Brick-wall structured segmentation for interpolative vector quantization of images

Abstract A new variable block size image segmentation scheme based on the modified split-and-merge (MSM) algorithm is devised in this paper. It is shown that the position of the sampling points obtained from the MSM algorithm are located at the edge boundaries and the distance between two consecutive sampling points indicates the local activity between them. An image is partitioned using the properties of the selected sampling points. The proposed method is applied for coding of monochrome image using interpolative vector quantization (IVQ) technique. Simulation results show that the proposed method can achieve 0.8 dB PSNR improvement at a bit rate lower than the fixed block size scheme. Besides, perfect reconstruction of edges is resulted and hence excellent reconstructed image visual quality is obtained.

[1]  Peter Strobach.  Tree-structured scene adaptive coder , 1990, IEEE Trans. Commun..

[2]  R. Gray,et al.  Vector quantization , 1984, IEEE ASSP Magazine.

[3]  Novel interpolative vector quantisation technique for image coding , 1993 .

[4]  Nasser M. Nasrabadi,et al.  Image coding using vector quantization: a review , 1988, IEEE Trans. Commun..

[5]  Allen Gersho,et al.  Variable block-size image coding , 1987, ICASSP '87. IEEE International Conference on Acoustics, Speech, and Signal Processing.

[6]  B.G. Haskell,et al.  Interpolative vector quantization of color images , 1988, IEEE Trans. Commun..

[7]  C.-K. Chan,et al.  Maximum descent method for image vector quantisation , 1991 .

[8]  Allen Gersho,et al.  Image compression with variable block size segmentation , 1992, IEEE Trans. Signal Process..