| With the rapid development of economy, personnel exchanges have become increasingly frequent, the population migration velocity is speeding up too. More and more public places are in a high density flow.The safety of these places is becoming a very urgent problem.With the high speed development of video surveillance technology, especially the rise of crowd density detection technology, which provides the guarantee for the safety of these places.With the help of the crowd density detection system in real time, manager or organizer can take corresponding measures timely and conveniently to avoid the dangers, guarantee the safety of life and property.Firstly, we introduced the current situation of research of people flow density detection at home and abroad. Secondly, this thesis studied some theory and knowledge of flow detection in the field of image process. most of them is the knowledge of digital image processing, including image enhancement, image denoising, morphological processing and edge detection etc. Then we discussed the feature extraction and tracking based on KLT algorithm.It laid foundation of collectiveness that reflected crowd density.There are two traditional methods in crowd density detection based on the statistics of pixel and based on texture analysis.We compared them with experiment.In the description of the SDP model we introduced the concept of Collectiveness, which indicates the degree of individuals acting as a union in collective motion, is a fundamental and universal measurement for various crowd systems.While people density is high, Collectiveness is high,and vice versa. Finally we combined flow density and Collectiveness and applied it to the pratical application, show its practicality and accuracy.We also give the range of collectiveness in five different crowd density.This combination was just the creation of this thesis and it provided a reference for designing a complte system in the future. |
PDF Full Text Request