Blind Deconvolution for Image Restoration Based on Text Characteristic

Along with the rapid development of computer technology and digital image processing application, words are frequently shown as images too. For little priori knowledge, blind image restoration should be used for blurry text image. We improve the blind deconvolution algorithm, one of blind image restoring methods, combining the text-specific bimodal characteristic about histogram with blind deconvolution method by means of probability and statistics theory. The improved algorithm increases the computation speed and optimizes the visual effect of recovery text image largely. It is verified that the advanced image restoration algorithm has better recovery capacity and certain practical value through the analysis and simulation of some typical experiments.