基于独立分量分析的高光谱影像小目标提取研究

基金项目: 国家863计划资助项目(2002AA783050)

详细信息

Research on Small Targets Extraction Based on ICA in Hyperspectral

¹ Institute of Surveying and Mapping,Information Engineering University,66 Middle Longhai Road,Zhengzhou 450052,China

Funds: 国家863计划资助项目(2002AA783050)

摘要: 借助快速独立分量分析(FICA)将高维数据中隐藏的目标信息集中投影到低维特征影像中,然后以峰度为特征度量指标选择特征影像,最后用以偏斜为指标的直方图分割方法提取小目标。实验证明,此算法精度较高,适用于对高光谱影像中的小目标进行提取。
- 快速独立分量分析 /
- 偏斜 /
- 峰度 /
- 目标提取
Abstract: This paper present a small targets detection approach based on independent component analysis in hyperspectral data.Firstly,this algorithm uses fast independent component analysis(FICA) approach to collect the targets information hided in the high-dimensional data and projects them into low-dimensional space.Secondly,selects feature images with kurtosis.At last,extracts small targets with histogram image segmentation method which has been labeled by skewness.
- fast independent component analysis /
- skewness /
- kurtosis /
- targets extraction