Abstract
The process of obtaining digitized copies of documents has become an indispensable part of many businesses around the world, including but not limited to Banking, Insurance and Telecom domain. The document capturing part is usually done with the help of scanning devices that have evolved from being fixed flat-bed ones' to hand-held devices. While capturing documents using the ADF scanners, skew related defects usually get inserted in the images. These defects degrade the legibility of the documents and hence the accuracy of the OCR and ICR extraction engines. The degraded accuracy of the extraction engines inversely affects the businesses that capture critical user information. Thus, it becomes necessary to correct such errors. Since the skew detection part holds a small but significant part in the whole process of form processing, it has to work accurately and efficiently to enhance performance of complete system. Here, a new faster content based approach for skew detection on document images is proposed. The proposed approach considers only the printed text present in the image, for skew detection, thus eliminating the errors arising due to the pictorial information, noise, scanning artifacts, etc.