Compressed Sensing (CS) Imaging with Wide FOV and Dynamic Magnification
Abstract
In this project we developed imaging systems based on the principles of compressive sensing. We developed advanced signal processing algorithms as well as a hardware testbed. The key idea in CS reconstruction is the realization that most signals encountered in practice are sparse in some domain and the theory of CS exploits such sparsity to dictate that far fewer sampling resources than traditional approaches are needed. Spectral image cubes are particularly well suited for sparse representation as images across different wavelengths exhibit strong correlation. We have thus developed a testbed to acquire and reconstruct natural images from a limited number of linear projection measurements at sub-Nyquist sampling rates. A key to the success of CS is the design of the measurement ensemble, which is based on the evaluation of the incoherence between the measurement ensemble and the sparsity basis. Due to the large scale nature of images, the generation of the measurement ensemble should be both computationally efficient and memory efficient. Based on these principles, we developed compressive cross-sectional imaging (confocal) systems, a single pixel compressive camera, and a multispectral imaging system. Advanced algorithms for the efficient reconstruction of the compressive measurements were developed.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 14, 2011
- Accession Number
- ADA538604
Entities
People
- Dennis W. Prather
- Gonzalo R. Arce
Organizations
- University of Delaware