摘要： 在基于内容的图像检索中，传统的颜色直方图和频域特征直方图不能满足颜色过渡、梯度方向和幅值变化等连续性和相似性。为此，提出非线性模糊颜色直方图和非线性模糊梯度向量直方图检索算法。该算法通过符合人类视觉非线性感知特性的方式描述图像，结合图像的统计和频域特征，构建一个特征组组合选取的方法，选出适合的标识图像特征组组合以提高图像检索系统的检索效率。实验结果表明，与其他检索算法相比，该算法具有较好的检索质量。

关键词: 基于内容的图像检索, 多尺度分析方法, 非线性模糊颜色直方图, 非线性模糊梯度向量直方图, 特征组组合选取

Abstract: In Content Based Image Retrieval(CBIR) system, the conventional color histogram and spectral histogram can not meet the continuity and similarity of color transition and changes in gradient orientation and magnitude. A image retrieval algorithm of the nonlinearly fuzzy chromatic histogram and the nonlinearly fuzzy gradient-vector histogram is presented. It can depict images with a form which can answer for the human nonlinearly cognitional nature in visual field well. Based on other existing features in spectrum and statistics, the features group combinatorics of selection method is constructed, it selects the most proper image features group combinatorics to promote the effectiveness of image retrieval system. Experimental result shows that, compared with other retrieval algorithms, the proposed algorithm has well retrieval quality.

Key words: Content-based Image Retrieval(CBIR), multi-scale analysis method, nonlinear fuzzy color histogram, nonlinear fuzzy gradient-vector histogram, features group combinatorics of selection

中图分类号: 

• TP391