基于视频流的运动人体检测和跟踪的研究实现

摘要

The detecting and tracking of human movement is a central part of video monitor. A method of detecting and tracking real - time is put forward. Background model is got through the background frame differential method and the frame difference method. Kalman filter is used for tracking. Then, the detected human burly is matched to the human body model. If it is correct, the background model is updated. Otherwise, tracking is continued, and the background model is updated.%运动人体的检测和跟踪一直视频监控的核心部分.提出一种将检测和跟踪实时结合的跟踪方法,检测时采用背景帧差法和帧间差分法得到背景模型,跟踪时采用Kalman跟踪,将跟踪得到的人体和检测到的人体匹配.如果正确,则更新背景模型,否则继续搜索跟踪,及时更新背景模型.