Abstract
Hyper-spectral remote sensing has the defects of huge data size and massive information redundancy, especially when processing pixel un-mixing, which always has high computation complexity. This paper proposes a band selection method for hyper-spectral pixel un-mixing, based on synthesized parameters of information content. It uses Kullback-Leibler divergence and mutual information with respective weights to construct a new comprehensive information matrix. The matrix can indicate the overall distribution of data's spectral information. According to the comprehensive information matrix, the method can select a small-number band combination from the massive bands of initial data in an iterative way. The experimental results show that, the method is effective in decreasing data volume and retaining effective spectral information, its result is better than those of similar algorithms. This method can be chosen as an effective preprocessing step for hyper-spectral pixel un-mixing.
Export citation and abstract BibTeX RIS
Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.