Abstract
Semantic image segmentation is one kindof end-to-end segmentation method which can classify the target region pixel by pixel. As a classic semantic segmentation network in optical images, DeepLabv3+ can achieve a good segmentation performance. However, when this network is directly used in the semantic segmentation of polarimetric synthetic aperture radar (PolSAR) image, it is hard to obtain the ideal segmentation results. The reason is that it is easy to yield overfitting due to the small PolSAR dataset. In this article, a lightweight complex-valued DeepLabv3+ (L-CV-DeepLabv3+) is proposed for semantic segmentation of PolSAR data. It has two significant advantages when compared with the original DeepLabv3+. First, the proposed network with the simplified structure and parameters can be suitable for the small PolSAR data, and thus, it can effectively avoid the overfitting. Second, the proposed complex-valued (CV) network can make full use of both amplitude and phase information of PolSAR data, which brings better segmentation performance than the real-valued (RV) network, and the related CV operations are strictly true in the mathematical sense. Experimental results about two Flevoland datasets and one San Francisco dataset show that the proposed network can obtain better overall average, mean intersection over union, and mean pixel accuracy than the original DeepLabv3+ and some other RV semantic segmentation networks.
Highlights
S YNTHETIC aperture radar (SAR) has all-day and allweather imaging capability, which is very important in military and civilian fields
We propose a lightweight complex-valued DeepLabv3+ (L-CVDeepLabv3+) for semantic segmentation of polarimetric synthetic aperture radar (PolSAR) image in order to obtain better segmentation performance than classic RV segmentation networks based on deep learning
FCN-8s is selected; softmax classifier is used in the multiclassification for U-Net; ResNet50 is used as the backbone network of PSPNet; and Xception is used as the backbone network of DeepLabv3+
Summary
S YNTHETIC aperture radar (SAR) has all-day and allweather imaging capability, which is very important in military and civilian fields. In order to improve the classification accuracy of land cover, some methods based on machine learning were used in PolSAR image classification. Because some superpixel segmentation methods were proved to be effective in preserving spatial structure information of PolSAR images [37]–[39], the superpixel-based graph convolutional network was proposed for PolSAR image classification [40]. Since the original semantic segmentation networks were proposed for optical images, the input data of these networks were real-valued (RV) Both single-polarization SAR and PolSAR data are complex valued (CV). We propose a lightweight complex-valued DeepLabv3+ (L-CVDeepLabv3+) for semantic segmentation of PolSAR image in order to obtain better segmentation performance than classic RV segmentation networks based on deep learning. The loss function is given, and the CV batch normalization and CV weight initialization are introduced
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
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 call
