運用凝視技術於自動導航車之避障系統

Apply Gaze Technology to Obstacle Avoidance System of Autonomous Land Vehicle

湯鈞驛(Chun-Yi Tang)

指導教授 ： 駱榮欽

國立臺北科技大學/機電學院/自動化科技研究所/碩士(2011年)

https://doi.org/10.6841/NTUT.2011.00259

摘要

在本論文中，提出三個模組(有興趣物體的建立、有興趣物體的辨識、凝視處理)將之串聯使用，並搭配兩台平行且可以水平移動攝影機來實現整個系統，在有興趣物體建立中，主要利用顏色來分割目標物和背景並進行框選的動作。有興趣物體的辨識中利用快速PCA演算法將高維度的影像資訊進行降維，使其達到及時性的效果，其輸出結果用來當作SVM的輸入，最後利用SVM辨識出所選取的有興趣的物體是否有效。凝視處理將會對有興趣的物體做凝視，利用左右攝影機回傳的移動角度將其分為三種情形進行定位有興趣的物體以及算出有興趣的物體到自動導航車之間的距離，並將其應用於自動導航車的避障系統中。

關鍵字

凝視 ； 自動導航車 ； 有興趣的物體 ； 支持向量機分類演算法

並列摘要

In this paper, we propose a method contain three modules (Object-of-Interest (OOI) generation, OOI Classification, Gaze System) and the system architecture includes two horizontally parallel cameras, which can rotate horizontally to gaze on OOI, in front of ALV and one desktop PC. The OOI generation module includes two steps: how to separate image and sieve out the candidate from the image. The OOI classification includes two blocks: Fast Principle Component Analysis (FPCA) and Support Vector Machine (SVM). The images which are captured from right and left cameras will be high dimensional information, that is why we should use FPCA to do dimension reduction, and the output of FPCA will be the input of SVM, and then SVM classifies the OOI which is correct. The gaze processing is the key point in this search. When two cameras gaze on OOI, we can find depth and position of OOI by calculating the rotation angles of cameras. Finally, the method can be applied to obstacle avoidance system of Autonomous land vehicle (ALV).

並列關鍵字

Gaze ； Autonomous Land Vehicle ； OOI ； SVM

參考文獻

[2] J. Ge, Y. Luo and G. Tei, “Real-Time Pedestrian Detection and tracking at Nighttime for Driver-Assistance Systems”, IEEE Trans. Intelligent Transportation Systems, vol. 10, no. 2, June. 2009.

[3] F. Xu, X. Liu and K. Fujimura, “Pedestrian Detection and Tracking with Night Vision”, IEEE Trans. Intelligent Transportation Systems, vol. 6, no. 1, March 2005.

[4] A. Bal and M. S. Alam, “Automatic Target Tracking in FLIR Image Sequences Using Intensity Variation Function and Template Modeling”, IEEE Trans. Instrumentation and Measurement, vol. 54, no. 5, October 2005.

[5] Z. Li and X. Tang, “Using Support Vector Machines to Enhance The Performance of Bayesian Face Recognition”, IEEE Transactions on Information Forensics and Security, Vol. 2, No. 2, June 2007, pp. 174-180.

[6] Limin Xia, “Vehicle Recognition Using Boosting Neural Network Classifiers”, Proceedings of the 6th World Congress on Intelligent Control and Automation, June 2006, Dalian, China, pp. 9641-9644.