提出了一种保护隐私的人群大小的估计，不依靠跟踪和计数的方法。-present a privacy-preserving system for estimating the size of homogeneous crowds, composed of pedestrians that travel in different directions, without using explicit object segmentation or tracking.

采用了Harrtraining和离子滤波算法实现了拥挤人群人头计数，非常有用。VC++和Opencv开发，采用Oepncv中的Harrtraining训练一个分类器即可运行-Human detection and tracking in high density crowds is an unsolved problem. Standard preprocessing techniques such as background modeling fail when most of the

巴基斯坦人做的拥挤行人检测与轨迹跟踪的硕士论文，基于opencv-Master degree thesis about DETECTION AND TRACKING OF MULTIPLE HUMANS IN HIGH-DENSITY CROWDS