Capture and location of blobs in digital images
Date
1993
Authors
Seelemann, Sven Alexander
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
This thesis describes three methods of detecting the location of active targets using machine vision. In the field of visual servoing, the detection and location of targets is the first step in controlling the motion of a platform with respect to the target. In certain long range, outdoor settings, active targets are the only feasible target type. For each method (centroid, linear parameter estimation, and non- linear parameter estimation), a theoretical resolution and accuracy analysis is performed. This analysis is followed by an experimental suite that determines real resolution and accuracy of each method. The linear parameter estimation technique does not properly locate active targets. Both the centroid and non- linear parameter estimation can achieve accuracies of 0.25 pixels in locating active targets.
Description
Keywords
UN SDG 11: Sustainable Cities and Communities