Post-graduate theses
Search command : Author="Στεφανίδης"
And Author="Κωνσταντίνος"
Current Record: 679 of 824
|
|
| Identifier |
uch.csd.msc//2000akoumianakis |
| Title |
Εξαγωγή κριτηρίων ασφαλούς πλοήγησης ρομπότ με χρήση πανοραμικής κάμερας |
| Alternative Title |
Criterion estimation for safely robots using a panoramic camera |
| Creator |
Akoumianakis, George A
|
| Abstract |
Autonomous robot navigation based on visual information provided by a camera, comprises a very important problem in the field of machine vision and robotics. Proposed methods usually depend on the computation of the optical flow to extract criteria for robot navigation. In addition, the majority of the approaches assume the perspective projection camera model. This is due to the fact that this type of cameras are commonly available. In recent years, the so called central panoramic cameras with upto a 360 degree field of view are also being used. The wide field of view of these cameras facilitates the resolution of important problems regarding autonomous visual robot navigation. In this thesis, two methods for calculating the egomotion of a panoramic camera are proposed, assuming that the distance of the camera from a point in the environment is known. Additionally, when the translational speed of the camera is known, the same two methods can be used to calculate the distance between a point and the camera. The two methods constitute different approaches to solving the problem. The first method is based on the calculation of a dense optical flow field due to the motion of the panoramic camera, while the second method is based on the correspondence of points between two panoramic images. If the egomotion and the distance of the panoramic camera from a point in the environment are not known, a method for calculating their ratio for every scene point is proposed. This ratio can be used as an estimation of time to contact between the camera and every scene point, assuming egomotion parameters are constant. The central panoramic camera was simulated using a raytracer program called \emph{POV-Ray}. The use of synthetic image sequences allows us to verify the accuracy of the proposed methods. The KLT software package was used for the extraction and tracking of suitable characteristic points between panoramic images. The experimental results prove that the second method provides satisfactory accuracy, regardless of the image resolution or the speed of the camera. Additionally, it is shown that the first method is more useful only when time to contact information is required for all image points as opposed to a set of characteristic points. Another conclusion is that the tracking algorithm should support affine transformations due to the optical deformations caused by the panoramic camera. Finally, future directions for research are suggested in order to improve the usability and effectiveness of the proposed methods.
|
| Issue date |
2000-11-01 |
| Date available |
2000-10-13 |
| Collection
|
School/Department--School of Sciences and Engineering--Department of Computer Science--Post-graduate theses
|
|
|
Type of Work--Post-graduate theses
|
| Permanent Link |
https://elocus.lib.uoc.gr//dlib/6/5/7/metadata-dlib-2000akoumianakis.tkl
|
| Views |
397 |
Collections
