A novel technique for estimation of defocus map from a single image

In this paper, we present a technique for defocus map estimation from solitary natural images that estimates the blur amount based on the radius of blurring kernel. First it estimate the blur amount at all the locations according to the window mean radius kernel on that position and then a full defocus map can be obtained by propagating the blur amount at all the locations over the entire image. Experimental consequences demonstrate that our proposed method is more consistent in terms of Universal Image Quality Index (UIQI) for defocus map estimation compared to diverse state-of-the-art methods.