Abstract
Hyperspectral image (HSI) classification has become one of the most significant tasks in the field of hyperspectral analysis. However, classifying each pixel in HSI accurately is challenging due to the curse of dimensionality and limited training samples. In this paper, we present an HSI classification architecture called camera spectral response network (CSR-Net), which can learn the optimal camera spectral response (CSR) function for HSI classification problems and effectively reduce the spectral dimensions of HSI. Specifically, we design a convolutional layer to simulate the capturing process of cameras, which learns the optimal CSR function for HSI classification. Then, spectral and spatial features are further extracted by spectral and spatial attention modules. On one hand, the learned CSR can be implemented physically and directly used to capture scenes, which makes the image acquisition process more convenient. On the other hand, compared with ordinary HSIs, we only need images with far fewer bands, without sacrificing the classification precision and avoiding the curse of dimensionality. The experimental results of four popular public hyperspectral datasets show that our method, with only a few image bands, outperforms state-of-the-art HSI classification methods which utilize the full spectral bands of images.
Highlights
Hyperspectral image (HSI) collects as many as hundreds of spectral bands of a scene, and has been widely used in the area of remote sensing
Due to the limited number of labeled training samples, some approaches are largely affected by the curse of dimensionality [6], which may lead to a drop in classification accuracy
We present a novel HSI classification method named the camera spectral response network (CSR-Net), which effectively investigates the optimal camera spectral response (CSR) and makes it possible to reach high classification accuracy with only limited number of image bands
Summary
Hyperspectral image (HSI) collects as many as hundreds of spectral bands of a scene, and has been widely used in the area of remote sensing. Given that the spectral and spatial information in HSI can provide discriminative features in identifying material characteristics, utilizing this information to classify HSIs has become an active topic in the hyperspectral community. HSI classification often identifies the category of the material at each pixel instead of the full image, where the high-dimensional spectral vector is supposed to provide sufficient characteristics and can be distinguished by classifiers. In order to reduce the Hughes effect when classifying HSIs, dimensionality reduction operation is often utilized to simplify the original high-dimensional data. Most HSI classification methods focus on transforming the high-dimensional HSI samples into lower ones, while maintaining the intrinsic and most discriminative features. This kind of dimensionality reduction method can be called feature extraction. The goal of feature extraction is to derive an effective representation of the original HSI in a certain feature space, and reduce redundant information within HSIs
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.
