Abstract
Even with the topical developments of numerous image Super Resolution (SR) algorithms, how to quantify the visual quality scores of a super resolved image is still an open research problem. Majority of SR images are evaluated by full-reference metric with the support of a reference image. There are some circumstances when a reference image is unavailable or is with degraded quality. We propose a super resolution benchmark Super Resolution Entropy Metric (SREM) which can be used to evaluate the effectiveness of pixel reconstruction and quality of the image in the absence of reference image automatically. SREM measures the experimental quality of an SR image based on the perceptions of acutance and spatial discontinuity features in the gradient domain and wavelet domain. Experimental scores illustrate that the SREM metric is competent for assessing the visual quality of super-resolved images.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
