Human Detection And Tracking System Design And Realization Of Nighttime Infrared Video Image

Intelligent video surveillance in the social security field is widely used, under normal circumstances the application scenarios are color cameras, color cameras, but strongly affected by the light in the evening light situations, the camera captured video quality is rather poor, to correct target monitoring had an impact. The infrared camera uses thermal imaging principle, the next night in a bad light or foggy weather conditions are also very good imaging results. Than color infrared surveillance monitoring machine has a distinct advantage, infrared surveillance equipment at night through infrared imaging principle, be able to work all day. This thesis is mainly through the study of the human body infrared video object detection and tracking technologies proposed algorithm to fit the design and realization of human target detection and tracking system.The research is mainly purpose to design a real-time human detection and tracking system at night for the purpose of proceeding from reality, the use of nighttime infrared camera equipment acquisition, a lot of experiments; through several structures and related algorithms to improve the system, using the most appropriate infrared video algorithm design the prototype of this system. In the video image de-noising processing stage, the main use of wavelet threshold for image denoising processing, reducing the impact of noise on useful information. Body moving target detection phase, the first use of the algorithm is simple and easy to implement background subtraction method using Gaussian mixture model to background modeling, the prospects for the separation of the video in the identification of human targets judged by SVM separator. Human target tracking using Kalman filter stage and combining MeanShift tracking algorithm to track targets on the human body, as a result of the Mean Shift tracking Kalman filter observations and experiments was carried out through a number of improvements, has been very good resultsOpenCV and VS2010 are used to build experimental platform to realize moving objects detection and tracking.System testing proved that the system can effectively detect and track the movement of human targets.