Spatial spectral band selection for enhanced hyperspectral remote sensing classification applications
| Autor: | Peter Godfree, Chris McCullough, Changfeng Yuan, Ruben Moya Torres, Peter W. T. Yuen, Johathan Piper |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2020 |
| Předmět: |
010504 meteorology & atmospheric sciences
Computer science hyperspectral imaging Feature extraction 0211 other engineering and technologies 02 engineering and technology lcsh:Computer applications to medicine. Medical informatics 01 natural sciences Convolutional neural network curse of dimensionality lcsh:QA75.5-76.95 Article spatial spectral band selection Preprocessor Radiology Nuclear Medicine and imaging lcsh:Photography band selection Electrical and Electronic Engineering Cluster analysis mutual information 021101 geological & geomatics engineering 0105 earth and related environmental sciences Hughes phenomenon accuracy-dimensionality characteristics business.industry Hyperspectral imaging Pattern recognition Spectral bands Mutual information lcsh:TR1-1050 Computer Graphics and Computer-Aided Design classification lcsh:R858-859.7 lcsh:Electronic computers. Computer science Computer Vision and Pattern Recognition Artificial intelligence business Curse of dimensionality |
| Zdroj: | Journal of Imaging Volume 6 Issue 9 Journal of Imaging, Vol 6, Iss 87, p 87 (2020) |
| Popis: | Despite the numerous band selection (BS) algorithms reported in the field, most if not all have exhibited maximal accuracy when more spectral bands are utilized for classification. This apparently disagrees with the theoretical model of the &lsquo curse of dimensionality&rsquo phenomenon, without apparent explanations. If it were true, then BS would be deemed as an academic piece of research without real benefits to practical applications. This paper presents a spatial spectral mutual information (SSMI) BS scheme that utilizes a spatial feature extraction technique as a preprocessing step, followed by the clustering of the mutual information (MI) of spectral bands for enhancing the efficiency of the BS. Through the SSMI BS scheme, a sharp &rsquo bell&rsquo shaped accuracy-dimensionality characteristic that peaks at about 20 bands has been observed for the very first time. The performance of the proposed SSMI BS scheme has been validated through 6 hyperspectral imaging (HSI) datasets (Indian Pines, Botswana, Barrax, Pavia University, Salinas, and Kennedy Space Center (KSC)), and its classification accuracy is shown to be approximately 10% better than seven state-of-the-art BS schemes (Saliency, HyperBS, SLN, OCF, FDPC, ISSC, and Convolution Neural Network (CNN)). The present result confirms that the high efficiency of the BS scheme is essentially important to observe and validate the Hughes&rsquo phenomenon in the analysis of HSI data. Experiments also show that the classification accuracy can be affected by as much as approximately 10% when a single &lsquo crucial&rsquo band is included or missed out for classification. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|