Algorithm And Application Of Passenger Flow Estimation Based On Machine Vision

In order to meet the requirement of estimation and analysis to the passenger flow of public place,a method based on multi-targets tracking to estimate passenger flow has been discussed in this paper,resolving the problem of compound movement in the watch area and having some theoretical and practical significance for the rapid and exact estimation of shopping malls,museums,libraries,tourist attractions,and some other public places.The main research contents and conclusions have been shown as follows:1) To research the theory of passenger flow estimation based on machine vision. The c飊clusion shows that it is feasible and effective that estimating the passenger flow by applying machine vision technology.2) To research the multi-target tracking method.Leading particle filter into passenger flow estimation system,integrating with the method based on the nearest-connected-region tracking method meet the requirement of real-time.3) To build the hardware sub-system of passenger flow estimation system based on machine vision.By using camera,lens,capture card,computer and LED display,the thesis constructs a hardware device based on PC platform with low cost.4) To program the software sub-system of passenger flow estimation system based on machine vision.By using target tracking framework provided by open-source architecture -OpenCV,encapsulating the algorithm and some necessary components of capture card's SDK in a dll file,programming a system UI in VC++,the study bring about the software sub-system.This system includes image capturing module,image segmentation module,fore-ground detecting module,target tracking module,behavior recognition module,and interface for repeated development.5) According to the field test result,the error ratio of in-bound passenger flow estimation is about +3.78%,the error ratio of out-bound passenger flow estimation is about +1.88%,and the total error ratio is about +2.87%.This result proves that the system can estimate the passenger flow of public places accurately...