A new algorithm for signal reconstruction from the phase of bispectrum, and its applications in image enhancement

There are applications in signal processing where the bispectrum magnitude of the source signal is distorted while the phase remains clean or shifts linearly. Many different algorithms have been developed, to reconstruct a signal from both the magnitude and phase of its bispectrum. We introduce an iterative scheme, the bispectrum iterative reconstruction algorithm (BIRA), for reconstruction of FIR sequence from the phase of the bispectrum. Then we propose procedures to optimize BIRA algorithm and introduce a new algorithm named adaptive BIRA (ABIRA). Finally, we reconstruct an image which convolved with a linear phase channel and also an image that had been corrupted by additive white Gaussian noise