摘要
该文提出一种基于小波变换方法对目标形状的描述方法．在给出小波轮廓描述符数学定义的基础上，详细分析和实验验证了描述符的各个性质，并讨论了描述符的实际计算问题．另外结合基于内容的图像查询对小波轮廓描述符与傅里叶轮廓描述符的描述和检索性能进行了比较，证实了小波轮廓描述符的优越性．
A method for object shape description based on wavelet transform is proposed. After giving the mathematical definition of wavelet boundary descriptor, the basic properties of descriptor, such as unity, comparability, are explained; the descriptive properties of descriptor, such as efficiency of description, ineffectiveness of displacement and scaling over description, are analyzed in detail and verified with experiments. Some problems that related to the practical computations of wavelet boundary descriptor, such as normalization of displacement and scaling, calculation and truncation of transform coefficients, are discussed. In addition, the retrieval performances of wavelet boundary descriptor and Fourier boundary descriptor are compared in the context of content based image querying, and the advantages of wavelet boundary descriptor are confirmed.
出处
《计算机学报》
EI
CSCD
北大核心
1999年第7期752-757,共6页
Chinese Journal of Computers
基金
国家自然科学基金
国家八六三高技术研究发展计划
关键词
小波变换
轮廓描述等
图像查询
图像数据库
Wavelet transform, shape descriptor, boundary descriptor, content based, image querying.
