Semi-Supervised Adaptive Pseudo-Label Feature Learning for Hyperspectral Image Classification in Internet of Things
Huayue Chen, Jie Ru, Haoyu Long, Jialin He, Tao Chen, Wu Deng
Abstract
Hyperspectral image (HSI) in Internet of Things (IoT) is a typical small sample data set, which is difficult and costly to label samples manually. In the feature extraction, it is difficult to increase the interclass distance and reduce the intraclass variance according to the limited label information, resulting in easy misclassification of the extracted features. To solve this problem, this article proposes an adaptive pseudo-label feature learning (APFL) model. In the APFL model, a hybrid distance pseudo-label generation (HDPG) method was designed to generate pseudo-labels by iterative multiscale superpixel segmentation using the spectral-spatial mixing distance information, while a pseudo-label feature generation (PFG) method was designed to generate pseudo-label features using pseudo-labels to capture the intraclass average vectors of HSI principal component features. Finally, the extracted pseudo-label features are classified at the pixel level. This APFL model can effectively reduce the intraclass variance and increase the interclass distance of the HSI data, thus improving the interclass separability. We have done comparative verification experiments on five commonly used HSI data sets in IoT. Compared with the current advanced feature extraction methods and classification methods, the proposed APFL model in this article has higher classification accuracy.