基于子块直方图的图像检索算法研究

Image Retrieval Arithmetic Research Based on Block Histogram

基于全图直方图图像检索算法过于简单及查全率、查准率过低的情况,本文提出把图像分割成子块的算法,寻找空间分布差异,间接捕捉到色彩的空间分布特征,从而能够区别色彩分布相似但空间分布不同的图像,尤其适用于纹理丰富的图像的检索,获得满意的结果。

Based on the simplicity of whole histogram image retrieval method and the low precision, a new method is proposed to divide an image into several blocks to search for the differences of spatial distribution, and to get the distribution features. By this way, we can differentiate the images of same in color but different in spatial distribution. This method is especially effective for texture images.