Microscopy image reconstruction method based on convolution network feature fusion

In order to solve the problem of Fourier ptychographic microscopy imaging reconstruction, this paper proposes a new convolutional network feature fusion Fourier stacked microscopy reconstruction method to achieve image reconstruction from low resolution to high resolution. Use the open-source deep learning framework TensorFlow to build the network model and continuously adjust the hyperparameters of the network model. In order to train and test the network model, the FPM imaging model is used to construct a data set for network training and testing. At the same time, it is also tested on traditional methods and neural network-based reconstruction methods to evaluate network performance. The experimental results show that the model has better reconstruction effect and better robustness to noisy images.