Date of Award

3-2000

Document Type

Thesis

Degree Name

Master of Science

Department

Department of Electrical and Computer Engineering

First Advisor

Mikel Miller, PhD

Abstract

Recent developments in aviation have made micro air vehicles (MAVs) a reality. These featherweight palm-sized radio-controlled flying saucers embody the future of air-to-ground combat. No one has ever successfully implemented an autonomous control system for MAVs. Because MAVs are physically small with limited energy supplies, video signals offer superiority over radar for navigational applications. This research takes a step forward in real time machine vision processing. It investigates techniques for implementing a real time stereovision processing system using two miniature color cameras. The effects of poor-quality optics are overcome by a robust algorithm, which operates in real time and achieves frame rates up to 10 fps in ideal conditions. The vision system implements innovative work in the following five areas of vision processing: fast image registration preprocessing, object detection, feature correspondence, distortion-compensated ranging, and multi scale nominal frequency-based object recognition. Results indicate that the system can provide adequate obstacle avoidance feedback for autonomous vehicle control. However, typical relative position errors are about 10%-to high for surveillance applications. The range of operation is also limited to between 6 - 30 m. The root of this limitation is imprecise feature correspondence: with perfect feature correspondence the range would extend to between 0.5 - 30 m. Stereo camera separation limits the near range, while optical resolution limits the far range. Image frame sizes are 160x120 pixels. Increasing this size will improve far range characteristics but will also decrease frame rate. Image preprocessing proved to be less appropriate than precision camera alignment in this application. A proof of concept for object recognition shows promise for applications with more precise object detection. Future recommendations are offered in all five areas of vision processing.

AFIT Designator

AFIT-GE-ENG-00M-08

DTIC Accession Number

ADA380765

Comments

The author's vita page is omitted.

Download

Share

COinS