基于分块颜色矩和纹理特征的图像检索方法

基于内容的图像检索技术是当今研究的热点领域之一.为了克服全局直方图检索方法不包含颜色的空间分布信息的不足,提出了一种采用固定块图像分割方法提取图像局部颜色矩的图像检索算法.在此基础之上,对检索结果进行基于纹理特征的二次检索.实验表明,圆环形分割块颜色矩算法不仅记录了颜色的空间信息,而且具有旋转、平移不变性的特点.综合利用图像颜色、纹理的检索方法能够大大提高检索结果的准确性.

An image search approach based on local main color feature and texture feature

The image search technique based on content becomes a hot research topic. In order to solve the problem of global histogram lacking the spatial distribution information of color, an image search method is presented. It uses fixed block segmentation for drawing the local color moments of an image. Based on this, the second search is completed by drawing the texture feature of the image. Experiments show that a ring block not only includes the spatial information of image color, but also it is characterized with the invariance of rotation and translation. The image search method of using both color and texture features can greatly improve the accuracy of the search results.