タイトル | A hybrid LBG/lattice vector quantizer for high quality image coding |
本文(外部サイト) | http://hdl.handle.net/2060/19920014178 |
著者(英) | Ramamoorthy, V.; Sayood, K.; Arikan, E. |
著者所属(英) | Nebraska Univ. |
発行日 | 1991-01-01 |
言語 | eng |
内容記述 | It is well known that a vector quantizer is an efficient coder offering a good trade-off between quantization distortion and bit rate. The performance of a vector quantizer asymptotically approaches the optimum bound with increasing dimensionality. A vector quantized image suffers from the following types of degradations: (1) edge regions in the coded image contain staircase effects, (2) quasi-constant or slowly varying regions suffer from contouring effects, and (3) textured regions lose details and suffer from granular noise. All three of these degradations are due to the finite size of the code book, the distortion measures used in the design, and due to the finite training procedure involved in the construction of the code book. In this paper, we present an adaptive technique which attempts to ameliorate the edge distortion and contouring effects. |
NASA分類 | COMPUTER PROGRAMMING AND SOFTWARE |
レポートNO | 92N23421 |
権利 | No Copyright |
|