Abstract
Synthetic aperture radar (SAR) image change detection (CD) aims to automatically recognize changes over the same geographic region by comparing prechange and postchange SAR images. However, the detection performance is usually subject to several restrictions and problems, including the absence of labeled SAR samples, inherent multiplicative speckle noise, and class imbalance. More importantly, for bitemporal SAR images, changed regions tend to present highly variable sizes, irregular shapes, and different textures, typically referred to as hybrid variabilities, further bringing great difficulties to CD. In this paper, we argue that these internal hybrid variabilities can also be used for learning stronger feature representation, and we propose a hybrid variability aware network (HVANet) for completely unsupervised label-free SAR image CD by taking inspiration from recent developments in deep self-supervised learning. First, since different changed regions may exhibit hybrid variabilities, it is necessary to enrich distinguishable information within the input features. To this end, in shallow feature extraction, we generalize the traditional spatial patch (SP) feature to allow for each pixel in bitemporal images to be represented at diverse scales and resolutions, called extended SP (ESP). Second, with the carefully customized ESP features, HVANet performs local spatial structure information extraction and multiscale–multiresolution (MS-MR) information encoding simultaneously through a local spatial stream and a scale-resolution stream, respectively. Intrinsically, HVANet projects the ESP features into a new high-level feature space, where the change identification becomes easier. Third, to train the framework effectively, a self-supervision layer is attached to the top of the HVANet to enable the two-stream feature learning and recognition of changed pixels in the corresponding feature space, in a self-supervised manner. Experimental results on three low/medium-resolution SAR datasets demonstrate the effectiveness and superiority of the proposed framework in unsupervised SAR CD tasks.
Highlights
Benefiting from the capability of all-weather and all-time Earth observation, the synthetic aperture radar (SAR) sensor has been used in numerous applications increasingly, including but not limited to urban planning, disaster monitoring, and land-cover/landuse (LCLU) analysis [1,2,3,4,5,6,7]
We propose to represent each pixel in the images using both the conventional singlescale patch (i.e., spatial patch (SP) feature) and the MS-MR patches simultaneously, which is capable of comprehensively describing pixel information through the complementary local spatial information and long-range context information
The results indicate that the SRS is critical in our hybrid variability aware network (HVANet), especially the scale-resolution information encoding module (SRE) module that can fully exploit MS-MR information and capture the informative scale and resolution clues to strengthen the description of the hybrid variabilities in change detection (CD) scenes
Summary
Benefiting from the capability of all-weather and all-time Earth observation, the synthetic aperture radar (SAR) sensor has been used in numerous applications increasingly, including but not limited to urban planning, disaster monitoring, and land-cover/landuse (LCLU) analysis [1,2,3,4,5,6,7]. SAR images by processing the radar backscatter responses coherently [10,11], speckle noise inevitably appears all over the images, which dramatically affects the intensity of the SAR images; changed pixels share a wide range of intensity values with unchanged pixels, namely intensity fluctuations. This is because speckle causes dramatic intensity fluctuations and further results in the overlapped nature between the changed and unchanged classes [9,10], which brings difficulty for accurate change feature extraction. Pseudochanges can be caused by the slight variation of the acquisition parameters, such as the imaging configuration, incidence angles, and radiometric variations, making it difficult to detect changes of interest precisely [14,15]
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
