This paper proposes the employment of both day-and-night thermal-infrared imagery density values in a multivariate model for automatic pattern recognition. The analytical model is discriminatory analysis, utilizing the discriminatory function, derived from the day-and-night- density vectors, for pattern recognition and mapping purposes. For mapping purposes, two models are developed further. First, to identify and map single objects, such as houses, roads, and water, a point-classification system is used. The discriminant function is derived from only two vectors. Second, to identify scenes, such as residential areas, factories and croplands, an areal-classification model is used. The discriminant function is then derived from 10 parameters (5 from X-Z axes, and 5 from Y-Z axes) extracted from the density surface constructed by densitometer scanning and computer graphics. The results indicated that an accuracy rate of well over 85 percent can be accomplished for automatic mapping purposes. A low-cost automatic imagery interpretation system can be obtained.

  • Availability:
  • Supplemental Notes:
    • Presented at the Fall Technical Meeting, American Society of Photogrammetry, Washington, D.C., September 1974.
  • Corporate Authors:

    American Society of Photogrammetry

    105 North Virginia Avenue
    Falls Church, VA  United States  22046
  • Authors:
    • Hsu, S
  • Publication Date: 1975-5

Media Info

Subject/Index Terms

Filing Info

  • Accession Number: 00097214
  • Record Type: Publication
  • Files: TRIS
  • Created Date: Oct 18 1975 12:00AM