A Method for Deblurring Estimation From a Single Image

This paper addresses the challenging problem of removing blurred regions from a single image. We propose a simple yet effective method for estimating the spatial variation of defocus blur at edge positions. The input defocused image is re-blurred using a Gaussian kernel, and the defocus blur amount is obtained by the ratio of gradients between the input image and the re-blurred image. By propagating the blur amount at edge positions to the entire image, we obtain a fully defocused image representing the depth of the scene. We further investigate deblurring methods for images with different degrees of blur, and ultimately achieve deblurring effects for various degrees of blur through parameter dynamicization. Experimental results on real images demonstrate a reliable defocus estimation method for removing blurred regions from images.