No-Reference Quality Assessment of Noise-Distorted Images Based on Frequency Mapping

In this paper, we propose a no-reference image quality assessment (IQA) metric for noisedistorted images specifically based on frequency mapping (FM), namely, FMIQA index. First, we decompose the image into intrinsic mode functions (IMFs) from small to large scale by using bidimensional empirical mode decomposition (BEMD), and perform the local feature analysis on the IMFs by Riesz transform. Considering that the combination of BEMD and Riesz transform can denoise the noise-distorted image, we use this method with appropriate application of visual contrast sensitivity function to get the denoised image. Then we calculate the similarity map of the Riesz transform feature maps from the distorted image and the denoised image to obtain the similarity indices. Finally, we combine these similarity indices to obtain the final index. Experimental results on three public databases show that the proposed FMIQA evaluates the noise-distorted image in consistency with subjective assessment and can obtain better performance in image quality prediction than other existing related methods.

Abstract licence: CC BY-NC-ND

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.