基於Adaboost演算法與對稱檢測的車輛偵測方法

Vehicle Detection based on Adaboost Algorithm and Symmetry Justification

陳彥東(Yen-Tung Chen)

指導教授 ： 林道通

國立臺北大學/電機資訊學院/資訊工程學系/碩士(2013年)

摘要

近年來，有關駕駛輔助系統的相關研究都是一個熱門的話題。原因除了擁有廣大的市場需求外，其所含括的應用範圍也相當的廣泛，包括車道偏離警示，碰撞預警，行人和車輛偵測系統等等。本文提出了一套即時車輛偵測和追蹤系統，能夠偵測駕駛人視野正前方的車輛。該系統主要有兩個核心技術:使用Adaboost演算法訓練分類器用以偵測道路上的所有可能車輛物件；再使用對稱軸檢測與水平垂直邊緣分析方法來判斷該物件是否為車輛。經過各種背景環境條件不同下影片的測試結果，證明本系統是具有相當大的可行性的。

關鍵字

車輛偵測 ； 車輛判斷 ； 對稱檢測

並列摘要

Recently, extensive research concerning driver assistance systems has been conducted. Related applications include lane departure warning, traffic sign recognition, and pedestrian and vehicle detection systems. This paper presents a real-time vehicle detection and tracking system capable of detecting the vehicles in front of a vehicle. The system consists of 2 main steps: generating candidates with respect to a vehicle by using the AdaBoost learning algorithm and verifying the candidates according to symmetry measurement and horizontal and vertical edge analysis. The proposed system is proven to be effective in various traffic scenarios through experiments.

並列關鍵字

adaboost ； symmetry measurement ； vehicle detection ； vehicle verification

參考文獻

environment monitoring for on-board driver support systems: Lessons learned from

design and implementation. Applied ergonomics, 41(2):225{235, 2010.

[2] Massimo Bertozzi, Alberto Broggi, Massimo Cellario, Alessandra Fascioli, Paolo Lombardi,

and Marco Porta. Arti cial vision in road vehicles. Proceedings of the IEEE,

90(7):1258{1271, 2002.