Comprehensive evaluation of restoration quality for weak image signals with fuzzy logic

Natural images captured in hazy, sandy or low-light conditions degrade significantly and are termed weak image signals (WIS) in this paper. Evaluating their restoration using multiple single-factor indices is challenging due to potential mutual exclusivity, hindering a holistic understanding. A novel no-reference image quality evaluation approach has been devised specifically to assess the visibility enhancement of restored WIS images. This approach constructs a comprehensive index, based on fuzzy theory, to holistically evaluate the restoration effect of WIS. The index is derived from five carefully selected single-factor indices: new visible edge, visible edge gradient, contrast gain, colour naturalness and colour richness. These indices capture improvements in contrast and colour post-restoration. Fuzzy logic is employed to meticulously analyze the contribution of each factor to the overall image performance. The weights assigned to each individual evaluation index are determined through a rigorous analytic hierarchy process (AHP) analysis, utilizing a judgment matrix. The comprehensive evaluation index, denoted as CNC, surpasses the limitations of single-factor indices by capturing even the subtlest changes across all factors, offering a more comprehensive and nuanced assessment. Experimental validation confirms its alignment with subjective evaluations, confirming its effectiveness in comprehensively evaluating the restoring effect of WIS images.