室內環境之弱互動物體追蹤

Weakly Interacting Object Tracking In Indoor Environments

萬高維(Kao-Wei Wan)

指導教授 ： 王傑智

國立臺灣大學/電機資訊學院/資訊工程學研究所/碩士(2008年)

https://doi.org/10.6342/NTU.2008.02196

摘要

在多目標追蹤(multitarget tracking)中，目標物之間的互動(interactions)已被研究用於解決遮蔽(occlusion)問題，但是多數現存的方法都把互動當作觀測資料上的變化，而沒有利用互動所包含更高階的資訊。相較於室外環境，室內環境中因為缺乏交通規則而較無限制，在室內環境的互動比在室外環境的互動更弱且多出更多可能性。弱互動(weak interactions)使得建構環境互動模型(scene interaction model)和鄰近物體互動模型(neighboring object interaction model)更加困難。這篇論文提出位置驅動(place-driven)的環境互動模型來表示目標物和環境之間的長期互動。為處理目標物之間複雜的短期互動，此篇論文提出鄰近物體互動模型，其中包含三個短期互動模型：跟隨(following)、逼近(approaching)、避障(avoidance)。此篇論文整合移動模型、靜態物體模型、和上述兩種互動模型，完成室內環境之弱互動物體追蹤(weakly interacting object tracking)。此外也達成更高階的場景了解(scene understanding)，如異常活動辨識(unusual activity recognition)和重要位置定義(important place identification)等。使用雷射測距儀所得到的實驗結果顯示提出方法的可用性和強健性。

關鍵字

多目標追蹤 ； 互動 ； 弱互動 ； 多模型追蹤

並列摘要

Interactions between targets have been exploited to solve the occlusion problem in multitarget tracking, but most of the existing approaches treat interactions in measurement level and do not take advantage of higher level information from the interactions. As indoor environments are relatively unconstrained than urban areas due to the lack of traffic laws, interactions in indoor environments are weaker and have more variants than in outdoor environments. Weak interactions make scene interaction modeling and neighboring object interaction modeling more challenging. In this thesis, a place-driven scene interaction model is proposed to represent long-term interactions between moving objects and the scene in indoor environments. To deal with complicated short-term interactions among moving objects, the neighboring object interaction model consisting of three short-term interaction models, following, approaching and avoidance is proposed. The moving model, the stationary process model and these two interaction models are integrated to accomplish weakly interacting object tracking in indoor environments. In addition, higher level scene understanding such as unusual activity recognition and important place identification is accomplished straightforwardly. The experimental results using data from a laser scanner demonstrate the feasibility and robustness of the proposed approaches.

並列關鍵字

multitarget tracking ； interaction ； weak interaction ； multiple model tracking

參考文獻

Ackerson, G. A. and Fu, K. S. (1970). On state estimation in switching environments. IEEE Trans. Automatic Control, AC-15(1), 10-17.

Blom, H. A. P. and Bar-Shalom, Y. (1988). The interacting multiple model algorithm for systems with markovian switching coefficients. IEEE Trans. On Automatic Control, 33(8), 780-783.

Chang, C. B. and Athans, M. (1978). State estimation for discrete systems with switching parameters. IEEE Trans. Aerospace and Electronic Systems, AES-14(5), 418-425.

Cox, I.J., H. S. (1996). An efficient implementation of reid’s multiple hypothesis tracking algorithm and its evaluation for the purpose of visual tracking. Pattern Analysis and Machine Intelligence, IEEE Transactions, 18, 138-150.

Cox, I. J. and Hingorani, S. L. (1996). An efficient implemenation of Reid’s multiple hypothesis tracking algorithm and its evaluation for the purpose of visual tracking. IEEE Trans. on Pattern Analysis and Machine intelligence, 18(2), 138-150.