Text Location And Recognition In Natural Scene Images Based On Mobile Terminal

With the growth of people's living standard and the development of Multi-Media technology, the mobile terminal product with a built-in camera has been in widespread use, smartphone is the most typical representative, which makes it easier for people to capture natural scene images, in addition, extracting and recognizing texts from natural scene images have a broad application prospect.The variations of text in terms of character font, size, style, orientation, alignment and color, as well as complex background, uncontrolled illumination and noise of images make the text location and recognition very hard.Based on mobile phone, this paper has done a research on text location and recognition in natural scene images, and a complete text analysis system based on mobile phone has been proposed. The main content is as follows:1. Development of Camera based on Mobile PhoneCombined with DirectShow technology, the camera of Windows Mobile phone is developed, and two image capture schemes are proposed to acquire images from natural scene.2. Text Location in Natural Scene ImagesWith adequate consideration of the limited hardware resources of mobile phone and the interference factors, a fast text location method based on the edge and connected domain is put forward.3. Text Recognition in Natural Scene ImagesThe text regions which are got after the location are to be binarized, segmented and normalized, then the feature of the direction of line elements is extracted, the final recognition results are given by the Euclidean distance classifier.