基于图像失真类型的自适应图像质量评价方法  

Self-adaption image quality assessment scheme based on distortion type classification

作　　者：王啸晨[1] 潘榕[2] 

机构地区：[1]天津大学电子信息工程学院,天津300072 [2]国家广播电视产品质量监督检验中心,北京100015

出　　处：《电视技术》2016年第2期137-140,共4页Video Engineering

摘　　要：提出了一种图像质量评价方法,所提出的评价算法抽取DCT系数特征来代表图像的失真类型,使用基于支持向量机的分类器对图像失真类型进行预测,随后针对每一种失真类型,给出相应的融合评价算法,对图像进行质量评价。实验结果表明,该算法在整体性能上要优于传统的PSNR及SSIM等图像质量评价算法。In this paper, a distortion type based image quality assessment algorithm is proposed. DCT coefficients are extracted as distortion feature vectors of images and the distortion type is predicted via support vector machine, then assess the image quality by a fusion assessment algorithm corresponding to each distortion type. Experimentation on the standard database is carried out, and it is found that the proposed method outperforms the traditional image quality assessment algorithms.

关 键 词：支持向量机 图像质量评价 特征提取 失真类型 

分 类 号：TN911.73[电子电信—通信与信息系统]