Abstract
Hyperspectral unmixing is a powerful tool for the remote sensing image mining. Nonnegative matrix factorization (NMF) has been adopted to deal with this issue, while the precision of unmixing is closely related with the local minimizers of NMF. We present two novel initialization strategies that is based on CUR decomposition, which is physically meaningful. In the experimental test, NMF with the new initialization method is used to unmix the urban scene which was captured by airborne visible/infrared imaging spectrometer (AVIRIS) in 1997, numerical results show that the initialization methods work well.
Highlights
Given a non-negative matrix M ∈ Rm×n and a desired rank k min(m, n), the non-negative matrix factoriza tion aims to find two matrices U ∈ Rm×k and V ∈ Rk×n with the non-negative elements such that M ≈ UV (1)Problem (1) is commonly reformulated as the following minimization problem: min M −UV s.t
Each row vector of the original matrix M represents an observed mixed spectrum at certain wavelength, while each column of M give the spectral signature of a given pixel
We compare our initialization strategy with the Urban hyperspectral image, which is taken from HYper-spectral Digital Imagery Collection Experiment (HYDICE) air-borne sensors
Summary
Given a non-negative matrix M ∈ Rm×n and a desired rank k min(m, n) , the non-negative matrix factoriza tion aims to find two matrices U ∈ Rm×k and V ∈ Rk×n with the non-negative elements such that. (2016) CUR Based Initialization Strategy for Non-Negative Matrix Factorization in Application to Hyperspectral Unmixing. Each row vector of the original matrix M represents an observed mixed spectrum at certain wavelength, while each column of M give the spectral signature of a given pixel. In such way, each element Mij of M is equal to the reflectance spectra of the jth pixel at the ith wavelength. Applying the column selection technique of CUR in the initialization stage of NMF is called Acol. We give two new initialization methods for NMF which is based on CUR
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
