Author: Ng, Wai Lam William
Title: Three-dimensional model acquisition and motion synthesis
Degree: Ph.D.
Year: 2013
Subject: Augmented reality.
Human-computer interaction.
Hong Kong Polytechnic University -- Dissertations
Department: Department of Electronic and Information Engineering
Pages: 141 leaves : ill. (some col.) ; 30 cm.
Language: English
Abstract: Augmented reality is an important extension of virtual reality that allows the interaction of real objects and computer generated graphics in a real or virtual environment. It has been a subject of extensive interest in recent years due to its widespread application in computer games, movies, simulations, etc. In this thesis, we focus on two important research areas in augmented reality: 3D model acquisition and character animation. To allow a real object to interact with computer generated objects in a virtual environment, it is desirable to obtain the three-dimensional (3D) model of the real object. One of the popular approaches to acquire high-fidelity 3D model of real objects is fringe projection profilometry. It projects parallel fringe patterns onto the object, and captures the deformed fringe by a camera for 3D reconstruction. Due to the imperfection of imaging devices and operating environment, the captured fringe images are inevitably to have many artifacts; one of them is noisy. It is especially the case when the acquisition process is carried out in real-time applications such that images are captured with short exposure time and perhaps high ISO value (for instance, in dark environment). Such artifacts can lead to serious distortion to the final reconstructed 3D model. In the first part of this thesis, we propose an efficient algorithm to enhance the noisy fringe images using 2-dimenional (2D) dual tree complex wavelet transform (DT-CWT). The proposed algorithm exploits the special analytic property of DT-CWT to obtain piecewise smooth representations of the fringe image. Then a new wavelet denoising procedure is applied for enhancing the noisy fringe image. The proposed algorithm outperforms the traditional approaches in enhancing the fringe image, and in turn significantly improves the accuracy of the reconstructed 3D model.
Another commonly found artifact in the captured fringe images is bias, which is a natural phenomenon due to the changing light intensity reflected from the object surface. If there is abruptly changing bias in the captured fringe image, severe reconstruction error will be resulted. To solve the problem, one popular approach is Phase Shift Profilometry (PSP) which removes the bias using two fringe images. It however is not suitable for real-time applications especially when the target object is moving. Any slight difference in the fringe image due to object{174}s motion can lead to serious distortion to the final reconstructed 3D model. Other approaches try to remove the bias by suppressing the zero spectrum of fringe image. It is based on the assumption that the energy of bias is located around dc (i.e. zero spectrum) and the aliasing between it and that of the fringe is negligible. It however is not the case in practice. In this thesis, we proposed a novel technique to eliminate the bias using a single fringe image. The new approach successfully identifies the feature of bias, fringe and noise in the DT-CWT domain, which allows the bias to be effectively extracted from a noisy fringe image. This bias removal approach can be integrated into the fringe enhancement algorithm as mentioned above so that the overall computational complexity can be reduced. The second part of this thesis is related to character animation, in particular synthesizing motion of computer generated characters. In augmented reality, one direction of study is how to integrate computer generated characters into real environment. To simplify the process of character animation, motion capture systems are often used so as to directly export the motion of real objects to the characters. While the motion data stored in the database are always limited due to various reasons, it is desirable if they can be reused to synthesize new motions. Traditionally, the standard motion graphs are used to facilitate motion data reuse. A motion graph keeps track of all possible transitions from one motion to another. It records how motion data can be spliced to form the motion required. Traditional motion graphs allow a motion clip to transit to another one only if the characters in the two motion clips have the same or very similar body posture. Such requirement however limits the possible transitions between motion clips and hence reduces their reusability. In this thesis, we present a new data structure called the Synchronized Partial body Motion Graph (SPbMG). The new SPbMG specifies possible transitions between upper body motions and between lower body motions separately; thus the connectivity of the graph is greatly enhanced. Besides, a new algorithm is proposed for exploiting the similarity between the lower body motions in order to derive the conditions for splicing the upper body and lower body motions synchronously. It ensures the coherence between the upper and lower body motions in the synthesized motion clips. Experimental results show that the proposed SPbMG improves the reusability of motion data while preserving the quality of motions as compared with the traditional motion graph techniques.
Rights: All rights reserved
Access: restricted access

Files in This Item:
File Description SizeFormat 
b26392148.pdfFor All Users (off-campus access for PolyU Staff & Students only)4.15 MBAdobe PDFView/Open

Copyright Undertaking

As a bona fide Library user, I declare that:

  1. I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
  2. I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
  3. I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.

By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.

Show full item record

Please use this identifier to cite or link to this item: