On the quantization distortion and coding efficiency in line drawing transmission using differential chain coding

Abstract

Differential chain coding (DCC) exploits the statistical correlation between successive vectors in encoding Freeman-vector-quantized line drawings to achieve data reduction. A quantitative analysis of the performance of DCC is presented. The authors develop relations between average quantization distortions (characterized by per-length error-of-area and per-length error-of-length) and the sampling step, and the tradeoff between the quantization distortions and the coding rate. The analysis is based on the chord approximation of curve segments and a series of a priori probabilities characterizing their statistical properties. In particular, by assuming a normal distribution for the angle difference of two consecutive segments of smooth curves, the authors establish the higher coding efficiency of DCC relative to standard chain coding. >