Development Of Passenger Flow Estimation Software Based On Video Analysis Technology

The main mode of traditional video surveillance is the automatic video recording and playback of human-computer interactive, virtually no video-based data mining functions. Current video surveillance has entered the era of intelligent video surveillance. With the using of computer vision theory on the video image processing and pattern recognition, we achieve intelligent video analysis the aim of intelligent video analysis. In the intelligent video analysis techniques, passenger flow estimation is an important means to solve the security problem of crowded places, and most concerned factor for our shopping malls and related sales companies. Based on passenger data, through the video data depth mining and analysis, passenger flow estimation system provides a series of important market management tools for businesses. Therefore, the research and development of passenger flow estimation analysis algorithms and related software has important significance and application value.The main contents are as follows:1, Through research of AdaBoost algorithm based on Haar features, we achieve the detection of passenger object on video.2, Through Pixel Matching method, we achieve the Tracking of passenger object on video.3, Based on the existing video surveillance equipment, we detect the target object which is pedestrian head. The program can either monitor, can also be achieved statistical functions.4, Pedestrian head is used as a detection target object, meanwhile Moving Foreground Extraction is combined to improve the accuracy of the algorithm.In this paper, we start the work of passenger flow estimation software development through C++and C language as a development language and VS2010combined with the OpenCV library as a development tool, and ultimately realize a successful demonstration on the PC. This task has laid a foundation for the latter part the secondary development and related applications software development that based on video data, and has done previous algorithm validation for the transplant in DSP platform.