一种奇异值与其向量联合评估图像质量的测度

An Image Quality Assessment Metric Based on Singular Values and Singular Vectors

提出一种新的利用奇异值分解来评估全参考图像质量的测度.该测度联合使用奇异向量和奇异值,试图从能量和一种新的结构来评估图像失真.理论分析表明,该测度不仅能评估图像的局部误差,而且能评价图像的全局误差.仿真实验表明,与传统的算法相比,该测度能更有效地评估失真图像,尤其是噪声图像的质量,且与人眼的主观评价更加一致.

A novel effective full-parameter image quality assessment metric is introduced from the viewpoint of energy and structure.The novel objective metric uses the singular vectors and singular values as the features to assess the quality of the distorted image.The novel metric can evaluate not only the local errors but also the global errors.Simulation shows that compared with traditional algorithms,the proposed metric can greatly improve the consistency with the DMOS,especially for the noisy images.