Bispectral reconstruction of 2-D signals in noise

摘要

Summary form only given. The properties of the bispectrum have been applied to two-dimensional signals to provide some possibilities for a variety of signal and image processing applications, ranging from signal recovery from noisy observations to reconstruction of 2-D objects moving in noise. The complex bispectrum B( omega /sub 1/, omega /sub 2/, lambda /sub 1/, lambda /sub 2/), which is a 4-D quantity for a 2-D signal h(m,n), can be expressed in terms of the signal Fourier transform H( omega , lambda ). The bispectrum preserves the magnitude and the phase information of the signals and is identically zero for Gaussian noise processes.