Phase extraction from electronic speckle patterns by statistical analysis

Abstract

In electronic speckle pattern interferometry (ESPI), speckles are information carriers as well as noise that hinders the extraction of high quality phase. This paper presents a phase extraction method based on the statistical property of speckles. Assuming that speckle related phase is a random variable having a uniform distribution, the grey level variance of a number of pixels is found to be related to the modulation intensity of a speckle pattern. The relation is used to establish a connection between the phase to be measured and the variance of grey level difference between two speckle patterns. Subsequently, a phase map wrapped in [0,π) is extracted. In order to obtain a standard 2π wrapped phase map, an initial one step phase shift is introduced. The phase value of a pixel under consideration is obtained from the grey levels of its N × N neighbouring pixels. The optimal value of N is obtained based on a qualitative analysis of the initial results. With an appropriate value of N, an accuracy of 1% can be achieved. © 2004 Elsevier B.V. All rights reserved.