On Skew Estimation and Correction of Text

Abstract

This paper, in addition to reporting some existing techniques, proposes some new techniques for skew correction. It includes two novel document skew detection algorithms based on histogram statistics and connected component analysis. The histogram based algorithm works much efficient for detecting skew angle in which we analyze the lines as peaks and valleys on histogram. The connected component analysis is based on finding the connected components within a single line and considers them as one blob to estimate skew angle. The two methods have been experimented on various text documents and very promising results have been achieved given more than 99% accuracy. A comparative study has been reported to provide a detailed analysis of the proposed methods.