An Evaluation of Shape Methods for Helicopter Classification and Orientation Determination
Abstract
The performance of two competing methods are evaluated for classification and orientation determination accuracy. Helicopter fuselage silhouettes are classified as to their type and orientation. The methods are tested for sensitivity to imaging noise and the reference library sampling density. Under all conditions of imaging noise and library sampling density, the Fourier descriptor method out performs the Walsh points method for helicopter type classification. Helicopter type classification performance for the Fourier method ranges from 99% to 84% even under the most severe conditions examined. In almost all cases, the Fourier method exhibits median angle errors one-half those os the Walsh points method. Median angle errors of one-half the library sampling interval have been attained using the Fourier descriptor method. A scheme for helicopter rotor segmentation and orientation determination including the necessary additional library enhancements is also proposed.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 1990
- Accession Number
- ADA227326
Entities
People
- Timothy A. Grogan
Organizations
- Battelle Memorial Institute