客观评估彩色图像质量的超复数奇异值分解法

Hypercomplex Singular Value Decomposition Approach to Objectively Assessing Color Image Quality

本文利用超复数直接对彩色图像建模,保存了彩色图像完整的信息;基于超复数奇异值分解(也称四元数奇异值分解QSVD)提出一种全新的图形化与数值化相结合的彩色图像质量评估测度,不仅能判断图像失真等级,还能判断不同的失真类型.测试结果表明,本文提出的算法比传统的MSE、PSNR以及MSSIM等算法性能更优.

We employ hypercomplex to describe a color image directly so that all information of the color image can be modeled. Based on hypercomplex singular value decomposition also called quaternion singular value decomposition （QSVD）, we present a novel color image quality measure that can be used as a graphical or a scalar measure to predict the distortion introduced by a wide range of noise sources. It reliably measures the distortion not only within a distortion type at different distortion levels, but also across different distortion types. We compare its performance with MSE,PSNR and MSSIM,the test results show that QSVD performs better.