Computational imaging: foundation, reality, and prospect

This paper supplies researchers and students with insights into the mathematical foundation and prospect of Computational Imaging, in juxtaposition with the Maximum Accuracy Computing (MAC), a new mathematical theory and method for data prediction (extrapolation, interpolation, approximation, or reconstruction), which yields 40 times more accurate data, compared with the computational imaging method of Convolutional Neural Networks (CNN): mse(CNN) = mse(MAC) x 40.18 (mse: mean squared error), leaping from utterly inapplicable to perceptually identical, equivalent to the speed change from horse wagon to jet engine, i.e., from 15 mph to 600 mph. Demo software, demo video, mathematical rationale, method description, and testing data are all downloadable at http://www.lucidsee.ca