小波模拟CSF加权的失真图像质量评价

Distorted Image Quality Assessment Using Weighted Method of Wavelet Simulated CSF Property

提出一种基于小波变换模拟对比敏感度(CSF)特性的加权结构相似度(W-MSSIM)评价方法.采用表达直观、计算简便的结构相似度(SSIM)模型替代模拟人眼视觉因素的各种公式,利用不同空间频带的小波系数进行对比敏感度特性加权,从而分别对被测试图像和参考图像提取出人眼感兴趣的中高频信息,然后进行结构相似度对比,以克服SSIM方法不能对失真图像有效评价的缺陷.经过实验验证,该方法的评价结果与HVSNR和ESSIM方法相比更符合主观评价.

Previous methods to evaluate image quality hardly can assess distorted image effectively. Their results usually can＇t match up to the human visual assessment. To assess distorted images in accuracy,this article proposes a W-MSSIM image quality assessment using weighted method of wavelet frequency band simulated CSF property. Since SSIM is simple and efficient,the proposed method adopts SSIM model to displace the formulas,which express the elements influencing human visual,and to simplify the process of human visual system simulation. CSF property can be weighted by wavelet coefficients of different frequency bands. Hence,the middle and high frequency information can be extracted from tested image and the reference image,respectively. Further,the information of two images is compared to overcome the limitation of SSIM,which can＇t effectively assess distorted image. The experimental test proves that the performance of proposed method is better than HVSNR and ESSIM.