An enhanced algorithm for Character Segmentation in document image processing

Optical Character Recognition consists of various steps like skew detection, segmentation of columns, lines, words, and characters before feeding the isolated character to an optical character recognition system. Several methodologies are followed to perform these steps using conventional Hough Transformation. In this paper, a new algorithm is proposed to perform all those steps involved in document image processing. The algorithm is implemented for skew detection, column and line segmentation and Character Segmentation. This can be extended to all other steps like character recognition. The novelty of this approach lies in “the consideration of any image, as one formed by several black and white lines of various lengths and at various angles”. The pixel values of the binary image are stored in an array. All the pixel values in the array are compared with their horizontally adjacent pixel values, row by row, for the presence of collinear points (i.e., a line). It is done by detecting the continuity of either the white or black pixels accordingly. Once the continuity is detected, the starting and end coordinates are displayed as an intermediate result. A new image will be generated as a result, which indicates the pixel area of line, identified from the input image. The algorithm is applied for English and other regional languages.