PROJECT TITLE :

SPSIM A Superpixel-Based Similarity Index for Full-Reference Image Quality Assessment - 2018

ABSTRACT:

Full-reference image quality assessment algorithms typically perform comparisons of features extracted from sq. patches. These patches don't have any visual meanings. On the contrary, a superpixel may be a set of image pixels that share similar visual characteristics and is so perceptually meaningful. Features from superpixels might improve the performance of image quality assessment. Inspired by this, we propose a brand new superpixel-based similarity index by extracting perceptually meaningful features and revising similarity measures. The proposed methodology evaluates image quality on the basis of three measurements, particularly, superpixel luminance similarity, superpixel chrominance similarity, and pixel gradient similarity. The primary two measurements assess the overall visual impression on native images. The third measurement quantifies structural variations. The impact of superpixel-primarily based regional gradient consistency on image quality is also analyzed. Distorted pictures showing high regional gradient consistency with the corresponding reference pictures are visually appreciated. Thus, the three measurements are additional revised by incorporating the regional gradient consistency into their computations. A weighting function that indicates superpixel-based mostly texture complexity is utilised within the pooling stage to get the ultimate quality score. Experiments on many benchmark databases demonstrate that the proposed method is competitive with the state-of-the-art metrics.