請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/69719
標題: | 精準六自由度物體姿態之估測與追蹤 Accurate 6DoF Object Pose Estimation and Tracking |
作者: | Po-Chen Wu 吳柏辰 |
指導教授: | 簡韶逸(Shao-Yi Chien) |
關鍵字: | 擴增實境,虛擬實境,六自由度,物體姿態,姿態估測,姿態追蹤, Augmented Realty (AR),Virtual Reality (VR),Six Degrees of Freedom (6DoF),Object Pose,Pose Estimation,Pose Tracking, |
出版年 : | 2018 |
學位: | 博士 |
摘要: | 本篇論文主要探討的問題是如何從已校準過的相機影像中穩定且可靠地計算出目標物體相對於拍攝相機的六自由度姿態,而此六自由度姿態是由三自由度的旋轉變量與三自由度的移動變量所組成。
雖然目前文壇上已有許多研究人員提出不同種解決此類問題的演算法,但是由於各方在做演算法間的交互評比時所使用的測試影像序列資料往往不盡相同,甚至在大部分情況下接近真實情況的影像序列是缺乏的,導致目前無法得到不同類型的演算法在多種情況與條件下的客觀表現優劣分析。因此在本篇論文中,我們提出了一個同時具備彩色與深度影像序列的大規模之物體姿態追蹤基準資料庫,此資料庫不僅包含數種不同型態的平面與立體目標物外,也同時提供物體真實姿態值。此外,我們也針對目前現有的演算法在此基準資料庫上做了完整的表現評估,也分析了現有演算法效能提升手段的可能切入點。 即使多點投影演算法通常可以精準地計算出目標物體相對於拍攝相機的姿態,此種演算法本質上需要目標物本身有足夠的纹理表徵,並且其特徵點群能成功與相機影像中的特徵點群正確配對時才能順利計算出準確的物體姿態,這也是傳統特徵點演算法的主要缺陷之一。因此,我們設計了一個用來估測物體姿態的二步直接法,此演算法無論目標物的纹理表徵足與不足皆可穩定且精準地計算出其姿態。基於此演算法,我們發展了一套能即時以六自由度追蹤被動目標筆件的系統,此系統的準確度甚至可達亞毫米的水準,使其足以在混合實境的環境中書寫及作畫。我們透過一系列在合成資料庫與真實資料庫上運行分析的實驗結果展現此系統在各種狀況下所成就的高效準確的姿態追蹤結果,其精準程度甚至可比由多台相機所組成的工業級動作捕捉系統。 This dissertation is concerned with the problem of determining the six degrees of freedom (6DoF) object poses from a calibrated camera. Given camera images which contain the target object, we wish to estimate the position and orientation of the target with respect to the camera accurately and robustly. Although a variety of algorithms for this task have been proposed, it remains difficult to evaluate existing methods in the literature as oftentimes different sequences are used, and no benchmark datasets close to real-world scenarios are available. In this dissertation, we present a large-scale object pose tracking benchmark dataset consisting of RGB-D video sequences of 2D and 3D targets with ground-truth information. In particular, we perform the extensive quantitative evaluation of the state-of-the-art methods on this benchmark dataset and discuss the potential research directions in this field. While advanced Perspective-n-Point algorithms perform well in pose estimation, the success hinges on whether feature points can be extracted and matched correctly on target objects with rich texture. Consequently, we develop a two-step robust direct method for 6DoF pose estimation that performs accurately on both textured and textureless planar target objects. Based on the proposed two-step direct approach, we present a system for real-time 6DoF tracking of a passive stylus that achieves submillimeter accuracy, which is suitable for writing or drawing in mixed reality applications. We demonstrate the system performance regarding speed and accuracy on a number of synthetic and real datasets, showing that it can be competitive with state-of-the-art multi-camera motion capture systems. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/69719 |
DOI: | 10.6342/NTU201800854 |
全文授權: | 有償授權 |
顯示於系所單位: | 電子工程學研究所 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-107-1.pdf 目前未授權公開取用 | 74.06 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。