| With the rapid development of image process technology and computer process power, vision tracking emerge as one of the most significant research branches in robotic technology. Specially, object tracking under infrared spectrum with its merits of non-illumination influence and insensitive to bad weather conditions becomes more and more populous in recent years. It is widely used in industry, medical care, aerospace industry, military etc. The moving object tracking under consecutive video images is the most important one. Because, it is vital to the applications in robot navigation, intelligence surveillance system, medical image analysis, industrial inspection, video image analysis and military radar video signal process etc. Through the detailed research and comparison of basic image process technology and vision tracking theory, methods, we adopted a Camshift algorithm to perform moving object tracking process under infrared spectrum, and also improve the deficient of the very algorithm.The main research of this paper is given as follow:Firstly, analysis the characteristics of infrared moving object, and the influence factors to moving object tracking, perform tailored image preprocess to the infrared images, to make preparation to the following tracking process.Secondly, on the basis of the existing color feature extraction algorithms, we make some improvements to fulfill the object tracking and infrared spectrum. By making use of the Kalman filter prediction and inverse projection's weighting process, we achieved a relatively high object tracking speed and insensitivity to large sheltering.Finally, we constructed computer simulation and test experiments to the proposed algorithm under Visual C++ based on the platform of Opencv. During the experiments of the proposed algorithm,we make the conclusion last: this paper's moving objects tracking system have a satisfied real time property and could perform moving objects real time test and tracking accurate. |
PDF Full Text Request