Accession Number : AD0768347

Title :   Space Object Identification by Filtered Fourier Transform Pattern Recognition Algorithm.

Descriptive Note : Master's thesis,

Corporate Author : AIR FORCE INST OF TECH WRIGHT-PATTERSON AFB OHIO SCHOOL OF ENGINEERING

Personal Author(s) : Malinski,William L.

Report Date : JUN 1973

Pagination or Media Count : 107

Abstract : Light intensity signals reflected from three classes of orbiting rocket bodies were analyzed using one-dimensional Fourier transforms. Low-frequency filtering in the transform domain and the Euclidean distance metric were used to classify the signals into the three classes. Using a portion of the data, linear decision boundaries were constructed by an adaptive training algorithm. It was found that the low-frequency filtered one-dimensional Fourier-transform domain gave good separation of the three classes of rocket bodies analyzed. A method of automated space object identification is proposed for non-stabilized satellites. It is suggested that the algorithm used in the study is also applicable to data collected via radar. (Author)

Descriptors :   (*PATTERN RECOGNITION, OPTICAL TRACKING), (*OPTICAL TRACKING, SPECTRUM SIGNATURES), (*SATELLITE TRACKING SYSTEMS, TARGET DISCRIMINATION), IDENTIFICATION SYSTEMS, ROCKETS, DATA PROCESSING, ALGORITHMS, CURVE FITTING, GRAPHICS, FOURIER ANALYSIS, THESES

Subject Categories : Optical Detection and Detectors

Distribution Statement : APPROVED FOR PUBLIC RELEASE