent pixels within the same frame or between pixels near the same position in the preceding and following frames. Bandwidth compression of a signal is performed by utilizing visual characteristics.

Predictive coding and transform coding are known as the band compression methods. In predictive coding, signal redundancy is removed by coding a difference value in corresponding pixels between frames or a difference value between neighboring pixels in the same frame.

Further, in the transform coding, the pixel data is orthogonally transformed, and the signal energy at that time is concentrated in the low frequency component, so that the low frequency component with large power is quantized with a fine step size. The average number of bits is reduced by performing quantization with a coarse step size for high-frequency components of low power. In this compression method, the distortion is difficult to detect even if coarse quantization is performed where the luminance changes rapidly, while the noise due to the quantization tends to stand out when the luminance changes slowly. Utilizes visual characteristics.

ISO-IEC / JTC1 / is a method of encoding moving image data for storage using a band compression technique.
MPEG (Mo which was discussed and standardized in SC2 / WG11
The vingPicture Coding Experts Group) method is known. In this method, predictive coding and transform coding are combined to compress moving image data.

Conventional image data encoding / adopting this system
As shown in FIG. 8, the decoding device includes a block division unit 1 for dividing a screen into blocks composed of a plurality of pixels, and a difference value between an image of a block to be encoded and a prediction image in pixel data. 2 to output as
A DCT processing unit 3 for performing a discrete cosine transform on the pixel data of this block, and a DCT obtained by the DCT transform.
A quantizer 4 that linearly quantizes coefficients with different step sizes for each coefficient position, an inverse quantizer 5 that reverses the data obtained by the quantizer 4 and restores DCT coefficients, and an inverse quantizer 5 Inverse discrete cosine transform (I
Inverse D for performing DCT) to restore pixel data of a block
The CT processing unit 6, the prediction unit 7 that collects the pixel data to restore the screen information, and outputs the predicted image corresponding to the input block, and the data output from the quantization unit 4 using entropy using the Huffman code. A variable length coding unit (VLC unit) 4 for coding and converting to variable length data is provided.

On the other hand, the decoder is an inverse VLC unit 11 which restores quantized data by subjecting an input signal to entropy decoding.
An inverse quantization processing unit 12 that restores DCT coefficients from the quantized data, an inverse DCT processing unit 13 that restores pixel data of a block by performing an inverse discrete cosine transform on the DCT coefficients, and a screen that collects the pixel data. A prediction unit 15 that restores information and outputs a prediction image corresponding to a block for reproducing an image,
Pixel data output from the inverse DCT processing unit 13 and the prediction unit 15
1 based on the pixel data of each block output from the adder 14 and the adder 14 that adds the prediction image output from
A frame playback unit 16 that plays back one frame screen and a frame buffer 17 that stores the played frame image are provided.

In this encoding device, the moving image data to be input is first divided into a plurality of pixels (for example, by the block dividing unit 1).
It is divided into data for each block composed of 8 × 8).
The pixel data for each block is input to the DCT processing unit 3 when being encoded in a frame,
Discrete cosine transform processing is applied to obtain 8 × 8 image information.
Converted to a DCT coefficient of × 8. The quantizer 4 linearly quantizes this coefficient data using a quantization table that defines a step size for each coefficient.

The VLC unit 5 encodes the quantized DCT coefficient using Huffman code. The Huffman code is a variable-length code in which the length of one word is not constant. By assigning a short codeword to a signal with a high occurrence probability and a long codeword to a signal with a low occurrence probability, the average codeword length is Can be kept short. As a result, the amount of information to be output can be reduced.

The quantized data output from the quantizer 4 is converted into DCT coefficient data by the inverse quantizer 5,
The inverse DCT processing unit 6 restores the pixel data of the block. The prediction unit 7 collects the pixel data of each block and stores the screen information.

On the other hand, when encoding is performed between frames, the pixel data for each block output from the block division unit 1 is input to the adder 2. This pixel data is also input to the prediction unit 7, and the prediction unit 7 uses the stored screen information to determine where in the peripheral area the block moves and the cumulative total of the difference values of the pixel data becomes minimum. Is calculated to obtain motion vector information, which is sent to the VLC unit 8 for coding. In addition, the prediction unit 7 outputs the pixel data of the destination block to the adder 2 as predicted image data.

The adder 2 subtracts the predicted image data output by the prediction unit 7 from the pixel data of the block output by the block division unit 1, and DCT the difference data between the frames.
Output to the processing unit 3. The DCT processing unit 3, the quantization unit 4, and the VLC unit 8 encode the difference data into variable length data by the above-described operation, and output the variable length data together with the motion vector information.

The inverse quantizer 5 and the inverse DCT processor 6 are also provided.
Restores the quantized difference data to the state of the difference data output from the adder 2. The prediction unit 7 adds this value and the predicted image data output to the adder 2 to restore the screen information of the next frame, stores the screen information, and uses it to create the next predicted image data.

On the decoder side, when the input signal is coded in the frame, the inverse VLC unit 11 restores the entropy code of the data received, and the inverse quantization unit 12 outputs D.
The data is converted into CT coefficient data, and the inverse DCT processing unit 13 restores the pixel data of the block. The restored pixel data is an adder
Input to 14, but in the case of intraframe coding, the prediction unit
There is no output from 15.

The frame reproduction unit 16 collects the block information output from the adder 14 and reproduces the screen information of one frame. This screen information is output after being stored in the frame buffer 17.

The pixel data of the block restored by the inverse DCT processing unit 13 is input to the prediction unit 15, and the prediction unit 15 restores and stores the screen information by collecting the pixel data of this block.

On the other hand, when the input signal is encoded between frames, the inverse VLC section 11, the inverse quantization processing section 12 and the inverse DCT processing section 13 convert the variable length encoded inter-frame difference data. Inverse DCT processing section 13
Outputs the interframe difference data of the block to the adder 14. At this time, the prediction unit 15 receives the information of the motion vector, reads the pixel data of the block that is the reference of the difference calculation from the stored screen information, and adds the pixel data.
Output to. The adder 14 includes an inverse DCT processing unit 13 and a prediction unit.
The original pixel data of the input block is restored by adding the outputs of 15. The frame reproduction unit 16 is an adder
The screen information of one frame is reproduced by collecting 14 pieces of restored block information, and this screen information is stored in the frame buffer 17 and then output.

The inter-frame difference data of the block output from the inverse DCT processing unit 13 is also input to the prediction unit 15, and the prediction unit 15 adds this data and the pixel data output to the adder 14 and then Restore the screen information of the frame and store it.

The difference between the frames can be obtained not only in the image of the frame preceding the image to be encoded but also in the image of the subsequent frame.

[0020]

However, in the conventional image data encoding / decoding apparatus, since the encoding is performed for each block, a discontinuous portion appears in the image at the boundary between the blocks. This phenomenon is called block noise.
Since this block noise always has the same start position of block division, block noise appears at the same position. Therefore, it looks like noise like a line stays at the same position visually.

The present invention solves these conventional problems, and provides an image data encoding / decoding device capable of visually obtaining a high quality image by making block noise inconspicuous. The purpose is to do.

[0022]

Therefore, in the present invention, the block division means for dividing the screen into blocks composed of a plurality of pixels, the DCT processing means for performing the discrete cosine transform of the pixel data of each block, and the DCT processing are used. An image data coding apparatus comprising a quantizing means for linearly quantizing a DCT coefficient in a block with different step sizes for each coefficient position, and a coding means for variable-length coding the quantized data. The means is provided with block start position selecting means for differently specifying a coordinate position serving as a reference for starting block division for each frame.

Decoding means for restoring the quantized data from the variable length code, dequantizing means for restoring the DCT coefficient from the quantized data, and inverse DCT means for restoring the pixel data for each block from the DCT coefficient. In an image data decoding device including a frame reproducing device for reproducing a frame image based on pixel data for each block, a frame reproducing device is provided with a coordinate position that is a reference for starting division of a block by the encoding device. A frame reproduction position selection means for designating a reproduction reference position in image reproduction is provided.

[0024]

As a result, the block noise is moved in each frame and becomes inconspicuous.

[0025]

Embodiments Image data encoding /
The decoding device, as shown in FIG.
The block division unit 1 includes a block start position selection unit 9 that selects a start coordinate when performing block division, and the decoder side includes a restoration start position when reproducing a frame from the pixel data of the block output from the adder 14. A frame reproduction position selection unit 18 for selecting is provided. Other configurations are the same as those of the conventional device (FIG. 8).

The block start position selection unit 9 includes a random number generator for generating a random number from 0 to 15 when the block is composed of 16 × 16 pixels, and the start coordinates of the block division are set to the random number generator. It depends on the value that occurs. Further, the frame reproduction position selection unit 18 has the same random number generator that always outputs the same value, and determines the value generated from this random number generator as the restoration start position of the frame image.

Block division on the encoder side is performed in the procedure shown in FIG.

Step 1: Block start position selector 9
When the frame image is input, the coordinates (3, 5) determined by the two values (for example, 3 and 5) generated from the random number generator are selected as the start coordinates for block division, and the coordinate data is selected. Output to 1.

Step 2: The block division unit 1 divides the frame image into blocks as shown in FIG. 3, using the coordinates (3, 5) output from the block start position selection unit 9 as the block start position.

When the next frame image is input, the block start position selection unit 9 determines the coordinates (7, 1) determined by the two values (for example, 7 and 1) generated by the random number generator in step 1. The result is output to the block division unit 1, and the block division unit 1 divides the next frame image into blocks as shown in FIG. 4 with the coordinates (7, 1) as the block start position in step 2. Thus, each time a frame image is input, this image is divided into blocks with a random offset.

The operation of encoding the pixel data divided into blocks is the same as the conventional device.

On the other hand, on the decoder side, the inverse VLC unit 11, the inverse quantization processing unit 12, the inverse DCT processing unit 13, the prediction unit 15 and the adder 14 restore the pixel data of each block by the same operation as the conventional device. Then, this pixel data is reproduced as a frame image by the procedure shown in FIG.

Step 3: When the first block of a frame is input, the frame reproduction position selection unit 18 outputs the same value as the random number generator of the block start position selection unit 9, and the coordinate value output by the random number generator. (3, 5) is output to the frame reproduction unit 16. Step 4: The frame reproduction unit 16 uses this coordinate (3, 5)
Starting from, the frame image is reproduced as shown in FIG.

When the first block of the next frame is input, the frame reproduction position selection unit 18 outputs the coordinate value (7, 1) generated by the random number generator to the frame reproduction unit 16 in step 3. The frame reproduction unit 16 performs step 4
As a result, with the coordinates (7, 1) as the starting point,
Play the frame image.

In this way, the restored block pixel data is moved to the correct position on the screen based on the offset value added for each frame at the time of encoding, and the frame image is restored. The restored frame image is stored in the frame buffer 17.

By doing so, the boundaries of the blocks in each frame are constantly moved, and even if block noise appears, they do not stop at the same place. Therefore, the block noise is less noticeable and is visually the same as an image without block noise.

In the embodiment, the same random number generator as the block start position selection unit 9 is provided in the frame reproduction position selection unit 18, but the coordinates output from the block start position selection unit 18 are coded and transmitted. The frame reproduction position selection unit 18 may be configured to select the frame reproduction position based on this encoded information.

[0038]

As is apparent from the above description of the embodiments, the image data encoding / decoding apparatus of the present invention can obtain a visually high quality image because the block noise does not stay in the same place. You can

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of an image data encoding / decoding device according to an embodiment of the present invention,

FIG. 2 is a flowchart showing the operation of a block start position selecting unit of the image data encoding device according to the embodiment.

FIG. 3 is a diagram showing a state of block division by the image data encoding device according to the embodiment;

FIG. 4 is a diagram showing another state of block division by the image data encoding device according to the embodiment,

FIG. 5 is a flowchart showing the operation of a frame reproduction position selection unit of the image data decoding device according to the embodiment.

FIG. 6 is a diagram showing a state of frame reproduction by the image data decoding device according to the embodiment;

FIG. 7 is a diagram showing another state of frame reproduction by the image data decoding device according to the embodiment;

FIG. 8 is a block diagram showing a configuration of a conventional image data encoding / decoding device.

[Explanation of symbols]

 1 block division unit 2, 14 adder 3 DCT processing unit 4 quantization unit 5, 12 inverse quantization unit 6, 13 inverse DCT processing unit 7, 15 prediction unit 8 VLC unit 9 block start position selection unit 11 inverse VLC unit 16 Frame playback unit 17 Frame buffer 18 Frame playback position selection unit

Claims (2)

[Claims] 1. A block dividing means for dividing a screen into blocks composed of a plurality of pixels, a DCT processing means for performing a discrete cosine transform (DCT) of pixel data of each block, and D.
Image data coding provided with a quantizing means for linearly quantizing the DCT coefficient in the block obtained by the CT processing with a different step size for each coefficient position, and a coding means for variable-length coding the quantized data. The image data coding apparatus according to claim 1, wherein the block division means is provided with block start position selection means for differently specifying a coordinate position serving as a reference for starting division of the block for each frame. 2. Decoding means for restoring the quantized data from the variable length code, dequantizing means for restoring the DCT coefficient from the quantized data, and inverse DCT means for restoring the pixel data for each block from the DCT coefficient. An image data decoding device comprising a frame reproduction means for reproducing a frame image based on pixel data for each block, wherein a coordinate position is used as a reference for starting division of the block by the encoding device with respect to the frame reproduction means. An image data decoding apparatus, characterized by further comprising frame reproduction position selecting means for designating as a reproduction reference position in reproduction of the frame image.