Abstract

This paper presents a dual-image deblurring algorithm which corrects image blur caused by moving object that follows a projective motion path. We first show how to modify the Richardson-Lucy (RL) algorithm to incorporate a general projective motion blur descriptor to estimate the clear image. Then, we will demonstrate that the Projective Motion RL algorithm can incorporate an iterative reweighted least square (IRLS) approach to improve the deblurred results, with two input images complementary to each other. Synthetic and real-life examples demonstrate that our dual-image algorithm improves significantly over single-image RL algorithm.

Paper Details