KR19980083167A - Transform encoding method and device considering background - Google Patents

Abstract

The present invention relates to a method and an apparatus for transform encoding and distinguishing a central portion having many movements of a frame and a background portion having little movement. According to the present invention, a background part of a frame is predetermined, and a macroblock corresponding to a predetermined part of an input image frame is downsampled to 8 × 8 blocks, and then discrete cosine transform (DCT) is performed on the macroblock of the background part. It is configured to do it only once without doing block unit DCT four times. Accordingly, the present invention provides an effect of reducing the amount of computation and shortening the overall compression encoding time.

Description

Transform encoding method and device considering background

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a video encoding system. In particular, since macroblocks (16 × 16) of a background portion having almost no motion of a video are downsampled to 8 × 8 blocks and transform-encoded, the overall compression encoding is reduced. The present invention relates to a conversion encoding method and apparatus in consideration of a background capable of shortening time.

Recently, international standardization of video compression technology for encoding video at very low bit rate for video telephony or video conferencing system such as Moving Picture Experts Group (IV) -IV has been actively conducted. have. Since MPEG-IV transmits video using a low-speed transmission channel of 64 kbps or less, high compression of video is essential. For this purpose, discrete cosine transform (DCT), quantization, variable-length-coding, dynamic estimation, and dynamic analysis are performed on pictures divided into M (horizontal) × N (vertical) blocks. An apparatus for encoding image data by performing compensation encoding and differential pulse coded modulation (DPCM) has been proposed.

Such an encoding apparatus encodes the entire block area in units of macroblocks having a size of 16x16 pixels. Here, the discrete cosine transform (DCT), one of the data compression techniques, uses a correlation between adjacent pixels in a screen to display image data in a spatial domain in units of 8 × 8 pixel blocks. domain). Transform coefficients are typically weighted, quantized and variable length coded. Data compressed in this way has a variable amount of data due to conversion and variable length encoding. For example, in the case of an image having low motion, the compressed data has a small amount of data, and in the case of a high-motion image, the compressed data has a large amount of its data.

However, since the entire frame is encoded by block, there is a problem in that the overall compression encoding time is long due to a large amount of computing power.

Accordingly, an object of the present invention is to reduce the number of transform blocks by downsampling the macroblocks of the background region into blocks in consideration of the fact that the motion mainly occurs in the center part of the frame and the background part has almost no change, thereby reducing the number of transform blocks. It is to provide a method for reducing the overall compression encoding time by reducing.

Another object of the present invention is to provide an apparatus for implementing a conversion encoding method in hardware in consideration of the above-described background.

1 is a block diagram showing a video encoding apparatus to which the present invention is applied;

2 is a view for explaining a background portion determination method.

Explanation of symbols for main parts of the drawings

11: schedule planning 12: background judge

13 Downsampler 14 Discrete Cosine Converter (DCT)

15 quantizer 15 encoder

In order to achieve the above objects, a conversion encoding method considering a background of the present invention includes a method of converting and encoding input frame video data, the method comprising: (1) dividing input frame video data into macroblock units; (2) receiving the partitioned macroblocks, determining whether the input macroblock corresponds to a center portion or a background portion of the frame, and (3) a macro corresponding to the center portion based on the determination result; Outputting the block as it is, down-sampling and outputting the macroblock corresponding to the background part in block units, and (4) converting the downsampled block unit image data and macroblock unit image data into block units. Outputting the transform coefficients, and (5) receiving and encoding the transform coefficients.

In order to achieve another object of the present invention, a conversion encoding apparatus considering a background of the present invention is a device for converting and encoding input frame-by-frame video data, wherein the constant size planning unit divides the video data of the input frame into a predetermined size. And a background determiner that designates a background portion of the frame in advance and determines whether the image data of a predetermined size unit partitioned by the constant size planning unit corresponds to the designated background portion, and a background according to the determination result of the background determiner. A downsampler which downsamples the partial image data by a conversion block unit of a predetermined size, and receives the image data of the conversion block unit downsampled by the downsampler and the image data of a predetermined size unit and converts the conversion data into a conversion block unit A discrete cosine transformer for outputting the Oh and a quantization, and an encoder for quantizing and encoding.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 shows a configuration of a video encoding apparatus to which the present invention is applied, which is a background determiner 12 and a down sampler 13 for judging and downsampling a background portion of an input image frame in a general video encoding apparatus. Configure additionally.

In FIG. 1, a moving picture signal in a frame unit photographed by a video input means (not shown) such as a camera is input to the constant size planning unit 11. The constant size planning unit 11 divides the video signal in a frame unit into a predetermined size, that is, a macroblock (MB) unit of 16 × 16, and outputs it to the background determiner 12. The background determiner 12 designates the background part of the frame in advance, and determines whether the input macroblock corresponds to the background part. A background part determination method of the background determiner 12 will be described with reference to FIG. 2.

FIG. 2 is a view for explaining a background part determination method of the present invention, and shows a frame of a QCIF structure composed of 11 (horizontal) x 9 (vertical) macroblocks.

2, the hatched portion of the frame is defined as a background portion with little movement. That is, 11 macroblocks (11 (horizontal) x 1/9 (vertical)) portions in the horizontal direction from the first row in the vertical direction of the frame, and then three macroblocks each in the horizontal direction from two rows in the vertical direction ( 3/11 (horizontal) × 2/9 (vertical) sections, followed by two macroblocks (2/11 (horizontal) × 3/9 (vertical)) sections in each of three rows in the vertical direction and horizontally, And then the background portion is defined by one macroblock (1/11 (horizontal) × 3/9 (vertical)) portions each in three rows in the vertical direction and in both the horizontal directions. This background part may vary depending on the type of frame.

Returning to FIG. 1, the background determiner 12 determines that the input macroblock corresponds to the hatched portion of the frame shown in FIG. The background determiner 12 outputs the macroblock determined as the center portion to the discrete cosine transformer (DCT) 14 as it is, and outputs the macroblock determined as the background portion to the downsampler 13. The downsampler 13 downsamples the macroblocks of the background portion input through the background determiner 12 into 8 × 8 blocks and outputs them to the discrete cosine converter 14. There are many ways of downsampling, which are not known because they are already known techniques. Here, the downsampler 13 performs downsampling by selecting any one of four blocks constituting the macroblock with respect to the macroblock determined as the background part. The discrete cosine converter 14 performs discrete cosine conversion on a macroblock output from the background determiner 12 or a block unit output from the downsampler 13 and outputs conversion coefficients. In this case, the discrete cosine transformer 14 has an advantage of reducing the amount of computation since only one block unit conversion is required without performing four block unit conversions for the macroblock of the background portion. The quantizer 15 receives and quantizes the transform coefficients output from the discrete cosine transformer 14 and outputs the quantized coefficients to the encoder 16. The encoder 16 receives quantized coefficients and variably encodes them to output encoded data.

When restoring the coded data as described above, the block corresponding to the background part is upsampled back to a 16 × 16 macroblock to restore the original frame.

As described above, the present invention relates to a method and an apparatus for transform encoding in consideration of a background. Macroblocks corresponding to a central portion of a moving screen perform discrete cosine transforms on a block basis, and macroblocks corresponding to a background portion Discrete cosine transforming into a single block obtained by downsampling reduces the amount of computation for the background part, thereby reducing the compression coding time of the entire frame.

Claims (8)

In the method for converting and encoding inputted frame-by-frame video data, (1) dividing input frame unit video data into macroblock units; (2) receiving the divided macroblocks and determining whether the input macroblock corresponds to a center portion or a background portion of the frame; (3) outputting the macroblock corresponding to the center portion as it is, and downsampling the macroblock corresponding to the background portion in block units based on the determination result; (4) receiving the downsampled block unit image data and macroblock unit image data and converting the result in block units to output conversion coefficients; And (5) A transform encoding method considering a background including receiving and encoding the transform coefficients. The method of claim 1, wherein the step (2) specifies a background part of the frame in advance, and determines whether the input macroblock corresponds to the designated background part. 2. The background as claimed in claim 1, wherein the step (3) comprises downsampling four 8x8 blocks constituting the macroblock 16x16 determined as the background part into one 8x8 block. Transform encoding method taking into account. In the device for converting and encoding the input frame-by-frame video data, A constant size planning unit for dividing the image data of the input frame into a predetermined size; A background determiner that designates a background portion of a frame in advance and determines whether image data of a predetermined size unit divided by the predetermined size planning unit corresponds to the designated background portion; A downsampler for downsampling the image data of the background part by a conversion block unit of a predetermined size according to the determination result of the background judge; A discrete cosine transformer for receiving down-sampled image data of a transform block unit and image data of a predetermined size unit and converting the transform block unit to output transform coefficients; And A transform encoding apparatus considering a background including a quantizer and an encoder for quantizing and encoding transform coefficients output from the discrete cosine transformer. The apparatus of claim 4, wherein the constant size planning unit divides the image data of the input frame into macroblock units of 16x16 pixels. 6. The image determining apparatus of claim 5, wherein the background determiner outputs the macroblock unit image data to the discrete cosine converter when it is determined that the image data input in macroblock units is the center portion of a predetermined frame. A conversion encoding apparatus considering a background, wherein the block unit image data is output to the downsampler. 6. The background processor of claim 5, wherein the downsampler selects and outputs any one block of four 8 × 8 pixel blocks constituting a macroblock determined as a background part by the background determiner. Conversion encoding device. 8. The discrete cosine converter converts image data of a macroblock unit corresponding to a center portion of a frame four times in block units and converts only one block that is downsampled with respect to a macroblock corresponding to a background portion. Conversion encoding apparatus considering the background, characterized in that.