遮蔽狀況下的行人追蹤採用卡曼濾波器和粒子濾波器

Occluded Pedestrian Tracking using Collaboration of Kalman Filter and Particle Filter

張雁翔(Yen-Hsiang Chang)

指導教授 ： 林道通

國立臺北大學/電機資訊學院/通訊工程研究所/碩士(2010年)

摘要

物件追蹤在智慧型影像監控中是一種關鍵技術。因為移動物體頻繁的遮蔽狀況以致於追蹤演算法是困難的議題。我們提出一個新的方法來說明追蹤的問題並計算在多人的遮蔽狀況中影像場景中有多少人。
我們的方法包含物件追蹤系統和頭部偵測系統。在我們的方法中，卡曼濾波器和粒子濾波器演算法提供準確的物件追蹤，使得我們能解決移動物件間的遮蔽問題。頭部偵測系統採用顏色和形狀特徵來做行人的計數。實驗結果顯示我們的方法能有效且有效率的解決遮蔽問題。

關鍵字

影像監控,遮蔽追蹤,卡曼濾波器,粒子濾波器,距離轉換,頭部偵測,行人計數

並列摘要

Object tracking is a one of the key feature in intelligent video surveillance. It is a challenging task in tracking algorithm due to the frequent occlusion encountered between moving objects. We propose a novel method to address the problem of tracking and evaluating the number of people in multiple people scenes with an occlusion condition.
The proposed method combines an object tracking system and a head detection. In our framework, Kalman Filter and Particle Filter provide robust object tracking for solving the occlusion between moving object. The head detection adopts the color model and shape-based object detection for counting the number of people. Extensive experimental results show that our method possesses effective and efficient performance.

並列關鍵字

Video surveillance, tracking in occlusion, Kalman Filter, Particle Filter, Distance transform, head detection, people counting

參考文獻

[1] CR Wren, A. Azarbayejani, T. Darrell, and AP Pentland. Pfinder: Real-time tracking of the human body. IEEE Transactions on Pattern Analysis and Machine Intelligence,

19(7):780–785, 1997.

[2] Z. Zivkovic. Improved adaptive Gaussian mixture model for background subtraction. In Pattern Recognition, 2004. ICPR 2004. Proceedings of the 17th International Conference

on, volume 2, 2004.

[3] L. Wang, T. Tan, W. Hu, and H. Ning. Automatic gait recognition based on statistical shape analysis. IEEE Transactions on Image Processing, 12(9):1120–1131, 2003.