A 2D hyperspectral Raman map (pixel size 20nm x 20nm) of polymer (PEDOT) nanowires (diameter 100nm) is decomposed into spectral sources and respective abundance maps using bounding hyperplane-based hyperspectral unmixing. The resulting abundance maps are independently deconvolved using an interior point least squares algorithm with a positivity constraint. Spectral changes along and across the wires, as well as differences between wires, can be resolved to better 50nm, a significantly better than the 400nm instrument resolution.

This is poster number 63 in Poster Session

Authors:

Dominik J. Winterauer (Renishaw plc,; Institut des Matériaux Jean Rouxel)

Tim Batten (Spectroscopy Products Division, Renishaw plc)

Said Moussaoui (Laboratoire des Sciences du Numérique de Nantes UMR CNRS 6004, Ecole Centrale de Nantes )

Bernard Humbert (Institute des Materiaux Jean Rouxel Nantes, University of Nantes)

Daniel Funes-Hernando (Institut des Matériaux Jean Rouxel)

Keywords:

hyperspectral image restoration, image deblurring, image enhancement, inverse problems, nonlinear optimization