Restoration of out of focus images using neural network

Abstract

Restoration of out of focus images is important role in imaging system. The lens defocus may cause image blurring. In this paper, a neural network approach to estimate the blur parameter for uniform out of focus blur is proposed. we estimate the parameter of defocused image in frequency domain by using circle Hough transform, and combine with neural network to have the relationship between the parameter of frequency response of out of focus image and the parameter of uniform out of focus blur model. Finally, we restore the out of focus image with its spatial continuous point spread function (PSF) and the trained neural network. The simulation result shows that the average error with the proposed estimation method is smaller than 0.48%, and more accurate than the existing methods.