Abstract

A feature space processor for multiclass distortion-invariant pattern recognition is detailed in Section 2. A moment feature vector space is considered. Test data [1, 2] on a robotic database are summarized in Section 3. Results on a ship database, using real input imagery with references from models is presented with attention to preprocessing, distortion parameter estimation, and class identification are advanced in Section 4.

© 1985 Optical Society of America