An approach for Skew Detection using Hough Transform

Detecting skew of any document image and correcting that are important issues in the preprocessing stage of OCR system. The Hough Transform is a technique that performs skew detection in the document images. In the present work, voting is done on the basis of angle from 0 to less than 90°. Moving from one angle say Θ1 to Ө2 , five partitioned are considered i.e. there would be 450 classes. Voting is the process to find the belongingness of a pixel to a particular class. Finally, each pixel present in skewed image is found to which class it belongs. The class which has maximum count of the pixels is taken as skewed angle class.Performance of our algorithm is analysed. It gives increasing results.