Comparative Analysis of Text Detection using SVM, KNN and NN

Text objects occurring in image can provide much useful information for content based information retrieval and counting applications, because they contain much minute information related to the documents contents. However, extracting text from images and videos is a very difficult task due to the varying font, size, color, orientation, and malformation of text objects. Although a large number of text extraction approaches have been reported in the past work, no specific designed text model and character features are presented to capture the unique properties and structure of characters and text objects. We have proposed SVM, KNN and NN Techniques in our research. The thesis was focused on the design of an algorithm to detect the text from images, extract it and convert it into speech for those who are unable to read and its implementation on MATLAB