Research On Mobile Phone Automatic Operation Based On Gesture Image

With the rapid development of hardware and software technology, the human-computer interaction technology by the mouse button to The Times that represented by voice and gesture recognition technology of natural interaction, the gesture recognition technology has become a more natural and intuitive way of input, can obtain better man-machine interaction experience. The Android mobile phone operating system favored by people, the intelligent mobile phone function as a micro computer processing data, the developer can make more complicated on the mobile phone operating system fast development, base on gesture recognition technology on the Android operating system has become a new generation of research direction, and will break the traditional concept of mobile operating mode.The present Paper study the palm segmentation, tracking and gesture recognition, and the android smartphone as a platform, as gesture Android music player for a example, illustrates detaily the gesture image automatic operating mobile phone.This article main study the content as follows:1. The gesture recognition technologyFirstly, introduces the method of commom palm detection,analyze the merits of the demerits of frame differential method and the color of skin detection, put forward the frame differential method combined with color extraction of palm segmentation method, then through the morphological processing, finally segment the palms; Then introduces the CamShift palm tracking method and analyze the shortcoming of CamShift algorithm and submit an improved CamShift algorithm, the improved CamShift algorithm not only automatically adjust the size of the search window to adapt to the tracked target in the image size, but also use the current frame palm prediction in the next frame size and the size of the location, the improved CamShift algorithm effectively solve the palm tracking which is interferenced by the large-area skin color.As result, using the algorithm of convex hull and Freeman Code extract the edge profile of palms, and using AdaBoost classifier to train the palms template, and using the Hausdorff torque of the contour matching algorithm and improve CamShift algorithm palm tracking to dentify the palms.2. The application of gesture image in Android phone platform.Firstly, This article introduce the Android gesture image algorithm platform construction process, then introduce OpenCV algorithm transplantation process of the mobile phone. In the system realization section, this paper briefly introduce the mechanism of Android applications, and detailedly introduces the Android JNI implement communication between the underlying gesture recognition and the top Android applications, and then use AdaBoost algorithm training the positive and negative gesture samples, and defines four gestures semantics which control the Android music player, finally realized the funcion of gestures control music player, the experiment results show that four kinds of gestures’ average recognition rate is92.5%.At the end, discuss the real-time feature of system, analysis shows that the system has lower memory consumption and meets the real-time requirements.Finally, make the summary and illustrates the main research result, At the same time pointed out shortcomings and problems which need to be further studied.