Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)


Electrical and Computer Engineering

Committee Chair

Charles A. Bouman

Committee Co-Chair

Gregery T. Buzzard

Committee Member 1

Mark R. Bell

Committee Member 2

Mary Comer

Committee Member 3

Michael A. Groeber


Sparse sampling schemes can broadly be classified in to two main categories: static sampling where the sampling pattern is predetermined, and dynamic sampling where each new measurement location is selected based on information obtained from previous measurements. Dynamic sampling methods are particularly appropriate for point-wise imaging methods in which pixels are measured sequentially in arbitrary order. Examples of point-wise imaging schemes include certain implementations of atomic force microscopy (AFM), electron back scatter diffraction (EBSD), and Scanning Electron Microscopy (SEM). In these point-wise imaging applications, dynamic sparse sampling methods have the potential to dramatically reduce the number of measurements required to achieve a desired level of fidelity. However, existing dynamic sampling methods tend to be computationally expensive and are therefore too slow for many practical applications. In this dissertation, we present a framework for dynamic sampling based on machine learning techniques, which we call a supervised learning approach for dynamic sampling (SLADS). In each step of SLADS, the objective is to find the pixel that maximizes the expected reduction in distortion (ERD) given previous measurements. SLADS is fast because we use a simple regression function to compute the ERD, and it is accurate because the regression function is trained using data sets that are representative of the specific application. In addition, we introduce an approximate method to terminate dynamic sampling at a desired level of distortion. We then extend our algorithm to incorporate multiple measurements at each step, which we call

group-wise SLADS. We then present simulation results on computationally-generated synthetic data and experimentally-collected data to demonstrate a dramatic improvement over state-of-the-art static sampling methods. 1 Next, we present implementations of SLADS for Scanning Electron Microscopy (SEM), Raman Imaging, Energy Dispersive Spectroscopy (EDS) and synchrotron X-ray imaging. In these imaging techniques, a measurement acquired at a pixel location may be a scalar or a vector, and therefore, to directly apply the SLADS framework we pre-process the measurements to convert them to scalar labels, using various classification techniques. We finally present results from SLADS implementations and simulations for these imaging techniques.