An efficient algorithm for text localization and extraction in complex video text images

This paper gives an efficient algorithm for text localization and extraction for detection of both graphics and scene text in video images. The Text size is a vital design parameter whose dimension should be properly elected to make the method more robust and insensitive to various font shapes and sizes, styles, color/intensity, orientations, languages, text directions, background and effects of illumination, reflections, shadows, perspective distortion, and the density of image backgrounds. Basically, it works in four steps: Edge generation using Line edge detection mask, Text localization using projection profiles based method, Text segmentation and Text recognition. The result of this paper demonstrates the capability of the proposed technique there by conducting experiments on the video images containing on a large group of complex video text images. The paper proves to be robust for various background complexities and text appearances. The method used in the paper has a high rate as good extraction result. This paper gives a method which has a high rate of good extraction results in complex video images and experimental results. The proposed technique gives better result than existing methods in terms of detection rate for large video image database with very few false alarms, reliable recall rate and precision rate.