以電腦視覺為基礎之行車安全輔助系統

Computer Vision Based Safe Driving Assistance System

孫瑞鴻(sun Sun Rui-Hong)

指導教授 ： 陳士農

亞洲大學/資訊學院/資訊工程學系碩士班/碩士(2008年)

摘要

現今的交通事故頻繁，而事故頻繁的原因可能是駕駛不專心沒有注意車輛周圍狀況，才會發生悲劇，尤其是在車輛周圍駕駛看不見的盲點，更是發生事故的主因。汽車雖有後視鏡可觀看左右與正後方的來車，但可看的範圍內仍有死角，當駕駛開車不專心，後方如有來車急速接近，從後視鏡中注意到時往往都已來不急反應，而容易發生事故。此時若能有一項電腦輔助設備，即時監控周圍車輛距離，並立即反應給駕駛，隨時提醒危險來車，便可有效防止事故發生。
本論文利用電腦視覺技術開發行車安全距離輔助系統，用以分析周圍車輛與自車之距離，判斷是否有車輛太過接近，或是急速接近之危險車輛，並立即警示駕駛，輔助駕駛注意自車周圍車輛距離，在此系統的輔助下，可提高駕駛之警覺性，確保生命財產安全。本系統所需之數位攝影機並無特殊限制，一般常見的數位視訊攝影機即可應用於本系統，透過擷取的畫面資料進行影像分析，以取得車道線位置，再取得車輛位置，接著利用比例尺計算距離，將其資料轉換成2D俯視圖，令駕駛更快獲得資訊，來達成行車安全距離之警示輔助。

關鍵字

電腦視覺 ； 行車輔助 ； 車輛偵測 ； 距離偵測

並列摘要

Nowadays, there have been very frequent traffic accidents, and the reason of frequent traffic accidents could be that drivers are not sufficiently concentrated and didn’t pay attention to neighboring conditions so that tragedy is brought about. Especially, when it is at the blind-spots of vehicle and one cannot see clearly, which is oftentimes the main reason why accident occurs. Though there is rear mirror for one to see the traffic on the left, right, and the right rear, there are still blind-spots within the domain. Thus, if the driver is not careful enough and there is car coming with fast-speed, one could not have enough time to respond when one has seen something from the rear mirror, and accidents have thus occurred. Therefore, if there could certain computer-assisted equipment to monitor the neighboring distance of cars, and gives real-time response to the driver to caution cars coming in dangerous distance it could then effectively prevent accident.
This paper has exploited to compute vision technology to develop the safe driving assistance system for car, which will analyze the distance of neighboring vehicles against one’s vehicle to see if they are too close, or they are of dangerously fast-approaching vehicles. Then, the driver will be cautioned at once so as to assist the driver to pay attention to the distance of his car to neighboring cars. Under the assistance of this system, it can help enhance the alertness of the drivers, and ensure the safety of their life and property. The proposed system can use a general digital camera to analyze the real-time picture to obtain car-lane position and the position of cars. Afterwards, scale measurement will be used to calculate the distance, and the information is transformed into a 2D vertical view so that the drivers can receive the information more quickly. Thus, it would help achieve the objective of warning alert for safety distance in driving.

並列關鍵字

vehicle detection ； distance detection ； computer vision ； driving assistance

參考文獻

[6] T. F. Reid, C. Kim, and M. R. Paulsen, “Computer vision sensing of stress cracks in corn kernels,” Transactions of the ASAE, vol. 34, no. 5, pp. 2236-2244, 1991.

[7] V. C. Patel, R. W. McClendon, and J. W. Goodrum, “Detection of cracks in eggs using color computer vision and artificial neural networks,” AI Applications, vol. 10, no. 3, pp. 19-28, 1996.

[8] M. Bertozzi and A. Broggi, “GOLD: a Parallel Real-Time Stereo Vision System for Generic Obstacle and Lane Detection,” IEEE Transactions on Image Processing, Vol. 7, No. 1, pp. 62-81, January 1998.

[9] M. Bertozzi and A. Broggi, “Real-Time Lane and Obstacle Detection on the GOLD System,” Proceedings of Conference on Intelligent Vehicles, Sep. 1996, pp. 213-218.

[10] M. Bertozzi and A. Broggi, “Vision-Based Vehicle Guidance,” Computer, July 1997, pp. 49-55.