基於動作識別為基礎之人類行為描述模型

本篇論文以人類動作識別為基礎，提出一個新的描述模型用來記錄人類的行為。因此本論文的重心是放在有關事件分析的研究上，而有關事件分析的研究目前仍屬於起步的階段。一個完整的智慧型監控系統，是由下列幾個區塊所組成；即物件偵測、物件追蹤、動作辨識、人類行為描述模型與事件偵測、事件記錄、事件的控制處理及事件的預測等。本論文提出的方法是利用動作辨識的結果，以及動作辨識所累積的時間資訊為特徵，將人類所發生的動作及發生動作的時間記錄下來，然後經由動作的組合判斷出事件，並將判斷出來的事件做有效的處理。為了驗證人類行為描述模型的可行性，我們以行人通過十字路口所產生的事件為例子。在實驗中以十字路口為場景，由五位不同的行人通過十字路口，共拍攝60個影片，產生了191個事件。在實驗影片中共正確偵測出了187個事件，實驗結果正確率為98%、錯誤率2%。在行為分析上能夠準確且穩定判斷出定義好的事件，並做妥善處理及控制。

關鍵字

行為辨識；人類行為分析；人類事件

並列摘要

Based on human action recognition, this paper proposes a new description model to record human behavior. Therefore, this paper mainly researches on relevant event analysis based on action interrelation; however, research relating to event analysis still stays in preliminary stage. A complete intelligent surveillance system consists of the following parts; object detecting, object tracking, action recognition, human behavior description model and event detecting, event recording, event control processing, and event prediction. The paper intends to make use of action recognition result and regard time information accumulated in action recognition as features, record human actions and time spent in these actions, then identify events through action combination and give effective processing toward these identified events. In order to prove feasibility of human behavior description model, we take events produced when pedestrians pass through cross-road as example. Under cross-road context in the experiment, total 60 films are shot when five pedestrians are passing through cross-road, producing 191 events. 187 events are correctly detected in the experiment with correct rate of 98% and error rate of 2%. On human behavior analysis, the well-defined events are able to be correctly and steadily identified and given with proper processing and control.

並列關鍵字

Behavior Identification ； Human behavior analysis ； Human event

國際替代計量

基於動作識別為基礎之人類行為描述模型

全文下載

主題瀏覽