The Application and Comparison of Feature Extraction and Digital Signal Processing Methods to Reconnaissance Data.

摘要

The research is concerned with the feature extraction,encoding,transmission,and interpretation of two different classes of reconnaissance information. The first type is temporal signals and is typified by acoustic,magnetic,and seismic sources. The second is imagery data and includes video,IR,and radar displays. Stress is placed on real time on-line digital processing of these types of information. During the first year,adaptive techniques have been developed for enhancing,extracting,and tracking nonstationary signals typical of the first class of sensors. Digital processing and interactive computer graphics for displaying the time varying spectrum is used to test the analytical results experimentally. Second,design techniques have been found which permit the design of long binary signals,particularly useful for synchronization,having prespecified autocorrelation properties using only shorter waveforms with desireable properties. (Modified author abstract)