Analysing aerial photographs with ADAM

Author

Smith, Guy ; Austin, James

Author_Institution

Dept. of Comput. Sci., York Univ., UK

Volume

3

fYear

1992

fDate

7-11 Jun 1992

Firstpage

49

Abstract

The use of the advanced distributed associative memory (ADAM) in the analysis of features in infrared line scan imagery is described. An ADAM neural network maps an input vector or image to an output vector or image. The ADAM neural network is capable of recognizing features in aerial images using a deterministic noniterative training algorithm. A novel form of weight update allowing a weighted training procedure and a binary runtime system to increase the classification success of ADAM is presented. The results of segmenting urban and field areas, as well as road identification, are discussed

Keywords

content-addressable storage; image recognition; image segmentation; neural nets; ADAM; IR line scan imagery; advanced distributed associative memory; aerial photographs; binary runtime system; deterministic noniterative training algorithm; image recognition; neural network; road identification; weighted training procedure; Associative memory; Computer architecture; Computer science; Face recognition; Image analysis; Image converters; Image segmentation; Roads; Sparse matrices; Testing;

fLanguage

English

Publisher

ieee

Conference_Titel

Neural Networks, 1992. IJCNN., International Joint Conference on

Conference_Location

Baltimore, MD

Print_ISBN

0-7803-0559-0

Type

conf

DOI

10.1109/IJCNN.1992.227038

Filename

227038