Multiresolution Compressive Feature Fusion for Spectral Image Classification

Abstract

In the compressive spectral imaging (CSI) framework, different architectures have been proposed to recover high-resolution spectral images from compressive measurements. Since CSI architectures compactly capture the relevant information of the spectral image, various methods that extract classification features from compressive samples have been recently proposed. However, these techniques require a feature extraction procedure that reorders measurements using the information embedded in the coded aperture patterns. In this paper, a method that fuses features directly from multiresolution compressive measurements is proposed for spectral image classification. More precisely, the fusion method is formulated as an inverse problem that estimates high-spatial-resolution and low-dimensional feature bands from compressive measurements. To this end, the decimation matrices that describe the compressive measurements as degraded versions of the fused features are mathematically modeled using the information embedded in the coded aperture patterns. Furthermore, we include both a sparsity-promoting and a total-variation (TV) regularization terms to the fusion problem in order to consider the correlations between neighbor pixels, and therefore, improve the accuracy of pixel-based classifiers. To solve the fusion problem, we describe an algorithm based on the accelerated variant of the alternating direction method of multipliers (accelerated-ADMM). Additionally, a classification approach that includes the developed fusion method and a multilayer neural network is introduced. Finally, the proposed approach is evaluated on three remote sensing spectral images and a set of compressive measurements captured in the laboratory. Extensive simulations show that the proposed classification approach outperforms other approaches under various performance metrics.

Using Super-Resolution Algorithms for Small Satellite Imagery: A Systematic Review

“…Another solution for assessing the correctness of the method is the correlation coefficient (CC) [186], which determines the correlation measure between the PAN and multispectral (MS) images determined according to (26) shown at the bottom of this page. Additionally, modification of the spatial correlation coefficient (SCC) metric [187] can be used, (26) shown at the bottom of this page where PAN represents the panchromatic image, MS the multispectral image, MS the mean value of the MS images, PAN the mean value of the PAN images, and n,m the image dimension [184].…”

Adaptive Multisensor Acquisition via Spatial Contextual Information for Compressive Spectral Image Classification

“…6 displays the labeling maps obtained by the proposed adaptive method from 3D-CASSI samples for different compression ratios ρ = {0.085, 0.125, 0.160, 0.250}. For comparative purposes, the classification maps yielded by the non-adaptive approach [22] are also shown. We use the support vector machine (SVM) method as a supervised classifier with the polynomial kernel of degree d = 3.…”

“…Recently, two spectral image classification approaches from multi-sensor compressive measurements have been proposed in [22], [23]. However, these approaches do not include an acquisition procedure that considers the spatial contextual information in spectral images to prevent the salt and pepper noise in classification maps.…”

Adaptive Multisensor Acquisition via Spatial Contextual Information for Compressive Spectral Image Classification

“…However, these methods have been designed to fuse multi-sensor data captured by conventional acquisition systems, challenging the storage and computing capabilities of processing systems. Various land cover classification approaches from multi-sensor compressive measurements have been recently proposed in [30]- [32]. These approaches obtain the target fused features by solving regularized optimization problems and the labeling maps are obtained from fused features by using pixel-based classifiers.…”