KR100228677B1 - Apparatus for encoding moving image by dwt - Google Patents

Abstract

Disclosed is a video encoding apparatus using a discrete wavelet transform (DWT) to increase the compression efficiency by the motion prediction based on the discrete wavelet transform (DWT) during video encoding. The present invention provides a frame-by-frame video signal input in an intra mode through intra wave mode through a more useful discrete wavelet transform (DWT) instead of the ordinary discrete cosine transform (DCT), and restores the previous frame video signal. Estimates and compensates the video signal of the current frame unit and inputs the interpolation between the video signal of the current frame and the motion compensated video signal, that is, the motion compensation error through a discrete cosine transform (DCT). Inter-frame coding in mode. Accordingly, the present invention can expect an improved compression ratio compared to the motion prediction based on the conventional discrete cosine transform (DCT).

Description

Video Encoding Device through Discrete Wavelet Transform (DWT)

The present invention relates to a video encoding apparatus based on motion prediction, and more particularly, to obtain a motion compensation error through a motion prediction on a discrete wavelet transform (DWT) and to encode on a discrete cosine transform (DCT). The present invention relates to a video encoding apparatus using DWT to improve compression ratio.

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. MPEG-IV uses a very low-speed network of 64kbps or less to transmit video at a very low bit rate, so high compression of video is essential. To this end, motion compensation is performed to reduce the redundancy in the time axis direction, and the difference between the images is taken. Then, discrete cosine transform (DCT) and variable length coding are used to reduce redundancy in the space axis direction. That is, in general, moving images are almost similar to the images before and after and the current image. Therefore, the difference between the current image to be encoded and the P image (forward prediction image) and the front image in time, or the B image (bidirectional prediction image) and the difference between the interpolation image made in front, rear or front and rear in time In this method, the amount of information to be transmitted is reduced by reducing the redundancy in the time axis direction by encoding and transmitting the difference value. However, not only take the difference of the same position but also use the motion compensation. This is a method of further reducing the amount of information to be transmitted by searching for the minimum difference between the previous image and the difference between the block positions in units of 16x16 pixels and taking the difference therebetween. In practice, in a P picture, a small amount of data is selected and encoded in units of 16x16 blocks (macroblocks), from the difference between the predicted picture after motion compensation and the difference.

In addition, image compression algorithms having a more advanced structure have been proposed. In other words, research on image compression based on fractal image compression theory, model-based image compression theory, object-oriented image compression theory, and wavelet transformation theory is in progress or at the stage of standardization. In particular, the wavelet transform can represent signals having locality with respect to time and frequency, which is advantageous for analyzing non-stationary video signals. In the field of image processing, he started to get the spotlight.

Therefore, the object of the present invention is to predict the motion efficiency on the discrete cosine transform (DCT), unlike the conventional motion estimation on the discrete wavelet transform (DWT) by compressing and encoding the video signal to improve the compression efficiency video through DWT The present invention provides an encoding device.

1 is a block diagram showing an embodiment of a video encoding apparatus through DWT according to the present invention.

<Description of the symbols for the main parts of the drawings>

11: Discrete Wavelet Converter (DWT) 12,19: Quantizer

13,20: encoder 14: inverse quantizer

15: Inverse Discrete Wavelet Converter (IDWT) 16: Motion Predictor

17: Subtractor 18: Discrete Cosine Converter (DCT)

The video encoding apparatus through DWT of the present invention for achieving the above object is a video encoding apparatus, the first encoding means for intra-frame encoding the input current frame video signal based on the discrete wavelet transform (DWT) and Motion estimation means for restoring a previous frame video signal from the video signal encoded by the first encoding means and using the previous frame video signal reconstructed with respect to the input current frame video signal, and for motion estimation and motion compensation. And a second encoding means for obtaining the difference signal between the current frame video signal and the video signal compensated for by the motion prediction means, and inter-frame coding based on a discrete cosine transform (DCT).

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

1 is a block diagram illustrating a video encoding apparatus through DWT according to the present invention. As shown, the apparatus of the present invention receives a video signal obtained through a video input device (not shown) such as a camera and performs a discrete wavelet transform (DWT) to output a conversion coefficient (DWT) ( 11), a first quantizer 12 for quantizing transform coefficients, and a first encoder 13 for encoding and outputting the quantized coefficients through Huffman coding or the like. The first quantizer 12 inversely quantizes the quantized coefficients, inverse discrete wavelet transform (IDWT), and outputs an image signal of the previous frame reconstructed, and an inverse discrete wavelet transformer (IDWT). 15 are connected in turn. The inverse discrete wavelet converter 15 is connected to a motion estimator 16 for motion estimation and motion compensation using the video signal of the previous frame reconstructed from the video signal of the current frame to be input. The motion predictor 16 is connected to a subtractor 17 which obtains a difference between the video signal of the current frame and the motion compensated video signal, that is, a motion compensation error. The subtractor 17 is connected to a discrete cosine transformer (DCT) 18 for discrete cosine transforming motion compensation errors. The discrete cosine transformer (DCT) 18 is configured such that a second quantizer 19 for quantizing the transform coefficient and a second encoder 20 for encoding and outputting the quantized coefficient through Huffman coding or the like are sequentially connected.

The operation of the video encoding apparatus through the DWT of the present invention configured as described above will be described in more detail.

First, a video signal of a frame unit obtained through a video input device (not shown) such as a camera is input to the discrete wavelet converter 11 and the subtractor 17 and to the motion predictor 16. The discrete wavelet converter 11 performs discrete wavelet conversion of the video signal of the current frame to be input to the first quantizer 12. Discrete wavelet transform is an example of subband coding, which consists of an analysis process and a synthesis process. A detailed description of the discrete wavelet transform is already mentioned in several documents and is not mentioned here. The first quantizer 12 changes to representative values of various levels through a predetermined quantization process for the video signal of the current wave-converted discrete frame. The first encoder 13 encodes the data through Huffman coding in consideration of the statistical characteristics of the quantization coefficients to generate compressed data. Here, intra frame intra-compression coding is performed through discrete wavelet transform, quantization, and encoding on the input video signal of the current frame. Meanwhile, the inverse quantizer 14 and the inverse discrete wavelet transformer 15 inversely quantize quantized data output from the first quantizer 12 and then inverse discrete wavelet transform to output the reconstructed previous video signal. do. The motion predictor 16 estimates the motion of each of the macroblock unit image data of the current frame video signal by using the input video signal of the current frame and the reconstructed previous frame video signal. The motion predictor 16 makes a motion estimation based on the correlation between the frames. That is, the motion predictor 16 selects a reference macroblock having image information almost identical to the image data of the current macroblock used for comparison in the search area that is part of the reference image signal based on the reconstructed previous frame image signal. Next, a motion vector representing a spatial position difference between the current macroblock and the reference macroblock is obtained. Then, the motion predictor 16 extracts a portion corresponding to the video information of the reference macroblock designated by the motion vector MV from the reconstructed previous frame video signal and outputs it as a motion compensated video signal. Here, the motion vector MV is transmitted together with the video signal to be compressed and encoded. The subtractor 17 subtracts the corresponding video signal applied from the motion predictor 16 from the input current frame video signal for differential pulse code modulation, that is, inter-frame encoding in inter mode, and divides the difference value obtained by subtraction. Output to cosine converter 18. The discrete cosine transformer 18 performs discrete cosine transform on the difference value applied from the subtractor 17, that is, the motion compensation error, to be converted into coefficients in the frequency domain. The second quantizer 19 changes the transform coefficients into representative values of various levels through a predetermined quantization process. The second encoder 20 encodes through Huffman coding taking into account the statistical characteristics of the quantization coefficients to generate compressed data.

As described above, the present invention relates to a video encoding apparatus using a discrete wavelet transform (DWT), which is more useful than the conventional wavelet encoding based on the discrete cosine transform (DCT) motion prediction. The motion prediction based on the transform (DWT) has the effect of improving the compression efficiency.

Claims (4)

In the video encoding device, First encoding means for intra-frame encoding the input current frame video signal based on a discrete wavelet transform (DWT); Motion estimation means for restoring a previous frame video signal from the video signal encoded by the first encoding means, and using the previous frame video signal reconstructed with respect to the input current frame video signal; And And a second encoding means for obtaining the difference signal between the input current frame video signal and the video signal compensated for by the motion predicting means and encoding between frames based on a discrete cosine transform (DCT). The method of claim 1, wherein the first encoding means A discrete wavelet converter for performing a discrete wavelet transform (DWT) on the input current frame video signal; A first quantizer for quantizing the discrete wavelet transformed video signal; And And a first encoder for encoding the quantized video signal through Huffman coding to output intra-frame coded video data. The method of claim 2, wherein the movement prediction means An inverse quantizer for receiving and inversely quantizing an image signal quantized by the first quantizer; An inverse discrete wavelet converter for outputting the inversely quantized video signal as an inverse discrete wavelet transform (IDWT) and restoring the previous frame image signal; And The motion vector is obtained by estimating the motion of each of the macroblock unit image data of the current frame video signal inputted by receiving the previous frame video signal recovered from the inverse discrete wavelet transformer, and the reference macro designated by the motion vector. And a motion predictor for extracting and outputting image data of a block from a reconstructed previous frame video signal. The method of claim 3, wherein the second encoding means A subtractor for subtracting the corresponding image data applied from the motion predictor from the macroblock unit image data of the current frame to be input, and outputting the difference value obtained by subtraction to the motion compensation error; A discrete cosine transformer for receiving a motion compensation error from the subtractor; A second quantizer for quantizing the discrete cosine transformed motion compensation error; And And a second encoder for encoding the quantized motion compensation error through Huffman coding to output inter-frame coded image data.