基于立体视觉的人体头部目标识别研究

使用立体视觉系统可确定任意物体的三维轮廓及其表面任意点的位置信息和深度信息。本文将立体视觉运用于人体头部目标检测，设计了一种基于图像深度信息处理的人体头部识别系统。系统采用Xtion摄像机采集场景的深度图像，并对其进行特征分析，根据深度图像的特点以及头部的特征确定头部目标区域。再对目标区域采用Mean Shift算法进行聚类处理，得到清晰的图像边缘。最后通过基于动态阈值的一维熵函数分割法实现头部的分割识别。该系统可快速锁定目标区域，减少了算法的计算量，大大提高了系统的识别速度。此外，采集深度图像的xtion摄像机悬挂在目标场景的正上方，因而较好的解决了目前人体目标检测受遮挡的问题。经实验论证，该系统有较高的识别精度。

The research of human head detection based on stereo vision

The three-dimensional contours of any object including location and distance information can be determined by using stereo vision system.In this paper,the stereo vision is used in human head detection by the system based on depth image processing.Firstly, the characteristics of the depth image collected by camera of Xtion are analyzed by the system. Then head target area is determined according to the depth of the image characteristics and the characteristics of the head.For the determined area, Mean Shift clustering algorithm is used to get a clear image edge.Lastly, head split is realized by using One-dimensional entropy function based on dynamic threshold. Through rapid target, the computational algorithm can be grately reduced and the speed can be greatly improved. Moreover, the issues that the accuracy of current target detection method is easily affected by obstacle can be solved by the system.It has been demonstrated by experiment that the system has high recognition accuracy.