摘要
针对非对称失真立体图像,提出了一种基于奇异值分解的无参考评价算法.该方法首先考虑人眼对空间频率变化敏感的特性和双目融合特性,对立体图像进行Gabor滤波,基于奇异值分解的融合策略生成融合图.然后,采用亮度加权直方图的局部二值模式算法分别对融合图、左右子图像提取特征,并将左右子图像的特征向量融合、采用欧几里得距离和夹角余弦进行向量之间的比较;为度量非对称失真差异,利用图像相似度算法计算左右子图像之间的相似性.最后,将融合图的特征向量、子图像的融合及比较特征向量、子图像的相似度特征向量级联,利用支持向量回归(SVR)算法完成特征到主观质量分数的回归映射.在LIVE3DⅡ、Waterloo-IVCⅠ和Waterloo-IVCⅡ立体图像库上对本算法进行测试.实验结果表明,本算法性能良好,优于目前主流的立体图像质量评价算法.
For asymmetrically distorted stereoscopic images,a no-reference evaluation algorithm based on singular value decomposition is proposed.First,considering visual sensitivity to spatial frequency variation and binocular fusion,Gabor filtering was performed on the stereoscopic image,and a fusion strategy based on singular value decomposition was proposed to generate a cyclopean image of the left and right sub-image pair.Then,the proposed luminance-weighted histogram local binary pattern metric was used to extract features of the cyclopean image and the left and right sub-images.Furthermore,feature fusion and comparison were conducted on the two feature vectors corresponding to the left and right sub-images,respectively.Euclidean distance and cosine were used to implement the vector comparison.Particularly,to measure the difference between asymmetrically distorted sub-image pair,image similarity metric was utilized to calculate the similarity between the left and right sub-image pair.Finally,feature vector of the cyclopean image,the fusion and comparison feature vectors,and the similarity feature vector were concatenated into a total feature vector,and regression mapping was performed from the feature vector to the subjective score using support vector regression.The algorithm was tested on the LIVE 3 DⅡ,Waterloo-IVCⅠand Waterloo-IVCⅡdatabases.The experimental results show that the proposed algorithm has an outstanding performance and is superior to other state-of-the-art image quality assessment metrics.
作者
沈丽丽
王莹
Shen Lili;Wang Ying(School of Electrical and Information Engineering,Tianjin University,Tianjin 300072,China)
出处
《天津大学学报（自然科学与工程技术版）》
EI
CSCD
北大核心
2020年第6期641-646,共6页
Journal of Tianjin University：Science and Technology
基金
国家自然科学基金资助项目(61520106002).
关键词
立体图像质量评价
非对称失真
奇异值分解
欧几里得距离
图像相似度
stereoscopic image quality assessment
asymmetric distortion
singular value decomposition
Euclidean distance
image similarity
