Abstract
Convolution-based autoencoder networks have yielded promising performances in exploiting spatial–contextual signatures for spectral unmixing. However, the extracted spectral and spatial features of some networks are aggregated, which makes it difficult to balance their effects on unmixing results. In this paper, we propose two gated autoencoder networks with the intention of adaptively controlling the contribution of spectral and spatial features in unmixing process. Gating mechanism is adopted in the networks to filter and regularize spatial features to construct an unmixing algorithm based on spectral information and supplemented by spatial information. In addition, abundance sparsity regularization and gating regularization are introduced to ensure the appropriate implementation. Experimental results validate the superiority of the proposed method to the state-of-the-art techniques in both synthetic and real-world scenes. This study confirms the effectiveness of gating mechanism in improving the accuracy and efficiency of utilizing spatial signatures for spectral unmixing.
Highlights
Hyperspectral images (HSIs) often contain affluent spectral information for the high spectral resolution
Considering the balance problem of spectral and spatial information, we introduce the gating mechanism to adaptively mitigate the negative effects of uncorrelated spatial distribution on unmixing results
Since the gating mechanism is of great importance to the unmixing performance, we investigate the effects through changing the penalty parameter of gating mechanism
Summary
Hyperspectral images (HSIs) often contain affluent spectral information for the high spectral resolution. This method avoids imposing abundance correlation regularization to areas of alternating materials, which is an improvement over TV He et al [17] continued to research on addressing spatial piecewise smooth structure and proposed a method based on sparse unmixing and TV. Spatial information was not incorporated in the network, making it hard to further enhance the unmixing results To this end, a fully convolutional encoder network was proposed in [28], where two-dimensional convolutional layers were adopted to exploit local spatial correlation. We propose a gated three-dimensional convolutional autoencoder network to extract spectral and spatial features simultaneously. We propose a gated dual branch autoencoder network to improve the exploitation efficiency of spatial and spectral information, respectively.
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
