A Dual-Tree Complex Wavelet Transform Based Complex-Valued Convolutional Neural Network for PolSAR Image Classification

In this letter, we propose an active complex-valued convolutional-wavelet neural network by incorporating with dual-tree complex wavelet transform (DT-CWT) and Markov random field (MRF). The DT-CWT constrained pooling layer in CNN and CV-CNN can remove the speckle noise and have a better performance in PolSAR image classification. Experimental results on three benchmark PolSAR datasets demonstrate that the proposed method can achieve a significant classification performance gain in terms of the effectiveness and robustness beyond some state-of-the-art deep learning methods.