影像追蹤及其在交通資訊估測的應用

Abstract

本篇論文以主動式輪廓模型為基礎，發展出影像追蹤技術並應用在交通資訊之估測。論文中發展出以主動輪廓模型求得車輛輪廓位置之影像量測方法，進而搭配卡曼濾波器進行追蹤車輛輪廓，由此可求得交通參數如車速、車輛數目等資訊。為了解決追蹤初始化的問題，本文運用輪廓成長的概念，設計一初始追蹤機制。最後以V8所拍攝路況影像來驗證影像追蹤的效果。

This thesis is mainly about image tracking techniques based on active contour model, and its application in traffic parameter estimation. The thesis introduces a algorithm of image measurement developed from active contour model to image to get the vehicle location. Further coordinating with Kalman Filter to tracking the vehicle location, traffic parameters like speed and numbers of vehicles would be thus acquired. The thesis utilizes the concept of contour growing to design an initializing tracking mechanism in order to solve the problems that could have occurred during the initialization of the tracking process. V8 visual images of traffic will be used to get the result of image tracking.