Skew Angle Estimation of Urdu DocumentImages: A Moments Based Approach

Abstract

The performance of an OCR system will not be satisfactory for most of the scanned images without accurate skew correction. This paper presents the skew angle estimation and correction for Urdu document images script using moments method. The basic idea is to draw a random polygon over the text in document. This leads to thinning free preprocessing. The skew angle is calculated using Central moments and centroid of the document image. Experimental results are found to be satisfactory and compared with other skew detection techniques.