PREPROCESSOR FOR SPECTRAL PATTERN CLASSIFICATION SYSTEMS

Abstract

The signal to be classified is sampled and the samples are multiplied by weighting functions prior to performing discrete Fourier transforms, power calculations, and normalization thereon so that the preprocessor is essentially a plurality of channel bandpass filter stages, the k-th one of which has a frequency response approximating an ideal filter defined by ##EQU1## where k=O, α, 2α, 3α, . . . , rα α=constant, 1≦α≦N/2 r+1=greatest integer in N/α This preprocessor transforms blocks of one dimensional data into spectra in such a way that data blocks coming from similar sources will have spectra that are close to one another in spectrum space. The discrete Fourier transform is modified to remove or reduce the phase dependency problem and to enhance the clustering of similar spectra by the weighting function above.

Source: Google Patents

35 USC §181 Secrecy Order

Inventor

• 1JERALD L. BAUCK