Answers are grounded in this paper and cite their sources.
A novel algorithm for no-reference blur assessment in digital pictures is proposed. This metric is based on the evaluation of local phase coherence across the scales of an overcomplete wavelet transform. We compare the performance of the proposed method with four state-of-the art metrics using a large database containing both simulated and real blur.
No takes yet. Share an insight, caveat, or question.
Ciancio et al. (2009) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: