Method for Line Segmentation in Handwritten Documents with Touching and Broken Parts in Devanagari Script

Now days, a vast research is going in Optical Character Recognition (OCR) of handwritten Documents in Indian scripts. A lot of handwritten data is existed in Devanagari script which is still to be recognized. Segmentation is the key step of OCR process. Segmentation is the process of extracting the valuable segments from the text document which are used in the process of recognition of characters. Line segmentation is the process of segmenting the text document into lines. Afterwards, word segmentation and character segmentation is carried out. This paper only deals with the Line segmentation of handwritten documents in Hindi. Devanagari script is the basic script to write Hindi, Marathi, Sanskrit and Nepali languages. In this paper the brief introduction of various existing techniques for segmentation of handwritten text is discussed. Also, develops an algorithm for segmentation of skewed lines, touching lines present in the text document and broken parts in upper modifiers or space present between the upper modifiers. This algorithm is implemented on large database collected from various writers. The proposed algorithm integrated the Projection based method, gap detection between text lines and neighbor pixel analysis method