Skew Correction Function of OCR: StrokeWhitespace based Algorithmic Approach

As the technology evolves, emergence of Optical Character Recognition (OCR) for both printed and handwritten documents of any language is obvious. In the process of developing an OCR for languages like Hindi, Bengali, Marathi that are among top 15, the most spoken language of the world, the task of skew correction still remains a challenging one as fewer research has been carried out in the field. In this paper, we confront this challenge and describe a stroke-whitespace based algorithmic approach that harnesses horizontal projection technique to correct the skewness of writings precisely for these languages. The paper proposes an easier and effective process named as OJ method that corrects the skewness of images for any degree of rotation. In essence, the paper deals with the images that are rotated by 180 ̊, using Stroke-Whitespace distance method. General Terms Optical Character Recognition, Handwritten OCR, Skew Correction, Stroke-Whitespace based text image rotation, Pattern Recognition.