影像追蹤系統及其在車輛安全之應用

Abstract

本論文發展一套應用於車輛安全之影像追蹤系統。文中提出一基於特徵追蹤法並結合擴散焦點(FOE)及場景分析之影像追蹤系統，此系統可自動偵測並同時追蹤多個動態物體，包括汽車及機車，並可在複雜的環境中有效地分離背景與動態物體，以提供碰撞偵測之資訊。本論文採用CMOS影像感測器及Nios embedded processor將此系統發展為一獨立式系統，使之能在車輛上獨立運作。我們將此獨立式影像追蹤系統置於車中進行實際測試。測試之結果顯示本系統具有追蹤車輛並提供碰撞偵測之資訊。

In the thesis, a visual tracking system for vehicle safety has been developed. We proposed a feature-based tracking system, which is combined with the concept of FOE and scenario analysis. This system can automatically detect and track multiple moving objects, including cars and motorcycles. The developed system can segment features of moving objects from moving background simultaneously and provide useful information of collision warning. Furthermore, we have realized the system in a stand-alone system, which can operate independently in the vehicle using a CMOS image sensor and Nios embedded processor architecture. The stand-alone visual tracking system has been validated in actual road environment. Experimental results show that the proposed methods successfully tracks front vehicles and provides information of collision warning.