The Research And Implementation Of Object Tracking Algorithm Based On PTZ Camera

Tracking moving objects in computer vision, pattern recognition and artificial intelligence is an important research subject, is widely used in security monitoring system. Target motion because of the diversity and complexity of the environment, including shelter, target deformation, illumination and other factors, how to accurately track a moving target in real-time target tracking algorithm is to design difficulties, it is a challenging field of computer vision of the subject.This paper studies based PTZ cameras (dome camera intelligent integration) of the moving object tracking algorithm, based PTZ cameras. Mean Shift-based method for target tracking and the ability to meet the objectives of the stability of the shape and size of continuous change, and fast computing speed, anti-interference ability, to ensure the system’s real-time and stability, so the static scene in the main study tracking algorithm based on Mean Shift and the algorithm was improved Mean Shift Method and the Kalman filter algorithm fusion, and target characterization by spatial color histogram features and enhanced tracking robustness.Finally,the paper build a prototype of the technology system,designing a high-speed do dome cameras in horizontal,vertical,axis in three directions to do intermittent movement of the control strategy.Ball machine by adjusting the P/T direction of attitude to be tracked object is always visible,when the vision center at the target,the zoom control to start the ball machine for the purpose of image scaling,so you can observe the target or capture the local details.As to the zoom control for the ball machine adaptive tracking window size problem,the use of SIFT feature matching algorithm to calculate the change rate of the method of the ball machine.Ultimately,the paper shows the effectiveness of technology for the actual scene by a large number of experimental tests.