Multispectral image classification using independent component analysis and data dimensionality expansion approaches

Abstract

In this paper we inestigate the application of independent component analysis (ICA) to multispectral image classification. In particular, the performance of Joint Approximate Diagonalization of Eigenmatrices (JADE) algorithm is studied. ICA is particularly useful for classifying objects with unknown spectral signatures in an unkown image scene, i.e., unsupervised classification, because it does not require any prior information about class signatures. Since the weight matrix in ICA is a square matrix for the purpose of mathematical tractability, the number of objects that can be classified is equal to the data dimensionality, i.e., the number of spectral bands. When the number of spectral bands is very small (e.g., 3-band CIR photograph), it is impossible to classify all the different objects present in an image scene with the original data. In order to solve this problem, we present a data dimensionality expansion technique to generate artificial bands for additional spectral measurements. The results from such a nonlinear band generation approach are compared with a linear band generation method using cubic spline interpolation of pixel spectral signatures. The experiments demonstrate that such a nonlinear band generation approach can expand the applicability of ICA and improve the classification accuracy.