JPH066782A - Moving image encoding device - Google Patents

Abstract

PURPOSE:To improve the encoding efficiency of a moving image. CONSTITUTION:When the movement degree of the moving image is high, a DCT block A is outputted from a switching circuit 16 and when the degree is low, a DCT block C is outputted from the circuit. When the DCT block A is outputted, if the fluctuation in the vertical direction is large, the DCT block A is outputted from a DCT block constituting circuit 20 and if it is small, a DCT block B is outputted from it. Similarly, when the DCT block C is outputted, if the fluctuation in the vertical direction is large, the DCT block C is outputted from the circuit 20, and if it is small, the DCT block D is outputted from it. Thus, the optimum mode can be selected according to the local property of the moving image.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a moving picture coding device, and more particularly to a moving picture coding device for interlaced scanning.
Divide the frame into blocks each consisting of multiple pixels,
The present invention relates to a moving picture coding apparatus that orthogonally transforms a pixel value on a block, or orthogonally transforms a residual with a value predicted from a pixel value on another frame or field.

[0002]

2. Description of the Related Art In a known moving picture coding apparatus 1 shown in FIG. 3, a DCT (Discrete Disc) is fixed by fixing it to either an intra-frame DCT selected by a switching circuit 2 or an inter-frame DCT.
te Cosine Transform: Discrete cosine transform) and other transform coding were performed.

[0003]

DISCLOSURE OF THE INVENTION Problems to be Solved by the Invention
When transform coding such as CT is performed, the coding efficiency differs between the intra-frame DCT and the inter-frame DCT depending on the local property of the image (moving region or still region, etc.), but in the conventional technique of FIG. Or D between frames
Since the coding is fixed to CT, there is a problem that the coding efficiency is lowered depending on the local property.

Therefore, a main object of the present invention is to provide a moving picture coding apparatus which can obtain high coding efficiency.

[0005]

SUMMARY OF THE INVENTION According to the present invention, one frame of an interlaced-scanned moving image is divided into blocks each having a plurality of pixels, and pixel values on the block are orthogonally transformed, or another frame is formed. Alternatively, in the moving picture coding method in which the residual difference between the pixel value on the field and the predicted value is orthogonally transformed, the first mode in which the original pixel value is orthogonally transformed while the block is divided for each frame is changed to the block for each field. A second mode in which the original pixel values are subdivided and orthogonally transformed, and a third mode in which motion compensation prediction is performed in field units and the prediction error values are orthogonally transformed in block units of each frame,
And motion-compensated prediction in field units, sub-divided into blocks for each field, and orthogonally transforming prediction error values
The moving picture coding apparatus is characterized in that an appropriate mode is selected from four kinds of coding modes of the mode and the coding is performed.

[0006]

When a moving image is DCT-encoded, for example, a first mode is used for a large motion area when the vertical variation is large, and a second mode is used when the vertical variation is small. For small motion areas,
If the variation in the vertical direction is large, the third mode is used, and if the variation in the vertical direction is small, the fourth mode is used. As a result, the image is encoded in the optimum mode in consideration of the local property of the image.

[0007]

According to the present invention, since the coding is performed according to the local property of the image, the coding efficiency of the moving image can be improved regardless of the magnitude of the motion. The above-mentioned objects, other objects, features and advantages of the present invention will become more apparent from the following detailed description of the embodiments with reference to the drawings.

[0008]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Referring to FIG. 1, a moving picture coding apparatus 10 of this embodiment includes a block dividing circuit 12 and divides an input image into DCT blocks of size 4 × 4 pixels, for example. For example, four DCT blocks form a macroblock of size 8 × 8 pixels, and the subsequent processing is performed in macroblock units. The block division circuit 12 is shown in FIG.
The DCT block A shown in (A) is output to the subtractor 14, the switching circuit 16, and the motion compensation frame memory 40 (described later). The DCT block A is configured as frame data by combining the pixels of the first field and the pixels of the second field in order to perform the DCT in the frame. However, the first field and the second field indicate that they are adjacent fields. The same applies below.

Next, the inter-field prediction error data obtained by taking the difference between the data of the current frame and the decoded data of the previous frame by the subtractor 14 in the DCT block A, that is, the DCT block C shown in FIG. 2C. In the frame /
The inter-frame determination circuit 18 compares. DCT block C
In order to perform inter-frame DCT, the inter-frame prediction error data of the first field and the inter-field prediction error signal of the second field are combined to form frame data. According to the result of the intra-frame / inter-frame determination circuit 18, the switching circuit 16 switches and codes the DCT block A or C, that is, the original image or the prediction error image. In-frame / inter-frame determination circuit 18
Then, for example, the averages of the variances of the DCT blocks A and C are compared, and the one with the smaller average value is selected. When the motion is large, the DCT block A has a smaller average variance than the DCT block C, so the DCT block A is selected. When the motion is small, the DCT block C is selected by the DCT block C.
The DCT block C is selected because the average of the variances is smaller.

When the switching circuit 16 outputs the DCT block A, the DCT block A and FIG.
DCT using any of the DCT blocks B shown in FIG.
Or when the DCT block C is output from the switching circuit 16, the DCT block C and the DCT block C shown in FIG.
DCT using any of the DCT blocks D shown in (D)
DCT block configuration circuit 20
Is determined by. Here, the DCT block B is configured as field data in which the pixels of the first field and the pixels of the second field are separate blocks. The DCT block D is configured as field data in which the inter-field prediction error data of the first field and the inter-field prediction error data of the second field are separate blocks. At this time, the DCT block configuration circuit 20 newly forms the DCT block B or D as needed, and makes the above-mentioned determination using, for example, the average of the variance of each DCT block.

When the vertical variation of the image is large,
When DCT blocks A and C have smaller average variances than DCT blocks B and D, respectively, while DCT blocks B and D, conversely, have smaller variances in the vertical direction of the image, respectively. The average dispersion is smaller than that of C. Therefore, when the motion is large and the vertical fluctuation of the image is large, the DCT
When the block A has a large movement and the vertical fluctuation of the image is small, the DCT block B has the small movement and the DCT block C has the small movement of the vertical fluctuation of the image. The DCT block C has a small motion and the vertical fluctuation of the image. , The DCT block D is selected.

Any of the DCT blocks determined by the DCT block configuration circuit 20 is a DCT circuit 22.
The DCT coefficient is quantized by the quantization circuit 24. The quantized data is stored in the variable length coding circuit 2
6 is variable length coded and transmitted or recorded. Also,
The quantized data is given to the adder 34 via the inverse quantization circuit 28, the inverse DCT circuit 30, and the DCT block inverse configuration circuit 32, and is added to the data from the switching circuit 36. Zero data and composited image data of the previous frame are input to the switching circuit 36, and either data is output according to the comparison result in the intra-frame / inter-frame determination circuit 16. That is, within frame /
In the inter-frame determination circuit 16, for example, the averages of the variances of the DCT blocks A and C are compared, and if the average of the variances of the DCT block A is smaller, zero data is output from the switching circuit 36 and the DCT block C If the average of the variances is smaller, the switching circuit 36 outputs the composited image data. In this way, the data locally decoded by the adder 34 is fed back to the subtractor 14 via the motion compensation frame memory 38.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention.

FIG. 2 is an illustrative view showing a configuration of a pixel of each DCT block A to D.

FIG. 3 is a block diagram showing a conventional technique.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 ... Moving image coding apparatus 12 ... Block division circuit 14 ... Subtractor 16 ... Switching circuit 18 ... In-frame / inter-frame determination circuit 20 ... DCT block configuration circuit 22 ... DCT circuit

Claims (1)

[Claims] 1. A frame of an interlaced-scanned moving image is divided into blocks each having a plurality of pixels, and pixel values on the blocks are orthogonally transformed, or pixel values on another frame or field are converted. In a moving picture coding system that orthogonally transforms the residual with the predicted value, the original pixel value is re-divided into blocks for each field in the first mode in which the original pixel value is orthogonally transformed while maintaining the block division for each frame. Is orthogonally transformed, a motion-compensated prediction is performed in a field unit, and a third mode is performed in which a prediction error value is orthogonally transformed in a block unit in each frame, and a motion-compensated prediction is performed in a field unit to re-create a block in each field. It is characterized in that an appropriate mode is selected from four kinds of coding modes of a fourth mode for dividing and orthogonally transforming a prediction error value for coding. , A video encoding device.