Research And Implementation Of Image Character Recognition System Based On Mobile Terminal

With the mobile operating system constantly upgrading, all kinds of applications based on mobile terminal devices is also constantly produced. A system of optical character recognition designed and implemented based on mobile terminal device will bring our daily life great convenience. Conventional image-based text recognition methods are mostly implemented on PC terminal. Although optical character recognition technology is quite mature, it can only provide the service of character recognition in a fixed position, which is not convenient for users. Optical character recognition technology based on mobile terminal is currently not mature, character recognition process requires a long time, and can only recognize a single language. Therefore, how to implement the text-based image recognition technology based on mobile devices will be more practical significance.Based on the above analysis, we designed and implemented image character recognition system for mobile terminals. The system not only implements the Chinese and English bilingual character recognition but also improves the character recognition rate. Specific steps are as follows:First, get real-time source images photographed by the phone camera, or read source image directly from the phone local gallery. Then preprocess the source image, including image cropping, image grayscale and image binarization. Among them, image cropping calls gesture cropping function provided by Android system; Gray-scale image processing is able to transform color image into gray image by using a well-known psychological formula, therefore the resulting image is closest to the human eye perception of grayscale images; the threshold used in image binarization is produced by iterative algorithm, the algorithm is efficient and binarization effect is also very good. Finally, Tesseract engine is applied to recogize the preprocessed image, and then the results are displayed in the user interface, and serve the user the function of modifing the recognition result. The system’s overall performance analysis of test has been accomplished. The results show that the proposed iterative algorithm is superior to the existing histogram algorithm and one-dimensional Means algorithm. In the test phase, the system integration test and analysis have been fulfilled. The experimental results show that the system can recognize the image text accurately and rapidly on the mobile terminal. It also can provide customers optical character recognition services anywhere and anytime.