Design of Optimal Illumination Patterns in Single-Pixel Imaging Using Image Dictionaries

Shuming Jiao

Abstract: A single-pixel imaging (SPI) system is often applied to a specific scenario. The target object images may belong to a specific category and have some common features. Fourier or Hadamard illumination patterns can be sub-optimal since they are fixed and non-adaptive. We propose a novel scheme for designing and optimizing the illumination patterns adaptively from an image dictionary by extracting the common features using principal component analysis (PCA), with enhanced imaging efficiency.