Accession Number : ADP000116

Title :   Matching Linear Features of Images and Maps,

Corporate Author : UNIVERSITY OF SOUTHERN CALIFORNIA LOS ANGELES DEPT OF COMPUTER SCIENCE

Personal Author(s) : Medioni,Gerard G.

Report Date : 1982

Pagination or Media Count : 9

Abstract : We present two methods to solve the problem of matching linear features extracted from an aerial image with a set of linear features derived from a map or another view of the same scene. The first method, using discrete relaxation, is very efficient but requires the model to have a small number of elements. The other one, called the kernel method, demonstrates how drastically we can improve the running time if we know a few pairs of matched elements. Illustrative examples are provided, and extensions are discussed. (Author)

Descriptors :   *Image processing, Optical images, MAPS, Edges, Aerial reconnaissance, Matching, Aerial photographs, Relaxation, Patterns, Pattern recognition, Computer applications, Flow charting

Distribution Statement : APPROVED FOR PUBLIC RELEASE