An Open-Set Domain Adaptation Framework for Hyperspectral Image Classification With Pixel-Aware Weighting and Decoupled Alignment

Abstract

Recent studies have shown that deep domain adaptation (DA) techniques perform excellently in cross-domain hyperspectral image (HSI) classification. However, these methods typically assume that the source domain and the target domain share the same class set, while in practice, the target domain may include unknown classes, and direct alignment can result in negative transfer. Moreover, in HSI classification based on deep learning, using the label of the central pixel to represent the label of the image patch may lead to feature bias due to the uncertainty of the labels of neighboring pixels, thereby reducing the generalization performance of the model. To address this, this letter proposes an open-set DA (OSDA) framework, including a pixel-aware adaptive weight learning (PAWL) module and a decoupled dual alignment (DDA) strategy. The PAWL module effectively reduces the feature bias caused by inconsistency in neighboring pixel labels by analyzing the uncertainty of neighboring pixel labels and using adaptive weight learning, thereby improving recognition performance in open-set environments. The DDA strategy decouples the features of the source domain and target domain into known and unknown classes and aligns them separately to mitigate negative transfer. Experiments on two cross-scene hyperspectral datasets validated the effectiveness of the method. Our source code is available at https://github.com/Li-ZK/PWDA-2025