Two Dimensional Spectral Representation

• Published in: IEEE transactions on geoscience and remote sensing Vol. 62; p. 1
• Main Authors: Kang, Xudong, Zhu, Yongxiang, Duan, Puhong, Li, Shutao
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.01.2024; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Classification; Computational modeling; Computer vision; Convolutional neural networks; Data analysis; Data processing; Deep learning; Feature extraction; Hyperspectral image(HSI); Hyperspectral imaging; Image classification; Markov processes; Representations; spectral representation; spectral visualization; Visualization
• ISSN: 0196-2892; 1558-0644
• DOI: 10.1109/TGRS.2023.3343909

In this paper, a two-dimensional spectral representation is proposed for the visualization and classification of hyperspectral images. First, several sequence data processing methods, i.e., Gramian Angular Field Algorithm (GAF), Markov Transition Field (MTF), and Recurrence Plot (REP) are applied to obtain multiple two-dimensional features of a one-dimensional spectrum. Second, the two-dimensional spectral features are stacked together to form the final two-dimensional spectral representation. Finally, many excellent classifiers in computer vision field are applied on the two-dimensional spectral representation to obtain the final classification result. Furthermore, 114 target spectral visualization maps are established based on their one-dimensional spectra. Experimental results reveal that the two dimensional spectral representation has multiple advantages in terms of better visual quality and classification accuracies. The code of this work is available at https://github.com/zhuyongxiang1/two-dimensional-spectral-representation.