Limb Motion Detection Based On FPGA

Limb motion detection in people’s daily life applications everywhere, and therefore the precise limb movements of people detected has important theoretical significance and economic value. Although the current theory of limb motion detection algorithm has matured, but the hardware implementation of these algorithms is still a hot topic of current research. FPGA (Field Programmable Gate Array) as a common hardware platform that is ideal for use in video and image processing because it can reach people in the video processing system for speed, integration, and reliability requirementsThis paper build FPGA-based video detection system limb movement to achieve a real-time motion detection video image is displayed in the upper body limb. The system includes video capture, image caching, pretreatment video, limbs moving target detection, video display module five core modules, including limbs moving target detection module for the system. In the limbs moving target detection module, first of all, based on frame difference algorithm to achieve the video moving target detection. On this basis, according to the data memory block frame, the skin will be applied to the pixel detector detects body movement detection module, each pixel VGA skin pixel detector must be to determine whether the pixel is a skin. After the detection of skin pixels, in order to better suppress noises, improving the stability the640×480frame is converted into the40×30samples, and the resulting image after down-sampling is very stable, and then after segmentation within a small pixel value averaged, and the result is added to the sample of a16×16pixel block, the final area of skin in order to reduce noise, its average space. We identify the skin pixels based on upper limbs, mainly head and left, rear right arm, respectively, to find its center of mass, with a16x16pixel block red marker out，and then with themovement of the head and letf arm to produce a smooth animation, an accuratedetection of upper limb movement in the video image.The system is in a static background can quickly achieve limb moving targetdetection, the light did not change significantly in the environment can be wellapplied, The next step will be to achieve a dynamic background moving targetdetection and tracking.