Your browser does not support JavaScript!

Doctoral theses

Current Record: 99 of 121

Back to Results Previous page
Next page
Add to Basket
[Add to Basket]
Identifier 299112
Title Motion Analysis and Modeling for Activity Recognition and 3-D Animation based on Geometrical and Video Processing Algorithms
Alternative Title Ανάλυση Κίνησης και Μοντελοποίηση για Αναγνώριση Δραστηριότητας και 3-Δ Συνθετική Κίνηση βασισμένη σε Γεωμετρικούς Αλγορίθμους και Μεθόδους Επεξεργασίας Βίντεο
Author Παναγιωτάκης, Κώστας
Thesis advisor Τζιρίτας, Γεώργιος
Reviewer Τραχανιάς, Παναγιώτης
Στυλιανού, Ιωάννης
Abstract The analysis of audiovisual data aims at a high level information extraction, equivalent with the one that can be extracted by a human. It is considered as a fundamental, unsolved (in its general form) problem. Even if the inverse problem, the audiovisual (sound and animation) synthesis, is judged easier than the previous, remains an unsolved problem. The systematic research on these problems yields solutions that constitute the basis for a huge number of continuously developing applications. In this thesis, we examine the two aforementioned fundamental problems. We propose algorithms and models of analysis and synthesis of articulated motion and undulatory (snake) locomotion, using data from video sequences. The goal of this research is the multilevel information extraction from video, like object tracking and activity recognition, and the 3-D animation synthesis in virtual environments based on the results of analysis. An important part of this thesis is dedicated to automatic human motion analysis from video and action/activity recognition. Moreover, we examine the problem of animal animation synthesis using parametric models and algorithms that are based on motion tracking data over appropriate video sequences. The problem of undulatory locomotion analysis led in the definition of a general geometrical problem, the curve EquiPartition (EP). In this thesis, we define, analyze and solve the EP problem. In the case of human motion analysis, we propose a general framework that focuses on automatic individual/multiple people motion-shape analysis and on suitable features extraction, which can be used on action/activity recognition problems under real, dynamical and unconstrained environments. In order to evaluate the robustness of the proposed scheme, we have applied it on various athletic videos from a single uncalibrated, possibly moving camera. Concerning the animals? animation synthesis, we have examined the problem of animal modeling, 3-D model construction and 3-D animation synthesis in complex 3-D virtual environments. The motion analysis and the 3-D animal model construction are performed using videos captured by a static camera from specific viewpoints. We have proposed distinct methods for articulated and undulatory motion analysis and synthesis. The whole methodology can be easily expanded in similar creatures. The general geometrical problem of partitioning a continuous curve into N parts with equal chords, under any metric distance, the curve EquiPartition problem (EP), is analyzed and solved. Moreover, we prove that the problem admits at least a solution under the Euclidean distance metric for planar curves, based on an introduced equivalent problem definition, the Level Set Approach. Finally, EP based applications, like polygonal approximation and key frames selection, are presented, and the special properties of their solutions are discussed.
Language English
Issue date 2007-05-03
Collection   School/Department--School of Sciences and Engineering--Department of Computer Science--Doctoral theses
  Type of Work--Doctoral theses
Permanent Link https://elocus.lib.uoc.gr//dlib/a/7/4/metadata-dlib-2007panagiotakis.tkl Bookmark and Share
Views 592

Digital Documents
No preview available

Download document
View document
Views : 16