Spatially varying blur estimation from a single image
Shereen El‐Shekheby, Rehab F. Abdel‐Kader, Fayez W. Zaki · IET Image Processing · 2019
Detection of single image spatially varying blur is a challenging task. Here, the authors propose a powerful spatially varying blur detection method. Blur is detected from a single image without requiring any prior knowledge about the parametric blur kernel, the camera settings, or information about the input image. The proposed approach detects both motion and defocus blurred regions in partially blurred images. First, blur‐type classification is performed using Hough transform and the structured learning edge‐detection method. Second, the initial blur kernel is estimated with a kernel direction that maximises the likelihood of a local window being blurred by incorporating either vertical or positive diagonal kernels. Blur detection is implemented using a kernel‐specific feature. Third, initial blur image regions are refined with the support of the reduced edges image segmentation (CCP) method. Finally, neighbouring information is utilised to refine detected blur results. Comparative results reveal accuracy improvements in the image blur detection results at a reasonable execution time.