基于视觉注意的全参考彩色图像质量评价方法

摘要

如何实现对图像质量的高效准确评价一直是图像和信号处理领域的一个关键问题,针对当前大多图像质量客观评价算法忽略人眼视觉系统(HVS)的影响以及只考虑灰度图像质量评估的不足,提出一种基于视觉注意模型的全参考彩色图像质量评价(IQA)算法;该算法将彩色图像视觉显著性(VS)作为相似性特征与彩色图像多通道边缘强度相似性测度相结合,最终实现对彩色图像的质量评估;LIVE2数据库中实验证明,该算法在复杂度、鲁棒性和与主观评价一致性方面都有明显优势.%How to achieve efficient and accurate evaluation of image quality is always a fundamental problem in image and signal processing.While most of the current image quality assessment algorithms ignore the influence of human visual system (HVS) and only considering gray images.To solve this problem,a full reference color image quality assessment (IQA) model is proposed in this paper,which is based on color image visual attention (VS).The procedures involved in the presented method include computing the VS similarity and multi-channel edge intensity similarity between reference and distorted images.The experiments on LIVE2 show that the presented method has very high consistency with the psychological evaluation,and has obvious advantages in complexity,robustness and consistency comparing with subjective evaluation.