摘要
离散余弦变换系数的变化能够有效反映图像的模糊程度变化,基于离散余弦变换系数提出了一种无参考模糊图像质量评价方法。该方法首先通过对图像进行离散余弦变换,得到图像的离散余弦变换系数作为图像质量变化的特征向量,然后使用支持向量回归模型对此特征向量进行训练学习,得到特征向量与模糊图像质量分值之间的映射关系模型,预测图像质量。在LIVE2、CSIQ和TID2008 3个数据库上检测该方法的性能,实验结果表明,新方法预测得分与主观得分有较好的一致性,获得了比较好的评价指标。
The change of discrete cosine transform （DCT） coefficients can effectively reflect image blur quality change ~ In this paper, a new no reference image blur quality assessment method based on discrete cosine transform is proposed~ First of all, the discrete cosine transform is performed on the images, the obtained discrete cosine transform coefficients are taken as the feature vectors of the image quality change, then the support vector regression model is adopted to train the feature vector, the mapping relation model between the feature vectors and the blur image quality score is obtained, and the image quality is predicted. The performance of the proposed method was tested on three public databases LIVE2, CSIQ and TID2008, the experiment results show that the score predicted with this method has a good correlation with the subjective quality score, and superior performance evaluation result is obtained~
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2013年第11期2599-2604,共6页
Chinese Journal of Scientific Instrument
基金
国家自然科学基金(61170120)
江苏省自然科学基金(BK2011147)资助项目
关键词
模糊图像质量评价
无参考
DCT系数
支持向量回归
blur image quality assessment
no reference
discrete cosine transform （DCT） coefficient
supportvector regression
