摘要： 提出一种古钱币图像的多尺度相对矩检索方法。利用小波多尺度特征模板提取边缘图像，基于多尺度相对矩进行特征提取，对 特征向量归一化并进行相似性度量。实验结果表明，该方法的检索效果较优，也同样适用于汽车图像的检索，具有较优的通用性与较强的鲁棒性。

关键词: 小波变换, 古钱币图像, 多尺度边缘, 多尺度相对矩, 相似性

Abstract: This paper proposes a multi-scale relative moment retrieval method for ancient-coin image. It takes the feature template of multi-scale wavelet to get the boundary images and extracts boundary image feature based on multi-scale relative moments. It normalizes the feature vector and measures the similarity of it. Experimental results show that the method can win better retrieval effect and also apply to car image retrieval and has good universality and robustness.

Key words: wavelet transform, ancient-coin image, multi-scale edge, multi-scale relative moment, similarity

中图分类号: 

• TP391