摘要
提出了一种基于图像信息熵的商标图像检索系统，该系统使用图像的单元熵作为特征来描述图像形状，用欧氏距离来度量熵矢量间的相似性。用１３１０ 幅图像构成的数据库进行实验，结果表明，该系统输出的相似图像与人的视觉感应吻合性较好，并且具有很好的比例不变性，抗噪声性和抗浓淡性。
In this paper,we present a method for such system based on the image content using unit entropy.Experimental results on a database of 1310 trademark images demonstrate that the proposed method retrieval visually similar trademarks which agree well with human perception, at the same time, it has scale invariance, noise invariance, thick and thin invariance.
出处
《计算机应用》
CSCD
2000年第2期21-23,共3页
journal of Computer Applications
