利用透視轉換技術估計人群數目

Abstract

近年來,在許多的自動化監控領域中,人群密度的測量是其中一項重要的主題。本篇論文提出的估測系統是利用單張影像,針對位於複雜背景的人群,更進一步地估測其人群的數目,以獲得較人群密度更具參考價值的數據。本系統包括了兩大主要步驟：頭部輪廓辨識與人群數目估測。在頭部輪廓辨識方面,先利用Haar小波轉換(HWT)產生特徵區域,再利用Support Vector Machines (SVM) 辨識該特徵區域是否為頭部輪廓。然後,利用透視轉換的技巧,進而更精準地估測人群的數目。最後,我們建立一個模型來評估本系統, 並更進一步地將本系統應用於實際環境的估測。

In the past years, the estimation of crowd density has become an important topic in the field of the automatic surveillance systems. In this thesis, the developed system takes one step ahead to estimate the crowd size at complex background by using single image. Therefore, more valuable information than crowd density can be obtained. There are two major steps in this system: recognition of the headlike contour and estimation of crowd size. In the part of recognition, the Haar Wavelet Transform (HWT) is used to extract the featured area of the headlike contour, and then the Support Vector Machines (SVM) is used to classify these featured area as the contour of head or not. Next, for estimation, the perspective transforming technique of computer vision is used to estimate crowd size more accurately. Finally, a model world is constructed to test this proposed system and the system is also applied in real world. Abstract 誌謝 Contents List of Figures List of Tables 1 Introduction 1.1 Survey 1.2 Motivation 1.3 Organization of the Thesis 2 Image Processing, Haar Wavelet Transform and Support Vector Machine 2.1 Haar Wavelet Transform 2.2 Histogram Equalization 2.3 Support Vector Machine 2.4 Linear Regression 3 Crowd Size Estimation System 3.1 System Overview 3.2 Feature Extraction 3.3 System Training 3.4 Crowd Size Estimation 3.4.1 Vanishing Point 3.4.2 Equidistant Parallel Lines in Computer Vision 3.4.3 Estimation of Crowd Size 4 Experimental Results 4.1 Experimental Environment 4.2 Experimental Results and Analyses 4.2.1 The Results of Estimation of Crowd Size in the Model World 4.2.2 The Results of Estimation of Crowd Size in the Real World 5 Conclusions Bibliography