A Compressive Sensing and Unmixing Scheme for Hyperspectral Data Processing

• Published in: IEEE transactions on image processing Vol. 21; no. 3; pp. 1200 - 1210
• Main Authors: Chengbo Li, Ting Sun, Kelly, K. F., Yin Zhang
• Format: Journal Article
• Language: English
• Published: New York, NY IEEE 01.03.2012; Institute of Electrical and Electronics Engineers; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Abundance; Algorithms; Applied sciences; Augmented Lagrangian method; Coding, codes; Compressed; Compressed sensing; compressive sensing (CS); Computation; Data processing; data unmixing; Detection; Exact sciences and technology; fast Walsh-Hadamard transform; Hyperspectral imaging; Image coding; Image processing; Information theory; Information, signal and communications theory; Mathematical model; Mathematical models; Numerical models; Reconstruction; Sampling, quantization; Signal and communications theory; Signal processing; Studies; Telecommunications and information theory; total variation (TV); Performance evaluation; Augmented Lagrangian method; data unmixing; Hyperspectral imagery; Data compression; Information rate; Data processing; Algorithm; Information transmission; Hyperspectral imaging sensor; fast Walsh-Hadamard transform; Television; total variation (TV); compressive sensing (CS); Hadamard transformation; Feasibility; Real time processing; Lagrangian method; hyperspectral imaging; Compressed sensing
• ISSN: 1057-7149; 1941-0042; 1941-0042
• DOI: 10.1109/TIP.2011.2167626

Hyperspectral data processing typically demands enormous computational resources in terms of storage, computation, and input/output throughputs, particularly when real-time processing is desired. In this paper, a proof-of-concept study is conducted on compressive sensing (CS) and unmixing for hyperspectral imaging. Specifically, we investigate a low-complexity scheme for hyperspectral data compression and reconstruction. In this scheme, compressed hyperspectral data are acquired directly by a device similar to the single-pixel camera based on the principle of CS. To decode the compressed data, we propose a numerical procedure to compute directly the unmixed abundance fractions of given end members, completely bypassing high-complexity tasks involving the hyperspectral data cube itself. The reconstruction model is to minimize the total variation of the abundance fractions subject to a preprocessed fidelity equation with a significantly reduced size and other side constraints. An augmented Lagrangian-type algorithm is developed to solve this model. We conduct extensive numerical experiments to demonstrate the feasibility and efficiency of the proposed approach, using both synthetic data and hardware-measured data. Experimental and computational evidences obtained from this paper indicate that the proposed scheme has a high potential in real-world applications.