A novel unsupervised bands selection algorithm for hyperspectral image

作者：Du, Xiaoping; Chen, Hang^*; Liu, Zhengjun; Yang, Chengwei

来源：Optik, 2018, 158: 985-996.

DOI：10.1016/j.ijleo.2018.01.001

• Hyperspectral
• Dimension reduction
• Bands selection
• ABS
• JSKF

A novel bands selection method based on ABS (Adaptive Band Selection) and JSKF (Joint Skewness-Kurtosis Figure) is proposed in this paper. The hyperspectral data is separated into different sub-spaces by employing ABS and JSKF respectively. Subsequently a novel optimal bands selection methodNIA (Normalization Index Algorithm) is proposed to select the optimal bands according to the value of ABS and JSKF. Both the high correlations of the adjacent bands and the richness of complementary information are considered in detail in this proposed method. Some numerical simulations are made to test the validity and capability of the proposed dimension reduction algorithm.

• 出版日期2018
• 单位哈尔滨工业大学