Improved Sensitivity of No-Reference Image Visual Quality Metrics to the Presence of Noise

Abstract

A problem of no-reference image visual quality assessment when images are corrupted by noise is considered in this paper. A specialized image set is proposed for the following two tasks: automatic verification of sensitivity of no-reference image visual quality metrics to noise, and analysis of blind noise level estimation methods. As a result, a method to improve the sensitivity of a given no reference quality metric to the presence of noise is proposed by combining this metric with a noise level estimator. The proposed method allows to significantly decrease a probability of wrong quality predictions for noisy images. Efficiency of usage of different noise level estimators in the proposed combined metrics is analyzed.