BLIND IMAGE BLUR ASSESSMENT BASED ON MARKOV-CONSTRAINED FCM AND BLUR ENTROPY

The image blur assessment is of various practical use such as feedback of microscope dynamic focusing and assessment of the quality of pictures in social media. However, the problem of providing a fast and sensitive assessment toward image blur is not easy to deal with. In this paper, we provide a new effective way to evaluate the blur level of the image. We first introduce Markov Constraints to the Fuzzy-C-Means (MC-FCM) clustering algorithm to improve the robustness to noise, then obtain the fuzzy membership of pixels via the MC-FCM, finally, to leverage fuzzy membership from MC-FCM, the blur assessment toward pixels in the edge zone is provided by modifying Shannon’s entropy. Comparisons are made on two public blur image database over five recent image blur assessment algorithms. The results demonstrate that the proposed algorithm has better resolutions for mildly blurred images and lower computation complexity compared with existing approaches.