Abstract

It has been shown¹ that imagery can be recovered directly from speckle patterns without the intermediate step of estimating image autocorrelation.² The present work demonstrates the imagery recovered from simulated speckle patterns with additive Gaussian noise. Excellent images can be reconstructed, even when corrupted by significant noise, if support constraints are tight and outline the objects. Images were reconstructed from 100 speckle patterns with an ideal reconstruction signal-to-noise ratio of 10. Gaussian-distributed speckle noise with 10% of the speckle-pattern energy had a negligible effect on the signal-to-noise ratio in the reconstructed imagery. When the noise content was increased to 50% of the speckle pattern energy, the signal-to-noise ratio decreased by a factor of 5. The image quality also dropped rapidly with decreasing support tightness.

© 1990 Optical Society of America