Design Of Embedded Motion Sensing Remote Control Based On Hand Gesture Recognition Technique

With the rapid development of computer technology, the way of human-computer interaction(HCI) has been transferred from single model to diverse ones and inclined to more user-friendly tendency. Approaches of Computer Vision based interaction have currently become the focus of research. Due to the versatile interpretation and flexible use of different expressions have the advantages of practicability and universality compared with other Computer Vision based interaction, the relevant techniques of hand gesture recognition have deemed as the key ones.This paper designs and implements a monocular vision based hand gesture remote control system which consists of camera, FPGA, PSoC infrared communication module. The specific work of this paper presents as below.The correlated algorithms research includes hand gesture detection, tracking, segmentation and feature extraction methods. After comparing and analyzing several classical algorithms, an adequate approach has been proposed. This paper employs Codebook method to construct the background model and Camshift algorithm combined with Kalman Filter to track the hand motion. For moving hand gesture, it extracts the features of motion vectors to recognize; and for static hand gesture, it extracts the features of hand contours to recognize.The algorithms are implemented on PC and embedded processor platform. They are verified by Intel’s computer vision library OpenCV. And then several popular embedded hardware platform are introduced and compared with each other. Finally, the system uses Altera company’s FPGA development board DE2as the hardware development platform. The paper introduces development and application of the soft core processor Nios II on video and images processing in SOPC environment. It gives out the realization of video signal acquisition, storage, processing and the output to VGA monitor. Besides, it illustrate the design of infrared communication module based on PSoC via serial port connected to FPGA. In order to take advantage of parallel processing and hardware reconstruction techniques of FPGA, part of hand gesture recognition algorithms are implemented to hardware circuits in FPGA.The experimental result indicates that this design can successfully track and recognize hand gestures in real-time monitoring. Accordingly, It lays the foundation and provides experiences for advanced improvements on this design.