Attention-based Sparse and Collaborative Spectral Abundance Learning for Hyperspectral Subpixel Target Detection

Dehui Zhu, Ping Zhong, Bo Du, Liangpei Zhang

Abstract: Highlights•An attention-based spectral abundance learning model is constructed for subpixel target detection in hyperspectral imagery.•Pixel attention-based linear prediction strategy is designed to select typical backgrounds, avoiding possible anomaly and noise contaminations.•Sparse and collaborative constraints are incorporated to govern abundance matrices for accurate target spectral abundance learning.•Optimization and iteration of the convex optimization model are realized via the alternating direction method of multipliers algorithm.