本文介绍了如何通过自适应阈值选择和区域分割来分割运动目标区域，进而利用霍夫变换在每个运动目标区域内检测和定位人类头部。一旦头部被发现，就很容易找到场景中的人体，因为两者之间存在对应关系。因此，采用此方法可以在场景中有效地检测和定位人类身体。
Human head detection
We use adaptive threshold selection and region segmentation to segment the moving target region. Then human head is detected and localized in each moving target region using Hough transform (the contour of head should be a circle or arc of a circle). Once the head is found, it is easy to find the human body in the scene because of the corresponding relationship. So with this method, we could detect and localize the human body in the scene effectively.
overflow of tracking
Target tracking with combination of Mean-Shift and Kalman filtering
demonstration of experimental results
扫一扫
3094
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
