Skew Detection Techniques used in Degraded Document Images
Snehal S. Kolhe, K. T. Jadhao · Zenodo (CERN European Organization for Nuclear Research) · 2016
In this paper skew detection technique for printed and handwritten devnagri scanned documents is proposed. Recognition methods for printed and handwritten texts in Scanned documents are significantly different. Skew detection technique consists of processing steps like preprocessing, segmentation, skew correction and detection with enhancement which is used to identify writer. Segmentation is used to extract text lines and words from handwritten and printed documents. A horizontal projection and vertical projection algorithm is used to segment the document into a number of lines. Nilback, Sauvola's and wolfs algorithm is used for binarization. Linear Otsu thresholding is used for filter. The scan line method is used for the skew detection and correction .Adaptive histogram algorithm is used for image enhancement. Various Printed and handwritten text documents are scanned for the different preprocessing methods. In segmentation, for the Line, word, character segmentation we used the separate algorithms. For the skew detection and correction, skew line method gives the better accuracy.