PASCAL - Pattern Analysis, Statistical Modelling and Computational Learning

Robust Shape and Polarisation Estimation Using Blind Source Separation
Lichi Zhang and Edwin Hancock
In: Computer Analysis of Images and Patterns - 14th International Conference, CAIP 2011, August 29-31, 2011, Seville, Spain.

Abstract

In this paper we show how to use blind source separation to estimate shape from polarised images. We propose a new method which does not require prior knowledge of the polariser angles. The two key ideas underpinning the approach are to use weighted Singular Value Decomposition(SVD) to estimate the polariser angles, and to use a mutual information criterion function to optimise the weights. We calculate the surface normal information using Fresnel equation, and iteratively update the values of weighting matrix and refractive index to a recover surface shape. We show that the proposed method is capable of calculating robust shape information compared with alternative approaches based on the same inputs. Moreover, the method can be applied when using uncalibrated polarisation filters. This is the case when the the subject is difficult to stabilse during image capture.

EPrint Type:Conference or Workshop Item (Paper)
Project Keyword:Project Keyword UNSPECIFIED
Subjects:Machine Vision
ID Code:8553
Deposited By:Edwin Hancock
Deposited On:13 February 2012