Noise estimation of polarization-encoded images by Peano-Hilbert fractal path

In the framework of Stokes parameters imaging, polarization-encoded images have four channels. The potential of such multidimensional structure comes from the set of physical information they carry about the local nature of the target. However, the noise that affects the intensity measurement may induce the non physical meaning of Stokes parameters making awkward their analysis and interpretation. This induces the need for a proper tool that allows the processing of polarization-encoded images while respecting their physical content. In this paper a new method to filter the additive noise of polarimetric measurement is addressed. This method is based on two multispectral filtering methods combined with a transformation of Stokes channels following a fractal path. The proposed algorithm is a tradeoff between the filtering of polarization-encoded images and the preserving of their physical content. The statistical performances of the method are tested on simulated and real images using Bootstrap re-sampling.