基于人眼视觉特性的彩色图像质量评价

图像处理系统的性能优劣的评判往往需要一个合理迅速的图像质量评价算法作为支撑.传统的图像质量评价算法由于没有充分考虑人眼的视觉特性,使得质量评价结果与实际图像的人眼感知质量不符.根据人眼对图像边缘信息非常敏感这一人眼视觉特性,提出一种综合图像边缘和背录相似度的算法(EBS)来评价彩色图像质量,即通过比较失真彩色图像与原始参考图像的边缘以及除边缘之外的背景相似程度最终确定失真图像的质量.应用于由779幅包含五种类型失真的图像质量评价库的实验结果表明,该算法的评价结果相比PSNR,MSSIM,IFC以及基于像素域的VIF等算法与图像的主观评价结果(由DMOS值表示--将背景不同的一组观察者对失真图像的评分进行统计平均后所得到的评价结果)更一致,也即该算法的评价结果更接近图像的实际视觉感知质量.

Color Image Quality Assessment Based on HVS

Measurement of visual quality is of fundamental importance to some image processing applications, so an efficient image quality assessment algorithm is greatly needed. And the evaluation results of the conventional algorithms do not match to the ones of human vision because those algorithms ignore the characteristic of the Human Visual System (HVS). According to the HVS that human have strong sensitivity to the image edge information, this paper propose a novel algorithm for image quality assessment based on the edge and background similarity between the distorted image and the reference (perfect) image. We test the validation of the proposed algorithm on the database of 779 images with five different distortion types, the results show our method has better performance than PSNR, MSSIM, IFC and VIF, because the evaluation results are closer to the ones of human vision.