A fast band selection method to increase image contrast for multispectral image segmentation

Multispectral microscopy for applications in histology and cytology has attracted much attention in recent years. It has been shown that the unique transmission spectra of biological tissue provides additional information that is potentially useful for better classification of the pathologies. However, irrelevant features in multispectral data may affect results and performance of data analysis methods. In this paper, a fast band selection method is proposed to increase the contrast-to-noise ratio such that relevant information is maximized while reducing the number of spectral bands. The new method obtains an optimal selection of bands by solving a specific objective function with low computational costs. A thresholding criterion is developed to decide on the number of bands to be selected.