Speckle noise in optical fibre sensing

A study using the Maximum Entropy method

L. Stergioulas, A. Vourdas and G.R. Jones
Department of Electrical Engineering and Electronics
University of Liverpool
P.O. Box 147
Liverpool L69 3BX

Abstract

The spatial distribution of the intensity of light at the output of an optical fibre can be useful in optical fibre sensing applications. Noise however limits the use of the detailed shape of this distribution for the development of highly sensitive sensors. In this paper we use the maximum entropy method to obtain the most likely spatial distribution consistent with the noisy data.

An image captured by a CCD camera at the output of an optical fibre has been divided into many small regions. In each region we find the first N moments and then calculate another distribution that has the same first N moments but also the maximum entropy. Noise reduction is thus associated with the elimination of the detailed information in the higher moments of the observed distribution. Results for various sizes of the regions and various values of the number of moments N will be presented.

A second approach has also been considered where many images have been captured at different times providing a distribution of the intensity at each pixel. We calculate the average \mu_i and standard deviation \sigma_i of the intensity at the pixel i and maximise the entropy with the constraint that

\chi^2 = \sum_i \frac{|I_i-I_i'|^2}{\sigma_i^2}

is smaller than a certain value. Several examples will be presented.

The work applies the powerful maximum entropy technique in the context of noise reduction in optical fibre systems and could play an important role in increasing the sensitivity of these systems.


MaxEnt 94 Abstracts / mas@mrao.cam.ac.uk