Text Detection and Extraction from Document Images using K-Nearest Neighbor Rule

Text extraction and text line detection is the foundation of document image analysis. Since many years, a large number of text detection methods have been proposed, where these methods depend on convinced assumptions of documents with various font style, font size, distorted text, uneven lighting, complex background and low resolution. In this paper, reveals k-nearest neighbor rule as a generic text-line detection and text extraction approach that can be applied on a complex mail document images. The performance evaluation of transition map generation and it compares with other two models is presented in this paper. Experimental analysis shows that image based text Optical Character Recognition (OCR) method is to extract the text from the colorful image and detection of advertised mails is very efficient than that of the other existing methods. KeywordsText extraction, Text-line detection, KNN rule, and mail document image.