Deep learning assisted far-field multi-beam pointing measurement

In this work, we present a deep learning approach to synchronously measure the multi-beam pointing error. This approach uses only one detector to identify the pointing change of the far-field spot by the deep convolutional neural network algorithm. It can be well applied to multi-beam coherent combination for high-power laser systems.Figure1.tif describes the simulated two-beam far-field interference pattern. Figure2.tif describes the training and measurement process of DCNN. Figure3.tif describes the experimental sample acquisition setup. Figure4.tif describes the experimental far-field interference pattern and the experimental results.In addition, DCNN train.py is the training file, Experimental_model.h5 is the trained model, inception_v3_weights_tf_dim_ordering_tf_kernels_notop.h5 is the transfer learning model, Test_train, Test_label, and Test_error are the test set results, label values, and errors of the experiment.