Dynamic Hand Gestures Control System Based On JZ4755 Platform

With the rapidly development of computer and network, a simple, convenient and natural Human-computer Interaction system is the field of human-computer interaction research and the trend of development. Gesture recognition as a kind of good experience with human-computer interaction technology is a hotspot of research on the current domestic and international numerous scholars. Gesture recognition technology is mainly used in the virtual environment of human-computer interaction and intelligent control, the gesture recognition in listed overseas products are Microsoft launched launched the device and Leap Leap Motion, etc., but these make gesture recognition application need to rely on high-end hardware equipment, which can lead to costly, has not been applied widely. Research, therefore, a low cost, simple and convenient to use the human-computer interaction system based on gesture recognition not only has important theoretical significance, but also has a broad market prospect and a high practical value.In this paper, in order to reduce the hardware cost and improve the real-time performance, for the purpose of analysis and research on the embedded system of video acquisition, gesture image preprocessing, segmentation and gesture trajectory tracking and recognition of the basic theory, designs and realizes dynamic hand gestures based on JZ4755 platform control system. The system can collect real-time gesture first video frame, and analyses the meaning of the gesture command, gestures map into infrared yards to send television operation, so as to realize the function of gestures to control TV sets. Research work of this article mainly has carried on the research and analysis from the following several aspects:(1) gestures video acquisition and preprocessing module, based on Linux video capture mechanism and combined with the characteristics of system platform, the video frames with V4L2 interface first acquisition gestures, then for each frame data using interval sampling method is used for data reduction to optimize the amount of data processing, improve the subsequent processing time; (2) the gesture segmentation and recognition module, each need separate pixels of the pixel value construct index number, according to the index number of Gaussian Color Hash Table lookup results determine whether the pixel for color points, thus replacing the mixed Gaussian color model calculation in the process of floating point arithmetic. Collected gestures frame 2 values into a binary chart of skin color and the color of skin, then the xor algorithm of binary chart with the frame to extract showed both skin color characteristics and show the dynamic characteristics of the movement of the color of skin area characteristic domain, according to the observed movement color features an area of mutation characteristics and movement of the color of skin area centroid point movement characteristics determine the gestures results; (3) the system verification, this system the raw video frames in the resolution of 640 * 480 * 2 byte conditions, can achieve 23 FPS recognition processing speed. By experimental verification shows that this system has high real-time performance, the algorithm of hardware platform dependent on low and humanized operation etc.