利用陰影去除及重疊偵測之人類行為分析

Abstract

監控系統在很多領域上有很廣泛的應用，如在家庭保全、保安、病人監控等。在這些應用中，一個智慧型的監控系統需要具有即時觀察環境以及即時發送訊息的能力。在這篇論文中，我們提出了一個以追蹤物件以及分析人類不正常行為的監控系統，例如：暈倒。首先，我們使用以背景作為參考之動態物件切割演算法把移動中的物體從背景中分割出來。接著，我們使用陰影去除演算法把陰影消除，以得到較佳的切割物件。經過陰影去除後，追蹤演算法使用了兩個配對涵式來找出人的移動軌跡。最後，我們使用找出來的移動軌跡以及一個事先定義好的有限狀態機來分析人的不正常行為。我們測試陰影去除、追蹤演算法和人類不正常行為分析的效果。我們使用了幾個監控影片來測試系統，獲得了滿意的實驗結果。

Surveillance is widely used in many fields, such as home care, security, patient monitoring and so on. In such applications, an intelligent system is needed to monitor the situation and give corresponding responses in real time. In this thesis, we propose an object-based video tracking and human abnormal behavior (e.g.faint) surveillance system. First we use a background-registration segmentation algorithm to segment the moving objects. In order to obtain better segmentation, we use a shadow cancellation algorithm to eliminate shadow. After shadow cancellation, two matching algorithms are used in tracking algorithm to extract the human trajectories. Finally, we use the trajectory of human and a pre-decided finite state machine to analyze the abnormal of human behavior. We perform the experiments on effectiveness of shadow cancellation, tracking algorithm and human abnormal behavior analysis. We test our system with several surveillance video sequences and we obtain a satisfactory experimental result.