作者: Xin Yuan
DOI: 10.1117/1.OE.55.12.123110
关键词: Artificial intelligence 、 Computer vision 、 High-dynamic-range imaging 、 Pixel 、 Demosaicing 、 Compressed sensing 、 Reconstruction algorithm 、 Computer science 、 Shrinkage 、 Shrinkage estimator 、 Detector 、 Image compression
摘要: We apply the Bayesian shrinkage dictionary learning into compressive dynamic-range imaging. By attenuating luminous intensity impinging upon detector at pixel level, we demonstrate a conceptual design of an 8-bit camera to sample high-dynamic-range scenes with single snapshot. Coding strategies for both monochrome and color cameras are proposed. A reconstruction algorithm is developed learn in situ on sampled image, joint demosaicking. use global-local priors coefficients representing data. Simulation results feasibility proposed superior performance algorithm.