Abstract
For hyperspectral image (HSI) classification, it is very important to learn effective features for the discrimination purpose. Meanwhile, the ability to combine spectral and spatial information together in a deep level is also important for feature learning. In this letter, we propose an unsupervised feature learning method for HSI classification, which is based on recursive autoencoders (RAE) network. RAE utilizes the spatial and spectral information and produces high-level features from the original data. It learns features from the neighborhood of the investigated pixel to represent the whole local homogeneous area of the image. In addition, to obtain more accurate representation of the investigated pixel, a weighting scheme is adopted based on the neighboring pixels, where the weights are determined by the spectral similarity between the neighboring pixels and the investigated pixel. The effectiveness of our method is evaluated by the experiments on two hyperspectral data sets, and the results show that our proposed method has a better performance.
Highlights
With the developments of remote sensing technologies, hyperspectral images (HSI) captured by hyperspectral imaging sensors have been successfully used to detect and classify objects
We propose an unsupervised feature learning method for HSI classification, which is based on recursive autoencoders (RAE) network
The contributions of the letter are two-fold: 1) a new unsupervised feature learning method based on recursive autoencoders is proposed for HSI classification; 2) a similarity weight between the investigated pixel and its neighboring pixels is considered in the RAE network to learn deep features
Summary
With the developments of remote sensing technologies, hyperspectral images (HSI) captured by hyperspectral imaging sensors have been successfully used to detect and classify objects. Spatial-spectral feature learning methods have found their applications in HSI analysis, such as 3D Gabor filter [10], 3D gray-level co-occurrence [11] and composite kernel SVM [12]. Inspired by semi-supervised RAE, we here propose an unsupervised RAE network with spectral-spatial learning for HSI classification. The learned features are classified using support vector machine (SVM) which has achieved good performance on hyperspectral classification [27]. The contributions of the letter are two-fold: 1) a new unsupervised feature learning method based on recursive autoencoders is proposed for HSI classification; 2) a similarity weight between the investigated pixel and its neighboring pixels is considered in the RAE network to learn deep features
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
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 callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
