SAR Image Change Detection Based on Heterogeneous Graph With Multiattributes and Multirelationships

Abstract

The performance of change detection between synthetic aperture radar (SAR) images mainly depends on the selection and utilization of image attributes. Nevertheless, most existing change detection approaches merely take the intensity attribute into consideration, constraining their capacities of detecting changes in complex situations. To solve this problem, this study develops an unsupervised SAR image change detection approach based on heterogeneous graph with multi-attributes and multi-relationships. First, the structural attribute, obtained by removing the complex texture, can be used to reflect the overall features of a SAR image. A heterogeneous graph is then constructed to encode the structural and intensity attributes as vertices, and two types of edges are hence connected to capture the intra-relationships and inter-relationships from these different attributes. With the support of this graph, the hyper-adjacency characteristics are proposed to quantify the multi-relationships and describe the global heterogeneous information. Constructing heterogeneous graphs respectively on bi-temporal SAR images, the changes of both intensity and structural attributes can be measured via comparing the hyper-adjacency characteristics of the bi-temporal graphs, thereby generating a difference image with good separability. Finally, the change map is obtained by cutting off the weakly related edges of heterogeneous graph on the difference image. Experiments on four real SAR datasets prove the effectiveness of the proposed graph-driven approach in improving the accuracy and robustness of change detection.

SAR Image Change Detection via Heterogeneous Graph With Multiorder and Multilevel Connections

“…Heterogeneous graph is a generalization of ordinary graph that allows for the definition of various types of vertices or edges [34][35][36]. This makes the heterogeneous graph ideal for representing a wide variety of relationships between vertices.…”

Nonlocal feature learning based on a variational graph auto-encoder network for small area change detection using SAR imagery

“…Thus, there were four methods for superpixel classification using the KM algorithm: SLIC+PCA+KM, SLIC+AE+KM, SLIC+SAE+KM and SCAE. Three stateof-the-art evaluation methods, applied to the benchmark datasets, were utilized to improve the comparison of the four methods, including two pixel-based approaches: nonlocal low-rank PCA and two-level clustering (NLR-PCATLC) (Sun et al, 2020), fuzzy local information c-means based on multiple features (MFFLICM) (Meng et al, 2020), and one object-based method, heterogeneous graph (HG) (Wang et al, 2022). The proposed Nonlocal Learning-Based Small Area Change Detection (NLBSACD) framework adopted the forementioned two methods (Eq.…”