Dual Difference Filtering: A Replacement for Interpolation and Subtraction to Detect Changes in Misregistered Signals
Abstract
Dual Difference Filtering (DDF), a symmetric signal processor developed as a means of detecting changes in a pair of displaced digital signals, supplants interpolation and subtraction. For oversampled signals, DDF performance can be described by an error spectrum that is the product of the spectrum of the underlying continuous signal and an attenuation factor that depends only on the filters. For undersampled or stochastic signals, performance is similarly describable, but in an average sense. Simple formulas are found for the attenuation factors in terms of the filters' real or frequency space descriptions. A fundamental equation for error minimization is also derived that is useful when the form of the power spectrum is known. Optimal DDFs are computed for a power-law spectrum that is appropriate for imagery. Also, flexible suboptimal design methods are developed that produce superior performance for a wide range of signal spectra. Typical signal to clutter improvements of 30 dB relative to standard interpolation/subtraction methods are demonstrated.... Change detection, Interpolation, Digital Filtering, Linear estimation, Nyquist sampling, Moving target indication, Image processing.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 30, 1992
- Accession Number
- ADA259440
Entities
People
- Alan P. Schaum
Organizations
- United States Naval Research Laboratory