状特征由于其简单、直观等优点在基于内容的图象检索技术研究中非常重要。本文在简要介绍了基于内容的检索技术（Content－basedRetrieval，CBR）的基础上，从不变特征提取。相似性度量、检索方式等方面，详细探讨了利用形状相似性进行图象CBR的技术关键。实验结果表明，不变矩和相对面积等特征相结合可以有效地进行形状检测。相信本文的工作，将对构建某些特定应用的系统具有很好的参考价值。 Shape-like features are very important in the research of content-based image retrieval because of its simplicity and intuition. Based on the content-based retrieval technology (Content-based Retrieval, CBR), this paper introduces invariant feature extraction. Similarity measure, retrieval method and so on, discusses in detail the key technology of shape CBR image CBR. The experimental results show that the combination of features such as moment invariant and relative area can effectively detect the shape. I believe this article’s work, will build a system for some specific applications have a good reference value.