Video surveillance for intelligent transportation system

• Main Author: Liu, Jiamin
• Other Authors: Chau Lap Pui
• Format: Final Year Project
• Language: English
• Published: 2019
• Subjects: Engineering::Electrical and electronic engineering
• Online Access: http://hdl.handle.net/10356/78696
• Institution: Nanyang Technological University

Video Surveillance has been the most important input for the Intelligent Transportation System (ITS). A significant amount of research on the Video Surveillance Intelligent Transportation System is focused on the automatic traffic flow rate counting and vehicle type detection. Artificial Intelligence (AI) with matching learning method had been widely used in this field in recent years. However, machine learning consumes a significant amount of computing power and unpredictable at some incidents. We are presenting a way to robustly detect the real-time traffic flowrate and classify the vehicle type (bus, car, motorbike) using basic video processing technologies from a surveillance camera video input. We are using video processing open and close operation to detect vehicles from the background as an object. Then we implement Kalman filter predict the movement of this vehicle object. Base on the vehicle object velocity and location from Kalman filter, we using Hungarian algorism to assign the vehicle object to vehicle tracks that detected before. Then we classify all tracks base on the vehicle object existing time and size inside a small region of interest (ROI) to recognize the vehicle object as an actual vehicle and they type of vehicle (bus, car, motorbike). Our major contribution is to find a way that is not consuming too many computational powers as machine learning to detect and classify the vehicle. The key technology is that we find out the width of the vehicle object size in a small region of interest (ROI) on the road can easily classify the type of vehicle (bus, car, motorbike). We also construct a robust vehicle counting and real-time traffic flow rate computing system based on the previous work. This software can easily be implemented in any traffic surveillance camera as an input of the Intelligent Transportation System.