An image quality assessment metric based on quaternion wavelet transform

In recent years, a great deal of efforts has been made to develop objective image quality assessment (IQA) that shows consistency with subjective quality evaluation. After structure similarity was found to be an important factor, IQA metrics such as SSIM showed a better performance than traditional ones. With the usage of phase congruency (PC) to obtain structure similarity measurement, a new metric named FSIM acquired even higher consistency. This paper is motivated to propose an improved structure similarity metric based on FSIM, where quaternion wavelet transform is exploited to extract the quaternion phase congruency map to represent the essential image structures. In addition, complex phase congruency map obtained as the co-product during quaternion wavelet transform, supplies complementary visual effects of detailed structure on image quality assessment. The proposed image quality assessment metric based on hybrid phase congruency (HPC) (IQA_HPC) is highlighted in emphasizing saliency distribution of different image structures to image quality assessment. As compared with the state-of-the art IQA methods, experimental results demonstrate that the new index IQA_HPC is able to gain a higher consistency with the subjective measurement.