Automatic Pedestrian Image Segmentation by Using Human Shape Prior

利用人形機率分佈之自動化行人影像分割

李克駿(Ke-Chun Li)

指導教授 ： 賴尚宏

國立清華大學/電機資訊學院/資訊系統與應用研究所/碩士(2010年)

https://doi.org/10.6843/NTHU.2010.00480

摘要

In this thesis, we present an automatic and accurate pedestrian segmentation algorithm by incorporating pedestrian shape prior into random walks segmentation from a static image. The Random Walks algorithm requires user-specified labels to produce segmentation with each pixel assigned to a label. This algorithm can provide satisfactory segmentation result with suitable input labeled seeds. Therefore, for taking advantage of this interactive segmentation algorithm, we improve the random walks segmentation algorithm by using prior shape information, which provides appropriate seeds for the pedestrian segmentation from the input image. By using the human shape prior information, we develop a fully automatic pedestrian image segmentation algorithm. The experimental results demonstrate improved segmentation results on some real images by using the proposed algorithm.

關鍵字

影像分割 ； 隨機慢步演算法

參考文獻

[2] R. Cucchiara, C. Grana, A. Prati, and R. Vezzani, “Probabilities posture classification for human-behavior analysis,” IEEE Trans. Syst., Man, Cybern. A, Syst., Humans, vol. 35, no. 1, pp. 42–54, Jan. 2005.

[3] J. Shi and J. Malik, “Normalized cuts and image segmentation,” IEEE Trans. on PAMI, 22(8):888–905, 2000.

[4] D. Comaniciu and P. Meer, “Mean shift: A robust approach toward feature space analysis,” IEEE Trans. on PAMI, 24:603–619, 2002.

[5] D. Martin, C. Fowlkes, and J. Malik, “Learning to detect natural image boundaries using local brightness, color, and texture cues,” IEEE Trans. on PAMI, 26(5):530–549, 2004.

[6] Y. Boykov and M.-P. Jolly, “Interactive graph cuts for optimal boundary & region segmentation of objects in n-d images,” In ICCV, 2001.