Determination of Aircraft Orientation for a Vision-Based System Using Artificial Neural Networks

  • Authors:
  • Sanjeev Agarwal;Subhasis Chaudhuri

  • Affiliations:
  • Intelligent Systems Center, University of Missouri, Rolla, MO 65401;Department of Electrical Engineering, Indian Institute of Technology, Powai, Bombay, 400 076

  • Venue:
  • Journal of Mathematical Imaging and Vision
  • Year:
  • 1998

Quantified Score

Hi-index 0.00

Visualization

Abstract

An algorithm for real-time estimation of 3-D orientationof an aircraft, given its monocular, binary image from anarbitrary viewing direction is presented. This being an inverseproblem, we attempt to provide an approximate but a fastsolution using the artificial neural network technique. A setof spatial moments (scale, translation, and planar rotationinvariant) is used as features to characterize different viewsof the aircraft, which corresponds to the feature spacerepresentation of the aircraft. A new neural network topologyis suggested in order to solve the resulting functionalapproximation problem for the input (feature vector)-output(viewing direction) relationship. The feature space ispartitioned into a number of subsets using a Kohonen clusteringalgorithm to express the complex relationship into a number ofsimpler ones. Separate multi-layer perceptrons (MLP) are thentrained to capture the functional relations that exist betweeneach class of feature vectors and the corresponding targetorientation. This approach is shown to give better results whencompared to those obtained with a single MLP trained for theentire feature space.