Study Of Pedestrians Counting Based On Face Detection

With the development of economy and technology, the demand of pedestrians counting inmany places is growing. While most of public places are configured with video surveillancesystems, especially there are camcorders installed at the entrance of these places to shoot peopleface for monitoring. This paper takes full consideration of this fact, and focuses on the study ofthe pedestrians counting technology based on face detection of the inclined still camera.Considering that the face can be used as an important feature to distinguish between humanand non-human, this paper uses a Pedestrians Counting algorithm based on face detection inreference to previous studies and in consideration of the advantages and disadvantages ofvarious algorithms. At the aspect of face detection, the paper adopts a improved Adaboost facedetection algorithm to get a good detection result for side faces,taking into account the existenceof the side faces of the case in the background of this application.Tracking statistics is an important part of the algorithm. Firstly, by analyzing thecharacteristics of the target motion and face characteristics, this paper chooses the method basedon combining the moving target tracking and the face detection and tracking technology. Duringtracking, the Kalman filtering prediction search area is used to reduce the matching time andachieve a real-time tracking. Secondly, the paper correctly handles the problems such as thehuman face is sheltered temporarily in the image sequences and the old face disappears while thenew face appearing. Then, the paper proposes a multi-count line traffic Statistics method on thisbasis of the actual application environment. Finally, the proposed method is tested throughexperiments. The experiments show that the proposed method is accurate, effective and reliable.