- 學位論文
駕駛輔助之超車車輛偵測
Overtaking vehicle detection and localization for driver assistance
若您是本文的作者，可授權文章由華藝線上圖書館中協助推廣。
摘要
傳統的車輛偵測系統，僅使用車尾偵測器來偵測行車前方的車輛，但針對駕駛輔助的需求，該系統無法對鄰近的超車車輛進行偵測，超車車輛有變換車道等發生碰撞的可能，所以我們提出超車車輛的偵測系統。 將車載攝影機行駛於道路所拍攝的影像作為輸入，使用車尾偵測器和車輪偵測器，來偵測實際駕駛環境中攝影機車輛的前方車輛和兩側道路的超車車輛，超車車輛的外觀變化非常急遽，不同的位置和角度，讓車輛的側面外觀變化複雜，所以我們選擇了外觀變化小的車輪部位來建立鑑別性良好的偵測器，達到超車車輛的完整偵測結果。 在影像上計算MB-LBP特徵，再使用AdaBoost所訓練的車尾和車輪這兩種偵測器進行掃描，有別於一般偵測物體的方式，除了變化搜尋的尺度，也建立了三種長寬比例的車輪偵測器，以克服在實際的行駛狀況下，畫面中不同位置和角度的車輛側面有不同的輪胎長寬比例，使用串聯分類器有助於提高偵測速度和選出具有鑑別性的特徵來提高偵測率，車輪偵測器的偵測率達0.96，車尾偵測器的偵測率達0.92。 一般車輪偵測只使用在靜態車輛的側面部位定位之用，從實驗結果能夠看出，將其應用到行車環境後依舊有良好可靠的偵測效果，最後利用偵測到的車尾和車輪位置來在畫面上標示出車輛的所在位置。
並列摘要
In this paper, we proposed an effective overtaking vehicle detection system. Using image from car-mounted camera as input, detect vehicle’s rear-view and side-view on actual driving scene to achieve overtaking vehicle detection. First, compute MB-LBP feature of image, then use cascade classifier as a detector trained by AdaBoost algorithm. Differing from the traditional detection approach, a detector using 3 different aspect ratios and finding the varying scales is established so as to overcome the fact that in real driving situations, the aspect ratio of the tire appearing in the image keeps varying. Using cascade classifier helps enhance the detection speed and keep the discriminative features so as to enhance the detection rate. Finally, both the information of the rear-view and the wheel of a vehicle are used to mark the position of it in the image. The performance of wheel detector is demonstrated with the precision rate as 0.91 and recall rate as 0.96. After wheel detection, the wheel matching is proposed which can be further applied for the overtaking vehicle localization. We achieve good detection results with two detectors.
參考文獻
[1] Zehang Sun, George Bebis, Ronald Miller, “On-Road Vehicle Detection A Review”, IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, vol. 28, no. 5, pp.694-711, 2006.
[2] Chih-Ming Fu, Chung-Lin Huang, and Yi-Sheng Chen, “Vision-Based Preceding Vehicle Detection and Tracking”, The 18th International Conference on Pattern Recognition, vol. 2, pp.1070-1073, 2006.
[3] Sayanan Sivaraman, Mohan Manubhai Trivedi, “A General Active-Learning Framework for On-Road Vehicle Recognition and Tracking”, IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS, vol. 11, no. 2, 2010.
[4] Yan Zhang, Stephen J. Kiselewich, William A. Bauson, and Riad Hammoud, “Robust Moving Object Detection at Distance in the Visible Spectrum and Beyond Using A Moving Camera”, Conference on Computer Vision and Pattern Recognition Workshop , pp.131-138, 2006.
[5] Fernando Garcia, Pietro Cerri, Alberto Broggi, “Data fusion for overtaking vehicle detection based on radar and optical flow”, 2012 IEEE Intelligent Vehicles Symposium , pp.494-499, 2012.