A distortion correction approach on natural scene text image

A method of distortion rectification on natural scene text images is presented. The method consists of two steps: 1) endpoints are extracted from the classified tip points using mathematical morphology operators, and then these tip points are classified in accordance with the location of the line using the nearest neighbor method. Secondly the text baseline is obtained based on the least-square method. Finally the RANSAC estimation method is used to select the optimal baseline and from which the line corresponding to the vanishing point is fitted. 2) The deformation parameters of images which contain in homography matrix are calculated, and then the front view image is achieved by homography matrix. The experiment results show that the recognition rate in deformed scene images has improved significantly after image correction using this method.