Texture Segmentation Using Localized Spatial Filtering

Abstract

Spatial filters based on two-dimensional Gabor functions are applied to the image segmentation problem using textural differences for discrimination. In order to provide class separability, the textural content of a scene must have spatial variations which exhibit characteristic differences in frequency and/or directional bandwidths. This idea stems from discoveries in vision research that the Gabor functions model effectively the architecture of the neural receptive fields in the striate visual cortex and in the belief that such functions can play an important role in the analytical study of machine vision, pattern recognition and image processing. This new technique in image segmentation does not required burdensome machine data processing as compared with other techniques based on pixel classification. In this paper the technique is applied to some SAR images of the open ocean surface and of some ice fields. The results are very encouraging.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jan 26, 1990
Accession Number
ADA218064

Entities

People

  • Li-jen Du

Organizations

  • United States Naval Research Laboratory

Tags

Communities of Interest

  • Energy and Power Technologies
  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Aspect Ratio
  • Availability
  • Boundaries
  • Computer Vision
  • Coordinate Systems
  • Frequency
  • Frequency Domain
  • Identification
  • Image Processing
  • Image Segmentation
  • Information Processing
  • Military Research
  • Pattern Recognition
  • Security
  • Spatial Filtering
  • Two Dimensional
  • Visual Cortex

Readers

  • Computer Vision.
  • Systems Analysis and Design
  • Vision Science/Vision Psychology/Cognitive Neuroscience.

Technology Areas

  • AI & ML
  • AI & ML - Machine Learning Algorithms
  • AI & ML - Neural Networks