This thesis addresses two important aspects in hyperspectral image processing: automatic hyperspectral image denoising and unmixing. The first part of this thesis is devoted to a novel automatic optimized vector bilateral filter denoising algorithm, while the remainder concerns nonnegative matrix factorization with deterministic annealing for unsupervised unmixing in remote sensing hyperspectral images. The need for automatic hyperspectral image processing has been promoted by the development of potent hyperspectral systems, with hundreds of narrow contiguous bands, spanning the visible to the long wave infrared range of the electromagnetic spectrum. Due to the large volume of raw data generated by such sensors, automatic processing in the hyperspectral images processing chain is preferred to minimize human workload and achieve optimal result. Two of the mostly researched processing for such automatic effort are: hyperspectral image denoising, which is an important preprocessing step for almost all remote sensing tasks, and unsupervised unmixing, which decomposes the pixel spectra into a collection of endmember spectral signatures and their corresponding abundance fractions. Two new methodologies are introduced in this thesis to tackle the automatic processing problems described above.
Vector bilateral filtering has been shown to provide good tradeoff between noise removal and edge degradation when applied to multispectral/hyperspectral image denoising. It has also been demonstrated to provide dynamic range enhancement of bands that have impaired signal to noise ratios. Typical vector bilateral filtering usage does not employ parameters that have been determined to satisfy optimality criteria. This thesis also introduces an approach for selection of the parameters of a vector bilateral filter through an optimization procedure rather than by ad hoc means. The approach is based on posing the filtering problem as one of nonlinear estimation and minimizing the Stein's unbiased risk estimate (SURE) of this nonlinear estimator. Along the way, this thesis provides a plausibility argument with an analytical example as to why vector bilateral filtering outperforms band-wise 2D bilateral filtering in enhancing SNR. Experimental results show that the optimized vector bilateral filter provides improved denoising performance on multispectral images when compared to several other approaches.
Non-negative matrix factorization (NMF) technique and its extensions were developed to find part based, linear representations of non-negative multivariate data. They have been shown to provide more interpretable results with realistic non-negative constrain in unsupervised learning applications such as hyperspectral imagery unmixing, image feature extraction, and data mining. This thesis extends the NMF method by incorporating deterministic annealing optimization procedure, which will help solve the non-convexity problem in NMF and provide a better choice of sparseness constrain. The approach is based on replacing the difficult non-convex optimization problem of NMF with an easier one by adding an auxiliary convex entropy constrain term and solving this first. Experiment results with hyperspectral unmixing application show that the proposed technique provides improved unmixing performance compared to other state-of-the-art methods.
Library of Congress Subject Headings
Multispectral imaging--Data processing; Remote sensing--Data processing; Signal processing--Digital techniques; Source separation (Signal processing)
Imaging Science (Ph.D.)
Department, Program, or Center
Chester F. Carlson Center for Imaging Science (COS)
Sohail A. Dianat
Peng, Honghong, "Automatic Denoising and Unmixing in Hyperspectral Image Processing" (2014). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus