Complex-Valued Multi-Scale Fully Convolutional Network with Stacked-Dilated Convolution for PolSAR Image Classification

Abstract

Polarimetric synthetic aperture radar (PolSAR) image classification is a pixel-wise issue, which has become increasingly prevalent in recent years. As a variant of the Convolutional Neural Network (CNN), the Fully Convolutional Network (FCN), which is designed for pixel-to-pixel tasks, has obtained enormous success in semantic segmentation. Therefore, effectively using the FCN model combined with polarimetric characteristics for PolSAR image classification is quite promising. This paper proposes a novel FCN model by adopting complex-valued domain stacked-dilated convolution (CV-SDFCN). Firstly, a stacked-dilated convolution layer with different dilation rates is constructed to capture multi-scale features of PolSAR image; meanwhile, the sharing weight is employed to reduce the calculation burden. Unfortunately, the labeled training samples of PolSAR image are usually limited. Then, the encoder–decoder structure of the original FCN is reconstructed with a U-net model. Finally, in view of the significance of the phase information for PolSAR images, the proposed model is trained in the complex-valued domain rather than the real-valued domain. The experiment results show that the classification performance of the proposed method is better than several state-of-the-art PolSAR image classification methods.

Mapping Vegetation Types by Different Fully Convolutional Neural Network Structures with Inadequate Training Labels in Complex Landscape Urban Areas

“…The highest UA improvement was obtained for grassland, with (a) being 6.74%, 9.46%, and 5.05% higher than the UA for (b), (c), and (d), respectively. Therefore, image features extracted via a single model are often inadequate compared to those extracted by multiple models [42], and the study was able to effectively improve the classification accuracy by fusing features with differences.…”

$C^{2}N^{2}$: Complex-Valued Contourlet Neural Network

“…The CV-EPLS [53] extracts nonredundant sparse features of the amplitude and phase information in different polarimetric channels, which is conducive to the subsequent classification. The CV-SDFCN [54] is a multi-scale FCN network based on the U-net structure, which is defined in the complex-valued domain with stacked dilated convolutions. Qin et al [55] introduced superpixel oriented (SPO) into CV-CNN, thus the computational cost is reduced and image details are preserved.…”