車輛側邊之盲點與車道偏離警示系統

Abstract

駕駛者於道路中駕駛車輛時，雖然可以經由後照鏡來彌補視野範圍的不足，協助其進行行車狀況的判斷，但是各種後照鏡都存在著大小不一的盲點區域，加上人類本身判斷錯誤的機率，使得所有道路使用者都必須承受意外發生的風險。 本篇論文將利用內嵌式數位攝影機搭配數位訊號處理器來實現行車時的即時車輛側邊路況偵測，偵測範圍包含了車輛周圍的視覺死角以及側後照鏡的盲點區域，對車輛左右兩側的各種車輛進行多車輛物件偵測，並輔以車道偏離警示系統，輔助駕駛者判斷自身的駕駛狀況與自身車輛周圍的道路情況。 本系統已於實際車輛上即時測試驗證成功，將可適應日間晴天、陰天、雨天等各種天氣狀況，適用道路種類包含高、快速道路以及一般市區街道。

When driving on the road, automobile drivers use side-view mirrors to increase the visible field and help give prompt traffic information. However, all side-view mirrors have its blind-spot region. In addition, if including the probability of incorrect judgements made by human, everyone of us on the road has to suffer the risk of getting into an accident. The purpose of this thesis is to use embedded digital cameras along with digital signal processors to realize the real-time traffic detection on a car. The detection area includes not only the invisible region around the car but also the blind spots of side-view mirrors. The systems can detect multiple vehicles at both sides at the same time of the car, and giving lane departure warnings to assist drivers to consider their own circumstances on the road. The real-time tests of the systems were successful. The adaptable weather conditions include sunny, cloudy and rainy. This system can also be used in many kinds of roadway, including highways, expressways, and common downtown streets.