Canonical relations of subspaces in multi-sensor data analysis

In multisensor data analysis, scene details can be extracted via subspace methods without any prior information on the scene. In these decomposition techniques, data is projected into a new space so that the information in the data is highlighted. In this study, Principal Component Analysis, Independent Component Analysis and Minumum Noise Fractions method are applied to a multi-sensor data composed of radar, visible, and infrared images. Canonical correlations between these subspaces are investigated via Canonical Correlation Analysis. This equalization subspace offers a new point of view in the realm of multi-sensor data analysis.