A Hybrid Approach To Detect And Recognize Text In Images
Poonam B.Kadam · IOSR Journal of Engineering · 2014
Text detection and recognition in images is a research area which attempts to develop a computer system with the ability to automatically read text from images or videos.This problem is challenging due to the complex background and large variations of these components features like color, size, shape, orientation or texture.Here new proposed framework is explored which can automatically detect and recognize aligned text from urban scene images e.g.shop name,landmark,street.The proposed framework evaluated on new generated dataset.It consist of a three main step 1) image partition perform to segment text based on color information.2) character grouping to detect text character in every text string depend on character size differences, distance between neighboring characters.3) the detected text recognition using neural network.This proposed method efficiently and accurately detects and recognizes the text with a low false positive.