混合行人與汽機車於路口之影像式偵測系統研究

本研究目的是利用影像處理技術來實現路口監控系統。路口系統使用影像資訊，偵測路口行人的流量，可以配合適當的紅綠燈轉換時間，有助於整體交通的流暢性，並且警示欲通過路口之車輛，注意行人安全。本系統包含了移動物件偵測模組、行人辨識模組以及移動物件追蹤模組。移動物件偵測模組混合用背景更新法得到的前景資訊與連續影像相減的移動輪廓資訊兩種方法來偵測行人與汽機車。當物體移動時用移動物件輪廓，物體停止時則是用前景影像。行人辨識模組則是偵測並辨識出行人與汽機車，行人偵測的困難點在於重疊行人，所以本論文使用三種不同的影像特徵來做行人偵測，第一種是行人頭部輪廓，配合本論文提出的角偵測找出可能的行人頭部；第二種是利用行人群的位置關係，搭配行人高度和寬度，將行人群做分割；第三種是行人頭髮灰階值的特徵，比對擷取到的頭髮位置和行人高度判斷是否為行人。汽機車偵測的特徵是用汽機車底部陰影，陰影區域符合汽機車的範圍當作偵測結果。移動物件追蹤模組將行人和汽機車偵測結果做追蹤處理，配合追蹤預測的位置，比對目前與追蹤的影像資訊，只有比對成功的移動物件才認定是追蹤成功。由實驗結果顯示，本系統能有效與強健的進行行人偵測，本論文所提出的方法，平均行人偵測率與正確率皆有九成以上，而且本系統在戶外長時間運作不受光線影響。

關鍵字

行人偵測；路口；汽機車偵測；影像

並列摘要

The goal of this study is using image processing technologies to implement the intersection monitoring system. The system calculates the pedestrian flow in intersection by means of the image information. Incorporating with the pedestrian flow and the transition time of red light, not only the traffic can be smoother but also the pedestrian safety can be improved by warning the passing vehicles. The proposed system composed moving object detection unit (MODU), pedestrian recognition unit (PRU), and object tracking unit (OUT). MODU, a proposed hybrid method, integrates the foreground information from background subtraction and the moving contour from consecutive frame difference to detect pedestrians, motorcyclists, and vehicles. The detection uses moving contour information while the object is moving; otherwise the foreground information is utilized. Pedestrian recognition unit distinguish the pedestrian from the motorcyclist and the vehicle. The difficulty in pedestrian detection is the overlap of pedestrians. Therefore, three different kinds of characteristic are used to detect and recognize the pedestrian. The first characteristic is head shape of the pedestrian. Through the corner detection, the positions of head candidates can be found. The pedestrians can be separated from the crowd by analyzing space information and considering the width and height of the pedestrian. Third characteristic is gray level of the hair. The pedestrian is found by comparing the height of object with the detected position of the hair. Since the shadow always exists in the bottom of the vehicle and motorcycle, the shadow characteristic is utilized. If the shadow region matches with size of the vehicle and motorcycle, the position of vehicle and motorcycle is detected. OTU tracks the detected pedestrian, the motorcyclist, and the vehicle. Comparing the current image information with the previous information, the object is tracked when the matching succeeds. The experimental results demonstrate that the proposed system can detect the pedestrian effectively and robustly. The average detection ratio and correct ratio are both above 90%. Moreover, our system performs very well in the daytime, no matter what kinds of lighting effect.

並列關鍵字

pedestrian detection ； vehicle detection ； intersection ； pedestrian

參考文獻

[1] N. Ikeda, A. Saitoh, T. Isokawa, N. Kamiura and N. Metsui, ”A Neural Network Approach for Counting Pedestrians from Video Sequence Images,” SICE Annual Conference, pp. 2485-2488, Aug. 2008.

[2] O. Masoud and N. P. Papanikolopoulos, ”A Novel Method for Tracking and Counting Pedestrians in Real-Time Using a Single Camera,” IEEE Trans. on Vehicular Technology, vol. 50, no. 5, pp. 1267-1278, Sep. 2001.

[3] I. Haritaoglu, D. Harwood and L. S. Davis, ” W4: Real-Time Surveillance of People and Their Activities,” IEEE Trans. on Pattern Analysis and Machine Intelligence, vol. 22, no. 8, pp. 809-830, Aug. 2000.

[4] C. R. Wren, A. Azarbayejani, T. Darrell and A. P. Pentland, “Pfinder: Real-Time Tracking of the Human Body,” IEEE Trans. on Pattern Analysis and Machine Intelligence, vol. 19, no. 7, pp. 780-785, July 1997.

[5] R. Cutler and L. S. Davis, ”Robust Real-Time Periodic Motion Detection, Analysis, and Applications,” IEEE Trans. on Pattern Analysis and Machine Intelligence, vol. 22, no. 8, pp. 781-796, Aug. 2000.

國際替代計量

混合行人與汽機車於路口之影像式偵測系統研究

全文下載

主題瀏覽