Automatic Garage-Parking System Implemented by Optical Flow Applied to Parking Space Detection and Fuzzy Control Decision

碩士 === 國立臺北科技大學 === 電腦與通訊研究所 === 101 === In this paper, we proposed a new method to detect parking space. First, we used four wide angle cameras around the vehicle to capture four images, through the spatial domain transformation which includes camera calibration and bird’s-eye views to construct a...

Full description

Bibliographic Details
Main Authors: Chien-Liang Lin, 林建良
Other Authors: Rong-Chin Lo
Format: Others
Language:en_US
Published: 2013
Online Access:http://ndltd.ncl.edu.tw/handle/y9j75g
Description
Summary:碩士 === 國立臺北科技大學 === 電腦與通訊研究所 === 101 === In this paper, we proposed a new method to detect parking space. First, we used four wide angle cameras around the vehicle to capture four images, through the spatial domain transformation which includes camera calibration and bird’s-eye views to construct a completely panoramic image of the vehicle surroundings; meanwhile, using Hough transform to find the reference line of parking space and according to the number of optical flow points to judge whether it is a parking space or not. Once a parking space is available, we can further get the distance between parking space and vehicle and also estimate the angle of vehicle via a reference line on the surround image. After feeding the distance and angle information mentioned above to the fuzzy inference engine which is used to control the steering angle of the vehicle, and we can move the vehicle to follow a trajectory consisting of a quarter-circle arc and a straight line, and then finish the automatic garage-parking task.