以移動式近紅外線相機進行夜間行人偵測

Near-Infrared Based Nighttime Pedestrian Detection on a Moving Platform

李忞蔚(Min-Wei Li)

指導教授 ： 傅立成

共同指導教授 ： 蕭培鏞

國立臺灣大學/電機資訊學院/資訊工程學研究所/碩士(2009年)

https://doi.org/10.6342/NTU.2009.00313

摘要

無資料

關鍵字

行人 ； 人形 ； 偵測 ； 特徵評估 ； 夜間

並列摘要

Pedestrian detection in video streams from a moving camera at nighttime is important for many applications. In this thesis, we present a nighttime pedestrian detection system that is based on the AdaBoost classifier with the symmetry weighted shapelet feature. There are three stages in the proposed system: a) find human candidates by segmentation with an adaptive thresholding method b) reject unreasonable candidates by some geometric constraints, such as the aspect ratio and the size of a candidate region, and c) verify the candidates by an AdaBoost classifier with symmetry weighted shapelet. The symmetry weighted shapelet is an enhanced version of the shapelet feature. We improve the shapelet feature to make it more informative for encoding pedestrians in a near-infrared image by embedding human symmetry property. The symmetry property is imposed on shapelet features by computing the similarity between itself and its symmetric pair to weight shapelet features. We also design a new feature evaluation approach, namely, Discrimination Measure Model (DMM), and by employing this approach we can more efficiently evaluate the discriminating power of the symmetry weighted shapelet, and then treat it as an additional support evidence for the proposed system.
In the experiments, our proposed nighttime pedestrian detection system demonstrates reliable results for pedestrian classification and the symmetry weighted shapelet also present a larger discriminating power than Histogram of Oriented Gradients (HOG) and original shapelet feature.

並列關鍵字

Pedestrian ； Human ； Detection ； Nighttime ； Feature Evaluation ； Shapelet

參考文獻

[1]V. N. Vapnik, The nature of statistical learning theory: Springer-Verlag New York, Inc., 1995.

[2]Y. Freund and R. E. Schapire, "A Decision-Theoretic Generalization of On-Line Learning and an Application to Boosting," Journal of Computer and System Sciences, vol. 55, pp. 119-139, 1997

[5]P. Viola, and M. Jones, "Rapid object detection using a boosted cascade of simple features," Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, pp. I-511-I-518, 2001

[6]C.-C. Chang and C.-J. Lin, "LIBSVM : a library for support vector machines.," Software available at http://www.csie.ntu.edu.tw/~cjlin/libsvm, 2001.

[7]W. Pratt, Digital Image Processing, 3rd ed. New York: John Wiley & Sons, 2001.