Multimodality Image Fusion for 3-D Model Building with Applications

Abstract

In this investigation, the authors propose a methodology for 3-D model building through the fusion of multimodality data provided from space-borne and/or air-borne sensors. A 3-D model of a target area can be built using different data types (e.g., Landsat MSS data, AVIRIS hyperspectral data, range and/or elevation (DEM) data). Data fusion and integration can be performed either at the data level or at the decision level. Multispectral and hyperspectral data sets can be classified locally (e.g., using the Fuzzy C-Mean classifier), then decision fusion is used to fuse the local decision classes, or data fusion techniques can be used to fuse the data sets into one data set for classification. Probabilistic and evidential methods for data fusion are investigated in this study. Topographic data from range scanners (ALTM) and/or radar (DTED) can be integrated, after registration, with the classification results of the multispectral and hyperspectral data to produce the final 3-D model of the sensed target area. Efforts addressed during 2001 included data collection, fuzzy classification for hyperspectral data, and fusion of range and stereo data. Efforts conducted during 2002 included updating the available programs and testing them with new data sets; statistical, fuzzy logic classifiers and data fusion algorithms; feature extraction; and image registration and fusion. Efforts conducted during 2003 included the development of high dimensional density estimation algorithms, statistical multi-stages classifier for multispectral data, and unsupervised statistical classification. Efforts during 2004 included mean field theory for density estimation using support vector machines and a unified framework for MAP estimation in remote sensing image segmentation. The titles of 13 seminars that were given during this contract period are listed. (14 refs.)

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Nov 02, 2004
Accession Number
ADA428114

Entities

People

  • Aly A. Farag

Organizations

  • University of Louisville

Tags

Communities of Interest

  • Materials and Manufacturing Processes
  • Sensors

DTIC Thesaurus Topics

  • Algorithms
  • Computer Vision
  • Data Analysis
  • Data Fusion
  • Data Sets
  • Feature Extraction
  • Fuzzy Logic
  • Image Processing
  • Image Registration
  • Image Segmentation
  • Machine Learning
  • Mean Field Theory
  • Probability
  • Remote Sensing
  • Stochastic Processes
  • Supervised Machine Learning
  • Three Dimensional

Readers

  • Computer Vision.

Technology Areas

  • AI & ML
  • Space