A review of vector quantization techniques used for encoding digital images is presented. First, the concept of vector quantization is introduced, then its application to digital images is explained. Spatial, predictive, transform, hybrid, binary, and subband vector quantizers are reviewed. The emphasis is on the usefulness of the vector quantization when it is combined with conventional image coding techniques, or when it is used in different domains. >

Image coding using vector quantization: a review

An algorithm introduced by L. Breiman et al. (1984) in the context of classification and regression trees is reinterpreted and extended to cover a variety of applications in source coding and modeling in which trees are involved. These include variable-rate and minimum-entropy tree-structured vector quantization, minimum expected cost decision trees, variable-order Markov modeling, optimum bit allocation, and computer graphics and image processing using quadtrees. A concentration on the first of these and a detailed analysis of variable-rate tree-structured vector quantization are provided. It is found that variable-rate tree-structured vector quantization outperforms not only the fixed-rate variety but also full-search vector quantization. The successive approximation character of variable-rate tree-structured vector quantization permits it to degrade gracefully if the rate is reduced at the encoder. This has applications to the problem of buffer overflow. >

http://www.angelfire.com/electronic2/ricardo_neri/00032124.pdf

Optimal pruning with applications to tree-structured source coding and modeling

Embodiments of the present invention relate to video coding for multi-view video content. It provides a coding system enabling scalability for the multi-view video content. In one embodiment, a method is provided for encoding at least two views representative of a video scene, each of the at least two views being encoded in at least two scalable layers, wherein one of the at least two scalable layers representative of one view of the at least two views is encoded with respect to a scalable layer representative of the other view of the at least two views.

Method and device for video coding and decoding

The problems of buffer overflow and underflow encountered when employing actually variable or effectively variable bit-rate channels for communicating encoded video images are overcome by jointly controlling the number of bits employed to encode each video frame and the transmission bit-rate of the variable bit-rate channel as experience by the encoder. The selection of the number of bits employed to encode each video frame, and hence the encoder bit-rate, as well as the associated channel bit-rate are determined from the encoder buffer fullness, a determination of the decoder buffer fullness and any constraint imposed on the channel. The encoder is responsive to the selected encoder bit-rate and accordingly adjusts its encoding parameters so as to achieve the selected encoder bit-rate.

Buffer control for variable bit-rate channel

Apparatus for decoding an HDTV signal conveyed as high and low priority data in high and low priority channels respectively, wherein said high and low priority data originated from blocks of hierarchically encoded compressed video data, with data of greater importance for image reproduction from each block allocated to the high priority channel and the remaining data from each block allocated to the low priority channel, includes circuitry for identifying and segmenting high and low priority data for corresponding blocks and recombining the high and low priority data into an hierarchically encoded compressed video data stream.

Apparatus for recombining prioritized video data

A method for data reduction of digital image sequences in heirarchic fashion by means of a multi-stage coder on the basis of vector quantizing and movement compensation, as well as reconstruction of these image sequences by means of a multi-stage decoder. An input image signal is represented in a plurality of image signals having different spatial resolutions by means of a cascade of L identical low-pass filters having subsequent sub-scanning, whereby the spatial resolution and the data volume of the input image signal in the horizontal and the vertical directions are respectively reduced by a factor R in every cascade stage, and whereby the input signal is referenced "resolution level L" and the low-pass filter image signal having the smallest resolution is referenced "resolution level .0.". The chronologically preceding, reconstructed, movement-compensated image signal is represented in a plurality of image signals having different topical resolution in the same way. The samples of the resolution levels are subdivided into blocks, whereby the "resolution level .0." has the block size "1×1" and, for the next, the block size is horizontally and vertically increased by the factor R with every further resolution level but a block assumes the maximum block size N×N.

Method for data reduction of digital image sequences

A method for data reduction of digital video signals by vector quantization of coefficients acquired by orthonormal transformation with a symmetrical, nearly cyclical Hadamard matrix, employs formation of a mean value coefficient and a plurality of structure coefficients from a plurality of blocks formed from the input signal in a Peano curve arrangement, these coefficients being in part only scalarly quantized and in part both scalarly as well as vectorially quantized. The quantization takes place in a coder having at least two stages interconnected cascade-like. The reconstruction of the video signal takes place in a decoder, having an identical plurality of stages. The method allows the noise amplitude of the video signal acquired after back-transformation to be adapted to the visibility thresholds of the human visual system.

Method for the data reduction of digital picture signals by vector quantization of coefficients obtained by orthonormal transformation by means of a symmetrical, nearly cyclical Hadamard matrix

Vector quantization is a promising encoding technique especially for low data rate image transmission. Due to the exponential growing of its computational complexity with the block dimension, however, only small blocksizes have been used in practical applications. This restricts the coding efficiency and produces some blockiness in the reconstructed images. Our proposal solves this problems by a combination of a block-overlapping pyramidal transform with multistage VQ. This concept enables VQ of large blocks in a hierarchical manner with small computational costs, while the block-overlapping principle gives rise to a smooth image reconstruction. The simulation results proved that picture phone sequences are reconstructed without any annoying artifacts.

Hierarchical encoding of image sequences using multistage vector quantization

For bidirectional motion-compensated interpolation in the decoding of compressed video sequences, the invention proposes that only one of the two reference images is completely decoded. From the other reference image, only those areas of the image are decoded which overlap with the area being processed of the interpolated image, allowing for the displacement related to motion compensation.

Method, requiring reduced memory capacity, for decoding compressed video data

For the purpose of bidirectional movement-compensated interpolation, it is proposed, when decoding compressed video sequences, completely to decode only one of the two reference frames. Only those respective picture areas of the other reference frame are decompressed which overlap with the picture area to be determined of the interpolated frame, taking account of a displacement corresponding to the movement compensation. As a result, a considerable reduction in the memory requirement of the decoding hardware is possible.

Bernard Dr. Hammer

Papers

Method for data reduction of digital image sequences

Method for the data reduction of digital picture signals by vector quantization of coefficients obtained by orthonormal transformation by means of a symmetrical, nearly cyclical Hadamard matrix

Hierarchical encoding of image sequences using multistage vector quantization

Method, requiring reduced memory capacity, for decoding compressed video data

Method for decoding compressed video data with a reduced memory requirement