A fast pedestrians counting method based on haar features and spatio-temporal correlation analysis
Dong Hao, Xue Feng, Fan Wu, Yong Chengxi · 2015
In recent years, with the development of computer vision technology, pedestrians counting is widely used in traffic, business and other applications. As the traditional pedestrians counting methods are susceptible to the influence of occlusion and large amounts of calculation, real-time performance and accuracy can not be solved very well. In this paper, we propose a novel method to fulfill pedestrians counting task accurately. First, we use Haar features and Adaboost algorithm to get a head classifier by sample training, which is consequently used to detect pedestrians in a predefined strip region inside the input video. At last, a statistics model called Spatio-temporal Correlation Analysis is designed to implement pedestrians tracking and counting. The experimental results show that our method is of low cost, high accuracy, and can be applied to many applications.