* Faculty       * Staff       * Students & Alumni       * Committees       * Contact       * Institute Directory
* Undergraduate Program       * Graduate Program       * Courses       * Institute Catalog      
* Undergraduate       * Graduate       * Institute Admissions: Undergraduate | Graduate      
* Colloquia       * Seminars       * News       * Events       * Institute Events      
* Overview       * Lab Manual       * Institute Computing      
No Menu Selected

* Research

Ph.D. Theses

Image Guided Borescope Tip Pose Determination

By Kenneth M. Martin
Advisor: Charles Stewart
February 23, 1998

This thesis describes a novel approach for solving the camera pose estimation problem commonly encountered in computer vision. In contrast to many previous approaches, this thesis considers the problems associated with pose estimation of typical industrial CAD models which are far larger than most pose estimation techniques can support. This thesis advances the state of the art in a number of areas. First, it provides a solution that is suitable for real time execution without specialized image processing hardware. It handles lens distortions common to borescopes, endoscopes, and other small cameras. Its on-line performance is interactive and unaffected by the size of the CAD model.

This approach consists of two components: off-line feature extraction from the CAD model, and on-line pose estimation. It is novel in its use of traditional computer graphics hardware to perform accelerated feature extraction off-line. The on-line approach is novel in that predicted features are matched against the incoming video image in an adaptive manner without requiring explicit feature extraction from the video image. The error vectors from this feature matching are used to solve for the delta pose vector using a new technique that includes radial lens distortions and support for different types of features. The framework for this thesis is borescope inspection and results are presented for both simulated and actual borescope video. The effects of feature density, motion prediction, and feature matching are considered as well as frame rate performance requirements.

* Return to main PhD Theses page



---