Abstract:  A new descriptor based on wavelet transform is proposed and used to match objects’ partial shape under affine transform. The algorithm firstly partitions the contour into many sub-curve based on division points. Then, in order to describe the curve more accurately, a new feature point, namely equal area segmentation point, is defined and extracted in this paper. According to the feature points, a new local feature descriptor is constructed to act as criteria for contour. Equal area segmentation point is sampled from each sub-curve and it can be employed to describe the curve more precisely than the common feature points such as corners, points of tangency, inflection points. It can solve the problem that the smooth curve has fewer feature points so that curve can not be described precisely. The equal area segmentation points and the recognition vectors constructed by this paper are local descriptors and invariant under affine transformation. So the matching method is robust under occlusion and affine transformation. The theory analysis and experimental results have shown the algorithm is effective.

Key words: match, partial shape, affine transform, wavelet transform

摘要： 为解决仿射变换下的局部形状匹配问题，提出了一种新的基于小波描述子的局部形状匹配方法。算法首先基于分割点将轮廓曲线分段，为了更精确地描述每段子曲线，定义一种新的特征点——等面积分割点，并在子曲线上提取，基于该特征点构造一种新的具有局部特征的小波描述子。新定义的等面积分割点有比一般的特征点（角点、拐点、切点）更精确描述曲线的特性，能解决轮廓曲线平滑特征点少而不能被精确描述的问题；定义的等面积分割点和提取的小波描述子都具有仿射不变性，且均为局部描述符，因而该方法适合于仿射变换且在轮廓局部遮挡和缺失的情况下仍然有效。理论分析和实验结果都证明了该算法的有效性。

关键词: 匹配, 局部形状, 仿射变换, 小波变换