An essential step in bispectral imaging is the recovery of the object’s Fourier phase from the bispectral phase. Such reconstruction can be performed by recursive or least-squares (minimization) methods. It is generally accepted that least-squares methods perform better because of their higher noise tolerance. Two weighted least-squares minimization schemes are compared, and the use of the error-reduction algorithm is suggested as a way to overcome the stagnation at local minima common to minimization problems.
© 1996 Optical Society of America
Pedro Negrete-Regagnon, "Phase recovery from the bispectrum aided by the error-reduction algorithm," Opt. Lett. 21, 275-277 (1996)