Heterogeneous Images Change Detection Based on Iterative Joint Global–Local Translation

Most heterogeneous change detection methods based on transfer learning may not yield satisfactory results due to the lack of comprehensive utilization global and local characteristics image. In this paper, we propose an unsupervised method iterative joint global-local translation (IJGLT). The two images are firstly segmented into superpixels with same boundary. Unlike for classification tasks that directly reduce distance distribution all data, construct a function decrease difference between unchanged while increasing changed superpixels. During process, overall is constrained using pseudo labels, manifold preserving utilized maintain spatial structure adjacency relationship By comparing translated images, map obtained then refined labels taking approach until convergence. Using public self-made optical-SAR datasets different resolutions, experimental demonstrate IJGLT superior several typical comparison maximum accuracy 0.98 kappa coefficient 0.9.