Improving Classification Accuracy of Radar Images Using a Multiple-Stage Classifier
Abstract
A simple method was introduced to classify radar image samples repeatedly for achieving a higher accuracy than by using a single-stage classifier. A Sobel edge operator was applied between the stages of classification to enhance the difference in texture between categories of radar image samples, thus reducing the overlap of image categories. Keywords: Radar image feature extraction; Texture; Histogram; Classification; Pattern recognition; and Edge operators.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 1988
- Accession Number
- ADA200291
Entities
People
- Neil D. Fox
- Pi-fuay Chen
Organizations
- Geospatial Research Laboratory