| Summary: | 碩士 === 淡江大學 === 運輸管理學系 === 88 === High-Occupancy-Vehicle is a new regulative strategy. The focal point of regulation is to control the quantity of passengers in passenger car. We use image processing techniques to find the positions of vehicle and adopt Neural Network to identify the quantity of front-seat passenger.
The paper is dividing into three. Firstly, in order to get the best height and angle of detector, we discuss the relation between vehicle and background in image. And we get the best height is 3 meters and the best angle is 60。. Secondly, we develop a Vehicle-Windshield-Found algorithm （VWFA） to get exact position of vehicle's windshield. Thirdly, according to the exact position of windshield, we can use Neural Network as identify tool to count front-seat passengers in vehicle.
We use 400 images as training samples and 100 images as test samples to build the Neural Network Model and to estimate the correct rates of VWFA. The research get some physical results including the windshield recognize rate is 84﹪and the quantity of front-seat passenger recognize rate is 94﹪,and the whole recognize rate is 80﹪.
|
|---|
