一种用于数码相机自拍系统的人体检测方法

在研发一种数码相机智能自拍系统的基础上，提出了一种简单有效的基于静止背景下目标人体的检测方法。该方法基于实时帧与参考背景帧的差分建立背景模型，进行前景目标分割，并基于颜色向量夹角消除阴影的影响，最后用投影阈值分析提取目标人体，并进行定位。该方法计算量低，环境适应性强,能满足数码相机自拍模式的目标跟踪要求。

Method of tracking human body for the self-timer system in the digital camera

Based on designing an intelligent self-timer system for the Digital Camera, a simple and effective approach was given to detect and locate the human body based on the still background. The background model was built based on the difference between real-timer image and reference image, in terms of which foreground objects were segmented. Then shadows were removed according to the contained angle of color vector. Finally threshold of the projection was used to extract and locate the human body. This algorithm can meet the object-tracking requirements of self-timer with low computation and good adaotability.