Application of blind image correction methods to Orlan hardware and software solution

This article provides an analysis of the algorithm for blind deconvolution of distorted images to determine the possibility of its application to computer vision systems. The original and blurred license plate images were analyzed by a computer recognition system used in Orlan complex. By studying the results, optimal parameters for image reconstruction were found. These parameters allow to obtain the maximum accuracy of license plate recognition. Compared to the other recovery algorithms, the method under consideration shows better results. Computational complexity of the considered algorithm allows it to be used in real time computing. The article also presents data on the comparative performance analysis of various image correction algorithms. The experiments show that the proposed algorithm can be used in embedded systems in order to improve the accuracy of subsequent analysis and classification of images.