Structural image codebooks and the self-organizing feature map algorithm

Truong, K.K.; Mersereau, R.M.

doi:10.1109/icassp.1990.116034

Cited by 14 publications

(3 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A key point of this technique is the design of a codebook such that the correlation among spatially close blocks is conveyed to the debook are shown as well. The first quantizer adopted is quite fine, the quantization rule being ej.i E &q = (0, k 1 , k 3 , & 6 , f 10,f 14, f 18, ...I (13) As a consequence, the bit-rate reduction is obtained with a moderate increase of the MSE and, what is more important, with limited subjective degradation (Fig. 16).…”

Section: Discussionmentioning

confidence: 99%

Address‐Predictive Vector Quantization of Images by Topology‐Preserving Codebook Ordering

Poggi

1993

Trans Emerging Tel Tech

View full text Add to dashboard Cite

Address vector quantization is a recently proposed image coding technique based on vector quantization (VQ) which exploits the interblock dependence by jointly encoding groups of VQ addresses. In this paper, by further developing the idea of address encoding, a new image coding technique is proposed. It is based upon the design of an ordered VQ codebook such that co‐devectors which are similar to each other have close addresses. As a result, the performance in terms of bit‐rate is very close to that of Address VQ, but much lower computational complexity and memory storage are required.

show abstract

Section: Discussionmentioning

confidence: 99%

Address‐Predictive Vector Quantization of Images by Topology‐Preserving Codebook Ordering

Poggi

1993

Trans Emerging Tel Tech

View full text Add to dashboard Cite

show abstract

“…The resulting codebook can be viewed as a hierarchy of networks [71], [70] with each level half the size of the previous level as illustrated in Fig. 14. The bottom layer contains all the codewords in the network, while each successive layer contains a subset of the codewords of the previous layer.…”

Section: G Hierarchical Vector Quantizationmentioning

confidence: 99%

Neural network approaches to image compression

1995

View full text Add to dashboard Cite

This paper presents a tutorial overview of neural networks as signal processing tools for image compression. They are well suited to the problem of image compression due to their massively parallel and distributed architecture. Their characteristics are analogous to some of the features of our own visual system, which allow us to process visual information with much ease. For example, multilayer perceptrons can be used as nonlinear predictors in differential pulse-code modulation (DPCM). Such predictors have been shown to increase the predictive gain relative to a linear predictor. Another active area of research is in the application of Hebbian learning to the extraction of principal components, which are the basis vectors for the optimal linear Karhunen-Loève transform (KLT). These learning algorithms are iterative, have some computational advantages over standard eigendecomposition techniques, and can be made to adapt to changes in the input signal. Yet another model, the self-organizing feature map (SOFM), has been used with a great deal of success in the design of codebooks for vector quantization (VQ). The resulting codebooks are less sensitive to initial conditions than the standard LBG algorithm, and the topological ordering of the entries can be exploited to further increase coding efficiency and reduce computational complexity.

show abstract

“…However, such schemes sacrifice either the simplicity of the coding system or the storage cost.' [1][2][3][4][5][6][7][8][9][10][11][12][13] In this paper. we propose an algorithm for structural codebook design based upon the SOFM algorithm.…”

Section: Introductionmentioning

confidence: 99%

<title>Structural codebook design and address-dependent vector quantization</title>

Wang

1996

SPIE Proceedings

View full text Add to dashboard Cite

A structural codebook design algorithm is proposed in this paper, which is aimed at considerably reducing the computational burden during the coding process of vector quanlization (VQ) while having almost no extra storage cost. Kohonen's self-organizing feature map (SOFM) algorithm is improved as to tackle two problems in codebook design. Since the address of currently coded block is often the same as that of the previously coded neighboring blocks, especially in the flat regions, an address-dependent vector quantization (ADVQ) scheme is proposed to further reduce the bit rate and computational complexity. The simulation results show that the ADVQ scheme can achieve a bit-rate reduction of 37% for a typical image "Lena while the computational complexity is reduced by a factor of 20.

show abstract

Structural image codebooks and the self-organizing feature map algorithm

Cited by 14 publications

References 13 publications

Address‐Predictive Vector Quantization of Images by Topology‐Preserving Codebook Ordering

Address‐Predictive Vector Quantization of Images by Topology‐Preserving Codebook Ordering

Neural network approaches to image compression

<title>Structural codebook design and address-dependent vector quantization</title>

Contact Info

Product

Resources

About