Object Identification and Manipulation Using Silhouette Moments.
Abstract
The theory of silhouette moments are presented along with results for classifying clean and noisy images. Noise reduction by image averaging is examined comparatively in image and moment spaces. Results are presented for classifications using subsets of the complete moment set. An inverse moment transform procedure is developed and demonstrated. Keywords include: Shape analysis; moments; Fourier descriptors; pattern recognition; noise reduction; FLIR imagery; silhouette imagery; inverse moment transform; Legendre moments; image averaging; moment subset classification; six airplane experiment; aspect ratio normalization; variance balancing.
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 01, 1985
- Accession Number
- ADA154755
Entities
People
- A. P. Reeves
- R. W. Taylor
Organizations
- Cornell University College of Engineering