提高SSIM-Map分辨率的研究

›› 2016, Vol. 1 ›› Issue (12): 21-25.

提高SSIM-Map分辨率的研究

（集美大学诚毅学院，福建厦门361021）

收稿日期:2016-10-10 修回日期:2016-11-01 出版日期:2016-12-15 发布日期:2016-12-15
作者简介:林佳楠，讲师，主要从事通信技术、图像视频质量评测研究.
基金资助:
福建省中青年教师教育科研基金资助项目（JA15606）

Research on Improving the Resolution of SSIM-Map

（Chengyi College, Jimei University, Xiamen, Fujian 361021, China）

Received:2016-10-10 Revised:2016-11-01 Online:2016-12-15 Published:2016-12-15

摘要/Abstract

摘要：

图像质量的客观评测对很多图像处理应用都很重要。和其他现有的算法相比，结构相似度（SSIM）模型最大的优势在
于它是基于图像结构失真，并且它与人类感知高度符合。通过对SSIM算法的深入研究，发现SSIM由于固有缺陷，很难清楚地
找出图像失真的详细部分。提出了一种新的SSIM算法的改进，采用循环改变分区的方法，以此提高SSIM-Map索引图的分辨
率。试验结果证明改进的新SSIM比SSIM更加符合人类视觉系统，对于纹理失真准确度更高。

关键词: 图像质量, 结构相似度, 客观评测

Abstract:

The objective evaluation of image quality is very important for many image processing applications. Compared with other
existing algorithms, the biggest advantage of structural similarity model（SSIM）is that it is based on image structure distortion, and it
is highly consistent with human perception. Through in-depth study of the SSIM algorithm, it is found that it is difficult to clearly
identify the details of the image distortion with SSIM's defects. In this paper, a new SSIM algorithm is proposed to improve the
resolution of SSIM index map. The image resolution of SSIM-Map's index plan can be elevated with the method of changing portion
circularly. The experimental results show that the improved SSIM is more consistent with human visual system than original one, and
the accuracy of texture distortion is higher.

Key words: image quality, structural similarity, objective evaluation

中图分类号:

TN919.85

林佳楠. 提高SSIM-Map分辨率的研究[J]. , 2016, 1(12): 21-25.

Lin Jianan. Research on Improving the Resolution of SSIM-Map[J]. , 2016, 1(12): 21-25.