The Design And Implementation For Big Ear Eva Image Recognition System Based On Android

Along with the increasing popularity of mobile terminals, mobile operating system has been becoming more and more mature, especially android system having the most prominent performance. At the same time a lot of applications based on Android system is developed and further improve people’s lives. Enterprises are expanding its products to the mobile terminal for improving the users’experience and attracting more users. So the developing of "Big ear Eva" graphic recognition system is put forward. It will connect the story of the book’s page and the video though recognition on mobile terminals.The system is the combination of Android and graphic processing technology. This paper first introduces composition structure of the android system and the composition of program development, and discusse restrictions and support for the image application on the android mobile. At the same time, this paper introduces some methods of image processing technology, for example image enhancement, image segmentation, image denoising, image correction, etc.The main work of this system is to design and implement the four modules of the system: encoding module, image acquisition module, image recognition module, decoding and displaying module. The image recognition module is divided into icon recognition module and the page number recognition module, the two modules are the core of the system. Coding module builds the relationship between image feature and the page number by encoding and rules the corresponding relation between page numbers and videos. Image acquisition module uses the Android development technology to call mobile hardware, previews and saves image, and makes the format conversion from YcbCr to RGB for further processing at the same time; Icon recognition gets the page number by the order of icons. This paper proposes a method called "kuang" to extract and segment icons. It can quickly obtain icon position and split out the statistical domain of the color of the icons in a unified specification. Icons are determined by the proportion of color, further order for icon can be get. The page number recognition is the recognition of digital. Considering the fact of digital in a ellipse icon, the correction and extraction of digital positioning will fully make use of the geometric characteristics of the ellipse. This process greatly simplified and accelerated the extraction of the digital domain. Then this paper will use template matching and logic analysis to confirm the page number. Decoding and playing module will use the correspondence corresponding in the image characteristics、the page number and video address to play videos.The system can run well in the android system. Through testing, the recognition rate of the two kinds of recognition methods are over98%and90%, and can meet requirement.