Development of a novel technique for skew detection, estimation and correction of scanned document

Abstract

We propose a novel technique for automatic skew detection, estimation, and correction. These defects are introduced in a document while scanning or capturing its picture. The process of automatic detection, estimation, and correction is involved in many interconnected operations. In the present work, we are able to remove any skewness introduced in a document with or without images, having bar codes, bar graphs, pie charts, logos, tags, tables, etc. The estimation of skew angle is calculated with help of probabilistic distribution of white and black pixels. Once the estimation is performed, the correction is carried on original document. Our technique is applicable for wide range of documents, which was not possible with state-of-the-art techniques available.