摘要：

目前常用的评估算法忽视了图像中携带重要信息的边缘特征。基于感兴趣区域的双尺度边缘结构相似度算法虽然考虑了边缘信息的重要性,但对边缘信息的识别效果不理想。针对以上问题,提出边缘结构相似度图像质量评估算法,将图像分为感兴趣区域与不感兴趣区域,分别用边缘结构相似度算法和结构相似度算法对2 种区域进行质量评估,并实现加权组合。实验结果表明,该算法对边缘信息的识别能力更强,对图像质量的变化更加敏感。

关键词: 图像质量评估, 边缘信息, 三次B 样条, 结构相似度, 感兴趣区域, 模糊

Abstract:

The normal evaluation algorithm does not bring out the edge features which are taking along important information in the image. Though Region of Interest(ROI) and dual-scale edge structure similarity algorithm take the importance of edge information into consideration, this algorithm is not an ideal one to the identification of edge information. Based on the above information,this paper proposes the edge structure similarity quality algorithm which is based on the interesting region. The algorithm departs the image into interesting and non-interesting region,and evaluates these two regions with the edge structure similarity quality algorithm separately,and realizes weighted combination. Experimental results show that the algorithm has a stronger ability to identify the edge information and is more sensitive to the variations of the image quality.

Key words: image quality assessment, edge information, cubic B-spline, Structural Similarity ( SSIM ), Region of Interest(ROI), blur

中图分类号: 

• TP391. 41