KR100192779B1 - A compatible encoder using lattice vector quantization - Google Patents

Abstract

The present invention relates to a compatible encoder using grid vector quantization, comprising: a first frame memory (10) for storing an input video signal; A subsampler (11) for subsampling the video signal of the first frame memory (10) to produce a low resolution video signal; A second frame memory 12 for storing the low resolution video signal of the subsampler 11; A TS vector quantizer encoder 13 for quantizing and outputting a TS vector quantized low-resolution video signal of the second frame memory 12; A TS VQ decoder (14) for inputting the encoded video sequence of the TS vector quantizer coder (13) and then performing TS VQ decoding to reproduce a low resolution video signal; An interpolator 15 for interpolating and expanding the reproduced low resolution video signal; A subtractor 16 for subtracting the interpolated and reproduced image signals from the high resolution image data of the first frame memory 10 and outputting a difference image; It consists of a lattice vector quantizer (LVQ) for outputting the difference image of the subtractor 16 by lattice vector quantization (LVQ) to efficiently perform compatible encoding.

Description

Compatible Coder Using Grid Vector Quantization

1 is a block diagram showing a compatible encoder according to the present invention.

* Explanation of symbols for main parts of the drawings

10: first frame memory 11: subsampler

12: second frame memory 13: TS VQ encoder

14: TS VQ Decoder 15: Interpolator

16 Subtractor 17 Grid Vector Coder (LVQ)

The present invention relates to a compatible encoding technique for dividing and encoding a video signal into a low resolution (or low level) video signal and a high resolution (or high level) video signal. Particularly, the present invention relates to a lattice vector quantization of a difference between a low resolution video signal and a high resolution video signal. (Lattice Vector Quantiztion; hereinafter referred to as VLQ) relates to a compatible encoder using grid vector quantization transmitted.

In general, digital image coding and technology are widely known as transform encoding, vector quantization, band division encoding, predictive encoding, and entropy encoding, while the International Standardization Organization establishes standards for encoding still and moving images. Efforts have been made to study encoding techniques to obtain more advanced encoding techniques.

On the other hand, compatible coding technology is widely studied for reasons different from the above coding technology, compatible coding (compatible coding) is a technology used to have compatibility between the existing TV technology and the next high-definition (HD) TV.

Here, the compatibility refers to the ability to decode the compressed information of the image having a high resolution into an image having a given resolution, and this compatibility is further divided into an upward compatibility and a downward compatibility. Upward compatibility is that a high-resolution receiver decodes a bit string transmitted by a low-resolution encoder, and then up-samples and reproduces it on a screen or reproduces a portion of a screen by the size of a low-resolution signal. Downward compatibility means that a low resolution receiver decodes and reproduces a required portion of a bit string transmitted by a high resolution encoder.

In addition, if compatibility is divided based on resolution and image quality, compatibility of resolution and image quality can be divided into resolution compatibility. It is the ability to decode and play the video, and the compatibility of the picture quality is that the receiver of the same resolution decodes all the transmitted bit strings when the transmission network is in good condition, and reproduces the high-definition video. It is the ability to play back images of basic quality.

The advantage of granting compatibility by the compatibility encoding technique is firstly that even a low-cost receiver can process a service corresponding to a high resolution signal, and as a result, a broadcast supporting multiple receivers is possible. This is useful in the middle of service improvement. A representative example thereof is compatibility with a TV in the field of HDTV.

Secondly, only a part of the entire bit string can be used. In a cable transmission network, the service can be connected to various data rates so that the user can select a service according to the cost.

In addition, another benefit of compatibility encoding is called graceful degradation, which guarantees the provision of basic video quality even when the transmission network is in a bad condition. Instead of leaving it unattended, an efficient error correction code is implemented in the low resolution part of the transmitted data so that the corresponding low resolution video signal can be received on the HDTV. In other words, this may be considered to be a case where a low-resolution signal is sent to a high priority cell in an ATM network.

As such, the encoding of video signals refers to a function of transmitting and receiving video signals of two or more standards or formats through the same transmission channel, thereby providing the same service to users with different receivers. Will be. In other words, in case of introducing HDTV broadcasting, it is a function to receive the service provided by HDTV as general digital TV since all users cannot purchase expensive HDTV receivers at the beginning.

Such compatibility can be classified into a spatial scalability technique and a frequency scalability technique. Spatial scalability is a technique of dividing an image into spatial encoding and linking it to compatible encoding. Spatial scalability is a technique of dividing an image into a transform region and linking it to compatibility encoding. As a technique, the existing block transformation encoding technique is applied to implement compatibility. In general, the filament coding technique that implements spatial scalability can optimally encode a TV signal, but it does not encode the excess signal used for reconstruction of the HDTV signal well, and the frequency scalability method using transform encoding is an HDTV signal. Can be optimally encoded, but has the disadvantage of not encoding TV signals well.

Accordingly, an object of the present invention is to provide a compatible encoder using a lattice vector quantization (LVQ) that transmits a difference between a high resolution video signal and a low resolution video signal by lattice vector quantization and transmits the encoded vector with the encoded low resolution image sequence.

In order to achieve the above object, the compatible encoder of the present invention subsamples an input video signal to obtain a low resolution video signal, and then outputs the low resolution video signal as a TS vector quantized to output a low resolution video sequence. After decoding and expanding the low resolution video sequence, the reproduced video is compared with the input video signal, and the difference is obtained. The difference video is then quantized by LVQ and output as a high resolution video signal.

Thus, the image quality of the image can be improved by transmitting the video sequence of TS vector quantization of the low resolution video signal and the video sequence of lattice quantization (LVQ) of the difference between the video signal reproduced from the video sequence and the input video signal. .

Hereinafter, with reference to the accompanying drawings will be described an embodiment of the present invention;

The compatible encoder according to the present invention includes: a first frame memory (10) for storing an input video signal as shown in FIG. A subsampler (11) for subsampling the video signal of the first frame memory (10) to produce a low resolution video signal; A second frame memory 12 for storing the low resolution video signal of the subsampler 11; A TS VQ encoder (13) for outputting a low resolution video signal of the second frame memory (12) by vector quantization (Tree Structured); A TS VQ decoder (14) which inputs the encoded video sequence of the TS VQ encoder (13) and decodes the TS VQ to reproduce the video signal of a low resolution; An interpolator 15 for interpolating and expanding the reproduced low resolution video signal; A subtractor 16 for subtracting the interpolated and reproduced image signals from the high resolution image data of the first frame memory 10 and outputting a difference image; And a lattice vector quantizer (LVQ) 17 for outputting the difference image of the subtractor 16 by lattice vector quantization (Lattice VQ).

Referring to the operation of the compatible encoder according to the present invention configured as described above are as follows.

First, as described above, since the input video signal is separated into a low resolution and a high resolution video signal, and then encoded and transmitted in a different or the same manner, the receiving side decodes only the low resolution video sequence by using a low resolution decoder. It is a technology that can satisfy both low resolution and high resolution services by receiving or decoding high resolution video by decoding both low resolution and high resolution video sequences with a high resolution decoder.

In the compatible encoder according to the present invention, the input video signal is stored in the first frame memory 10 once, and the sub sampler 11 obtains the low resolution video signal by subsampling and then transmits the result to the TS VQ encoder 13. Is encoded and then transmitted as a low resolution video sequence. On the other hand, the low resolution image sequence is demodulated again in the TS VQ decoder 14 and then sampled by interpolation in the interpolator 15.

Here, TS VQ will be described.

First, a top-down method and a bottom-up method for representing an image as a block having a variable size, that is, a quad tree, according to the amount of information or the detail of the image, etc. There is a method, of which the bottom-up method is more efficient. In the bottom-up method, the merge quad tree is the most common, and the image is divided while merging the blocks through homogeneity checking between four neighboring blocks. After the merge quad tree process is completed, a tree and a tree structure code representing the result are generated. For example, the tree structure code allocates 1 to a parent node and 0 to a leaf node. Each terminal node allocates a pixel value of a corresponding subpicture in the entire picture. TS VQ divides an image into a tree structure as described above, and then encodes the image by vector quantization using different codebooks for each divided region.

Returning to FIG. 1 again, the subtractor 16 subtracts the interpolated video signal from the interpolator 15 from the original input video signal to obtain a difference video signal and supplies it to the grid vector quantizer 17.

The lattice vector quantizer (LVQ) 17 performs a lattice vector quantization (LVQ) on the difference image signal output from the subtractor 16 and transmits it as a high resolution image sequence.

Here, grid vector quantization is proposed to overcome the shortcomings of vector quantization by LBG (Linde, Buzo, Gray) algorithm which is widely used, and is designed to have a structured codebook in consideration of the statistical characteristics of a given input signal source. Therefore, by storing only the size information of the codebook without storing all the codebooks, the same code can be obtained at the transmitting end and the receiving end, and there is no restriction to search all the areas to find the divided areas. Therefore, the complexity of the quantizer is close to the scalar quantizer, so that the quantizer can be applied to a system requiring real-time coding, and the size of the quantizer is not limited because the size of the codebook is not limited.

The lattice vector quantizer 17 selects an infinite number of lattice, which is a set of n-dimensional vectors defined as follows, according to a given coding rate and dimension, and selects a certain number of lattices as output vectors and uses them as codebooks.

Where ai is the basic vector of the lattice, ci is an integer, and n is a dimension.

As described above, the compatible encoder according to the present invention encodes and transmits a low resolution video signal by tree structure vector quantization, and improves compatibility by encoding and transmitting a high resolution video signal by lattice vector quantization. There is an effect that can improve the image quality.

Claims (1)

A compatible encoder for dividing an input video signal into a high resolution video signal and a low resolution video signal, the encoding apparatus comprising: a first frame memory (10) for storing the input video signal in units of frames; A subsampler 11 subsampling the video signal of the first frame memory 10 into a low resolution video signal; A second frame memory 12 for storing the low resolution video signal of the subsampler 11 in units of frames; A TS VQ encoder (13) for encoding and outputting a low resolution video signal of the second frame memory (12) by tree structure vector quantization; A TS VQ decoder (14) for inputting the encoded video sequence of the TS VQ encoder (13) and decoding the same by tree structure vector inverse quantization to reproduce a low resolution video signal; An interpolator 15 for interpolating the reproduced low resolution video signal; A subtractor 16 for subtracting the interpolated image signal from the high resolution image data of the first frame memory 10 and outputting a difference image; And a lattice vector quantizer (LVQ) configured to output a lattice vector quantization of the difference image of the subtractor 16 as a high resolution image sequence.