基于人眼感知度的分块加权图像检索

Block-weight image retrieval based on human eye perception measure

提出了一种基于人眼感知度的图像检索算法。该算法首先由像素的人眼感知度的计算得到各分块的权值,然后对分块的LBP纹理直方图和HSV颜色直方图进行加权处理,综合纹理和颜色特征进行图像检索。实验结果表明,该算法结合了局部图像的相关性,利用了纹理和颜色特征的互补能力,检索结果符合用户所需并具有较好的查准率和查全率。

This paper proposed a image retrieval based on human eye perception.Firstly,computed the weight of each block according to HEPM of pixel.Secondly,used the weight to process LBP texture-histogram and HSV color-histogram of each block.Finally,executed the retrieval.The experiment results demonstrate that the algorithm combined texture and color feature with coherence of local image,show the effectiveness in improving the retrieval performance.