A Effective Algorithm for Skew Correction in Text Images

Abstract

This paper proposes an effective corner detection method for skew correction in text images. The idea of corner detection is to build an RGB background mapping function based on similar features of the same region in the background. The RGB background mapping function separates the target area from the image and determines the corner points. Finally, the target area is corrected according to the corner point using perspective transformation. From the experimental results, it is shown that the proposed method can correctly make the corner point detection in the image and produce easy-to-read text images after skew correction.