Abstract: Content-based Image Retrieval（CBIR） has become a hot research field of image processing. Features extraction and matching of image are the primary means of CBIR. However, it is extremely difficult to extract effective characteristics. This paper combines the HSV color space’s characteristics with laws of human perception of color, to propose a color recognition method. Applying this method, pixels in image are divided into several classes which keep the structure in the original image. Structure feature extraction is implemented in each class. Image feature matching is done between the corresponding classes in different images. This method reduces the complexity of the image features extraction and matching. The experiments show that the proposed image retrieval method has good results.

Key words: image retrieval, fuzzy sets, color recognition, feature extraction

摘要： 基于内容的图像检索（Content-based Image Retrieval，CBIR）以其极高的理论与应用价值成为了图像处理领域的研究热点。提取和匹配图像特征是CBIR的主要手段。然而提取图像的有效特征是极其困难的。利用HSV颜色空间特性以及人类对颜色的感知规律，提出一种颜色识别方法。应用此方法对图像的像素进行一种保持结构的分类，并在类内提取结构特征。图像的特征匹配将在同类像素集合间进行，降低了图像特征提取与匹配的复杂性。实验表明，提出的图像检索方法有良好的效果。

关键词: 图像检索, 模糊集, 颜色识别, 特征提取