首先对曲率尺度空间（CSS）算法进行改进，并用改进的CSS算法对任意形状物体的形状信息进行特征点的分层提取；其次，用自适应算术编码算法对固定长度的八分圆编码算法进行改进，并对所提取的特征点进行算术编码，实验结果表明，这种新的基于特征点提取的形状编码方法比MPEG-4校验模型中基于上下文的算术形状编码算法，在Dn相同的条件下，主观质量更好，且压缩比更高。

In this paper, a new shape coding algorithm called curvature scale space based shape coding(CSS) is proposed. For dealing with video image, we regard a shape as a set of points that are parameterized by arc length. Then, the evolution of curve at different resolution levels in Gaussian scale space is achieved by convoluting the curve with Gaussian kernel. By calculating the curvature, the feature-points including significant information of shape contour can be found. But for shape coding, this shape representation will not very efficient when the shape consists of arcs. So, the CSS is modified, then extract the feature points of shape using the modified CSS algorithm. The octants algorithm is modified and encode the feature points by using adaptive arithmetic code algorithm. Simulation results show that our proposed algorithm has higher compress ratio, better subjective quality compared with other results from the CAE in MPEG-4.