Connected Component Feature Analysis Based Handwritten Uyghur Text Lines Detection And Separation Algorithms

Text line segmentation is a hot topic towards handwritten document, it providescrucial infromation for the tasks of text block segmentation, keyword matching,character segmentation and recognition. The effective segmentation would derectlyaffect the tasks of extraction and recognition.The Uyghur handwritten document contains many kinds of combination towardsmain strokes with additional strokes and handwritten style. Handwritten textlines areoften un-uniformly skewed and curved, full of many additional strokes and small gaps.The statistic and analysis of connected components is a basic job for the segmentation.This paper focus on segmentation of handwritten document and make a deep research.more details are as follows:1. To deal with the issues of text lines segmentation in Uyghur handwrittendocuments which contains many additional strokes, we proposed a classificationprocessing method which based on connected components.2. Taking the characteristics of main neighbour text lines which located veryclose, we propose an adaptive piecewise algorithm.3. To overcome the problems of text lines construction, by using text linesdetection of post-processing, a merging technique is applied to correct possible falsealarms, text lines that did not detect and touching problems between two connetedcomponents are revised, at last, the center of gravity radiation scanning method whichsolved the belonging problems for remaining small strokes.4. A tool soft ware for Uyghur handwritten documents is developed, and manyhandwritten documents are segmented. The processed results are evaluated in detail.