基于内容的图像检索（CBIR）中图像颜色特征提取方法的研究和改进

摘要

The paper presents a method for image retrieval with the aid of image segmentation. This method is based on color distribution histogram and uses the HSV color model which is consistent with human visual. With the help of MeanShift Algo-rithm,we use quadtree to divide the image and extract the main color,which is helpful for us to get the distribution of color his-togram. Last but not least we use the EMD algorithm to measure the distance between images ,so that we can implement the im-age search.%该文给出了一种基于颜色分布直方图借助四叉树图像分割进行图像检索的方法，该方法使用与人类视觉感知相符合的HSV颜色模型，通过构造四叉树借助MeanShift算法对图像进行分割，提取出特征颜色，得到颜色分布的直方图。最后利用EMD算法衡量图像与图像之间的距离，实现图像的搜索。