Tongxin xuebao (Aug 2014)
High efficient shape coding based on the representation of contour and chain code
Abstract
A high efficient lossless shape coding scheme was proposed based on the representation of contour and chain code. The object contours are firstly extracted and thinned to be single-pixel width. Then the object contours are trans-formed into chain-code-based representation and divided into different sub-segments based on link directions to make each sub-segment consist of up to two types of links. Thirdly, straightness detection is performed and long straight-line segments within contours are separated. Finally, different schemes are proposed to encode the straight-line sub-segments and ordinary curve sub-segments by employing the spatial correlations among contours aiming to acquire high coding performance. Experiments are conducted and the results show that the proposed scheme outperforms the existing tech-niques and can improve the compression efficiency by 36.5% on average compared with the state-of-the-art method.
