Detect And Localize Text In Natural Image Using Wavelet Transformation

With increasing use of capturing device in recent year. Content based image serves as important clues for many image based application. Scene text contains significant and beneficial information. Detection and localization of scene text is used in many applications. In this paper, a new approach is developed which locate text in different backgrounds. However, variation of text due to differences in size, style, orientation and alignment, as well as low image contrast and complex background make the problem of automatic text localization extremely challenging. The text localization algorithm system is designed to locate text in different kinds of images. Firstly, the color image is converted into grayscale image, and apply segmentation algorithm for segment gray scale image. After that, Haar Wavelet Transform (DWT) is employed. Haar DWT decompose image into four sub image coefficients, one is average and other three are detail. Which is find the approximately text area in the given image (confident region). After that apply binary and unary properties of conditional random field (CRF) on confident region image for connected component labeling finally, using some specific condition, the text is obtained in bounding box.