3D Wavelet Convolutions with Extended Receptive Fields for Hyperspectral Image Classification

TL;DR

This work tackles hyperspectral image classification under challenges of high dimensionality, sparse ground truth, and spectral redundancy. It introduces WCNet, a 3D-DenseNet-based architecture that embeds Wavelet Conv to extend receptive fields through cascaded Haar wavelet transforms, focusing on low-frequency components with minimal parameter growth. Two design choices—exponentially increasing growth rate and fully dense connectivity—improve feature reuse and efficiency. Across Indian Pines, Pavia University, and KSC, WCNet achieves state-of-the-art or competitive performance with lower parameter counts and reduced redundancy, demonstrating the practical value of wavelet-based extended receptive fields for joint spatial-spectral hyperspectral analysis.

Abstract

Deep neural networks face numerous challenges in hyperspectral image classification, including high-dimensional data, sparse ground object distributions, and spectral redundancy, which often lead to classification overfitting and limited generalization capability. To better adapt to ground object distributions while expanding receptive fields without introducing excessive parameters and skipping redundant information, this paper proposes WCNet, an improved 3D-DenseNet model integrated with wavelet transforms. We introduce wavelet transforms to effectively extend convolutional receptive fields and guide CNNs to better respond to low frequencies through cascading, termed wavelet convolution. Each convolution focuses on different frequency bands of the input signal with gradually increasing effective ranges. This process enables greater emphasis on low-frequency components while adding only a small number of trainable parameters. This dynamic approach allows the model to flexibly focus on critical spatial structures when processing different regions, rather than relying on fixed receptive fields of single static kernels. The Wavelet Conv module enhances model representation capability by expanding receptive fields through 3D wavelet transforms without increasing network depth or width. Experimental results demonstrate superior performance on the IN, UP, and KSC datasets, outperforming mainstream hyperspectral image classification methods.

Figures (9)

• Figure 1: Performing convolution in the wavelet domain results in a larger receptive field
• Figure 2: The workflow of WTCONV operation
• Figure 3: Wavelet convolution in 3D-DenseNet's dense block
• Figure 4: The proposed DenseNet variant. It differs from the original DenseNet in two aspects: (1) layers with different resolution feature maps are also directly connected; (2) the growth rate doubles whenever the feature map size shrinks (features generated in the third yellow dense block are much more numerous than those generated in the first block).
• Figure 5: The network structure of our WCNet, incorporating the basic architecture of 3D-DenseNet.