This is the Algorithm for the Paper: "Using Machine-Learning to Optimize phase contrast in a Low-Cost Cellphone Microscope" implemented in TENSORFLOW.
This produces a dataset of object/optimized illumination source patterns pairs for the learning approach. The Algorithm were tested (tf_illopt_v4_Matrixsolution_GPU_wo_memoryproblem.py) using a Nvidia GTX Titan using TF 1.3 on Anaconda
Beamerscope_CNN: This learns the implicit model from the relationship between the phase-object and its optimized illumination source which will enhance the phase contrast. The Algorithm were tested (input_data_v1.py.py) using a Nvidia GTX Titan using TF 1.3 on Anaconda