Unsupervised Change Detection in Multitemporal VHR Images Based on Deep Kernel PCA Convolutional Mapping Network

• Published in: IEEE transactions on cybernetics Vol. 52; no. 11; pp. 12084 - 12098
• Main Authors: Wu, Chen, Chen, Hongruixuan, Du, Bo, Zhang, Liangpei
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 01.11.2022; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Change detection; Change detection (CD); Clustering; Convolution; Datasets; deep siamese KPCA convolutional mapping network (KPCA-MNet); Feature extraction; Feature maps; Image resolution; Image segmentation; Kernel; kernel principal component analysis (KPCA); Kernels; Machine learning; Principal component analysis; Principal components analysis; Remote sensing; Task analysis; Training; unsupervised multiclass change detection; very-high-resolution (VHR) images
• ISSN: 2168-2267; 2168-2275
• DOI: 10.1109/TCYB.2021.3086884

With the development of Earth observation technology, a very-high-resolution (VHR) image has become an important data source of change detection (CD). These days, deep learning (DL) methods have achieved conspicuous performance in the CD of VHR images. Nonetheless, most of the existing CD models based on DL require annotated training samples. In this article, a novel unsupervised model, called kernel principal component analysis (KPCA) convolution, is proposed for extracting representative features from multitemporal VHR images. Based on the KPCA convolution, an unsupervised deep siamese KPCA convolutional mapping network (KPCA-MNet) is designed for binary and multiclass CD. In the KPCA-MNet, the high-level spatial-spectral feature maps are extracted by a deep siamese network consisting of weight-shared KPCA convolutional layers. Then, the change information in the feature difference map is mapped into a 2-D polar domain. Finally, the CD results are generated by threshold segmentation and clustering algorithms. All procedures of KPCA-MNet do not require labeled data. The theoretical analysis and experimental results in two binary CD datasets and one multiclass CD datasets demonstrate the validity, robustness, and potential of the proposed method.