The Design And Implementation Of Distributed Motion Detection And Tracking System

Visual Sensor Networks are widely applied in the field visual surveillance system, real-time human posture reconstruction-analysis, environmental and ecological monitoring, health monitoring, home automation and traffic control. Smart camera networks gradually gain its attention in modern world. As opposed to the traditional centralized visual surveillance under human’s supervision, Smart camera networks has the advantages that:high computing speed, needing no human supervision, computing visual data inside camera nodes, low network load high performance of the surveillance system. Smartcamera networks mainly based on the camera to capture the visual data and then perform the analysis of human posture. As a hot issue in visual research field, tracking is the base technology in visual computing. How to perform target tracking directly has greatly influent the performance of smart camera networks and the accuracy of human posture analysis.Motion detection a basis of target tracking, is also a basis of visual computing. In this article, we mainly focusing on the motion detection and tracking technology. Combined withsignificance detection we improved the performance of the motion detection module, improved the performance adaptive Gaussian mixture-model based motion detection module and enhance the accuracy of adaptive Gaussian mixture-model algorithm. At last we use the particle filter algorithm to implement the tracking module ofdistributed motion detection and tracking system. The content of this thesis includes the following aspects:1) This thesis studies motion detection algorithm based on adaptive Gaussian mixture-model as well as significance detection based on global contrast, and implement the combination of this two algorithms. The significance detection algorithm helps the adaptive Gaussian mixture-model to improve its accuracy and also the speed defect of significance detection.2) The thesis also studies the tracking algorithm based on the particle filter algorithm. The advantages of this algorithm, robust and strong tracking performance, enhanced the tracking module of the system.3) This thesis designs and implements distributed motion detection and tracing system. Requirements analysis performed in the thesis helps us to determine the functional requirements, including distributed motion detection unit and target tracking unit.The thesis combined different type of algorithms to form distributed motion detection and target tracking system. By the cooperation of different types of algorithm to improve the performance of the system.