Linear blur compensation in digital images using Lucy-Richardson method

This paper considers the problem of image restoration, in particular, linear blur compensation. To restore image we use Lucy-Richardson iterative method and its modifications. The main disadvantages of the basic method are edge artifacts in the form of horizontal and vertical stripes on the image and lack of information about the optimal number of algorithm iterations. To decrease the negative impact of those disadvantages we propose to extend the image beyond its original borders reducing its brightness to zero at new borders, as well as use an empirical criterion to define the point where the iterative Lucy-Richardson procedure should be finished. The criterion is based on certain analysis of the changes in the image at each iteration. The proposed adaptations resulted in decrease of image distortion by more than 50% in terms of rms metric, making it possible to estimate the number of iterations required for better performance of the algorithm.