【摘要】 针对无参考噪声图像质量评价问题,提出基于视觉对比度敏感函数(CSF)和仿射重建模型的噪声图像质量评价方法.引入CSF对待评价噪声图像进行滤波,利用图像分块技术,建立基于最优化问题求解的仿射重建模型,得到图像信号成分,估计出残差信号图像.计算各分块的噪声强度点分布,选取噪声强度点数量分布最多的区间,最终的噪声图像质量评价算子由该强度区间内的所有强度点的均值计算得到.在LIVE、TID2008及CSIQ数据库上开展评价算法主客观一致性评估实验,与其他几种评价算法进行对比,比较算法客观评价性能的表现.实验结果表明,提出的算法具有很好的准确性和主客观评价一致性.更多还原

【Abstract】 An image quality assessment method based on contrast sensitive function(CSF)and affine reconstruction model was proposed for no-reference noisy image quality assessment.The visual contrast sensitivity function was introduced to apply the filtering process for noisy image.The image segmentation algorithm was utilized and the affine reconstruction model was applied to solve the optimal problem.Then image signal was obtained and the residual signal image was estimated from the input image and the signal image.The noise intensity sample of each block was calculated to select the interval with the most noise samples falling in.The final noise image assessment metric was obtained by the mean value of all the noise intensity samples belonging to the selected interval.Experiments were conducted on LIVE,TID2008 and CSIQ image data base in order to evaluate the subjective and objective consistency of the proposed method.The objective performances were assessed compared with other image quality assessment methods.Experimental results illustrate that the presented algorithm has a good performance on accuracy and subjective and objective consistency.更多还原