A 3-stage Spectral-spatial Method for Hyperspectral Image Classification

Abstract

Hyperspectral images often have hundreds of spectral bands of different wavelengths captured by aircraft or satellites that record land coverage. Identifying detailed classes of pixels becomes feasible due to the enhancement in spectral and spatial resolution of hyperspectral images. In this work, we propose a novel framework that utilizes both spatial and spectral information for classifying pixels in hyperspectral images. The method consists of three stages. In the first stage, the pre-processing stage, Nested Sliding Window algorithm is used to reconstruct the original data by enhancing the consistency of neighboring pixels and then Principal Component Analysis is used to reduce the dimension of data. In the second stage, Support Vector Machines are trained to estimate the pixel-wise probability map of each class using the spectral information from the images. Finally, a smoothed total variation model is applied to smooth the class probability vectors by ensuring spatial connectivity in the images. We demonstrate the superiority of our method against three state-of-the-art algorithms on six benchmark hyperspectral data sets with 10 to 50 training labels for each class. The results show that our method gives the overall best performance in accuracy. Especially, our gain in accuracy increases when the number of labeled pixels decreases and therefore our method is more advantageous to be applied to problems with small training set. Hence it is of great practical significance since expert annotations are often expensive and difficult to collect.

Unsupervised Spatial-spectral Hyperspectral Image Reconstruction and Clustering with Diffusion Geometry

“…This section reviews HSI reconstruction, which efficiently denoises hyperspectral data by reconstructing HSI pixels using the spectra of spatial nearest neighbors. HSI reconstruction has been successfully used as a preprocessing step for semi-supervised learning [4], [6] and is expected to be useful for unsupervised learning [3].…”

“…A pixel is considered a spatial neighbor of x if it is contained in a spatial window centered at x in the original image. While simple spatial squares have been successfully used as spatial windows in unsupervised and semi-supervised algorithms, the spatial radius generally requires tuning in practice [4], [6]- [9]. In contrast, shape-adaptive (SA) regions may be used for parameter-free HSI reconstruction [4].…”

“…Real-world HSIs often encode strong spatial regularity; i.e., pixels that are spatially close are more likely to contain the same materials. As such, incorporating the rich spatial structure contained in HSIs often leads to substantially higher performance among algorithms for HSI clustering and segmentation [4], [6]- [9]. This section introduces the DSIRC clustering algorithm, which explicitly incorporates HSI reconstruction into D-VIC.…”

Semi-supervised Change Detection of Small Water Bodies Using RGB and Multispectral Images in Peruvian Rainforests

“…Multi-stage SVM-based methods, such as [14], [19], [20], make use of spatial and spectral information in different phases separately. The rich spectral information is used to discriminate classes, and the spatial information is utilized to reduce the noise in images.…”