應用於車輛側邊安全輔助之影像偵測系統

Abstract

本論文發展一套應用於車輛側邊安全之影像偵測系統。文中提出混合式光流法結合物體特徵點之影像運動偵測系統，此系統可在移動環境之連續影像中分割出疑似移動物體特徵，包括汽車和機車。本論文運用CMOS感測器作為低成本的側邊安全偵測之影像感測器，及採用Nios embedded processor將此偵測系統發展為一獨立式裝置。本文提出之影像動態偵測法則，能在移動背景中偵測移動物體特徵，在經由安全性分析使本系統能適時發出警示資訊(聲音、LED顯示)提醒駕駛者。我們將此獨立式系統置於車中進行實際測試。測試之結果顯示本系統具有在連續影像偵測出疑似移動物體特徵之位置和相對應的光流特性，並提供行車側邊之安全輔助。

In this thesis, an image-based motion detection system has been designed and realized for lateral driving assistance of a vehicle. We proposed a feature-based detection system, which combines with a mixed optical flow estimation algorithm for motion detection. The developed system can segment moving objects from a moving background. The objects contain cars and motorcycles. We have realized the system in a stand-alone image processing system, which can operate independently on the vehicle using a CMOS image sensor and Nios embedded processor. Experimental results show that the proposed methods are effective to segment features of moving objects from a moving background for lateral safety assistance.