Dynamic threshold algorithm for removal of Back-to-Front noises of visual document image

As we all known, the Back-to-Front noises may affect the feature extraction and classification of text when using ORC to identify. In this paper, we take advantage of dynamic threshold method to process the vision document image of Back-to-Front noise with unimodal or bimodal histogram characteristic. Experimental results show that this algorithm has a better effective.