Local mapping for multispectral image visualization

In this paper, fusion of multispectral images for visualization is aimed at. The techniques that are studied are mappings of the scattergram onto a one-dimensional feature space. Linear (Principal Component Analysis) as well as non-linear mapping (Self-Organizing Maps) are discussed. In this paper, local mappings are studied. Two spatially local techniques are proposed and discussed: a pixel-based and a block-based technique, where a local map of the scattergram is calculated for each pixel and for a block of pixels respectively. In the experimental section, the proposed techniques are aplied to PCA and SOM. For a test image, they are compared to each other and to global PCA and SOM.