Underwater Microbial Detection Based On Mach-zehnder Interference Holography

Digital holography is an optical imaging technique that can capture the diffraction front of an object by recording an interference pattern with an electronic sensor.Since the hologram retains the entire wavefront,the amplitude and phase information can be reconstructed using a reconstruction algorithm.For reconstruction,we need to know the depth position of the recorded object in advance to reconstruct the image at the focused position.We need to do super resolution reconstruction to enlarge the resolution of the object and improve the clarity of the object.In this paper,we propose a deep neural neural network for predicting focal distance based on convolutional neural network(CNN).The neural network can obtain depth position information accurate to millimeters directly from the hologram,and saving a lot of time.In this method,CNN can directly obtain the depth position information from the hologram,which is accurate to millimeter,saving a lot of time.We also proposed a super-resolution neural network based on holography,which can save the time of super-resolution reconstruction,and the resulting image has a better effect than the image obtained by using the traditional super-resolution reconstruction method.The specific research work is as follows:1.We built the off-axis holographic device,and collected microbial images of paramecium.And we use the convolution reconstruction algorithm to accomplish the reconstruction of the hologram.2.We studied the self-focusing algorithm and simulated a variety of traditional self-focusing algorithms.Also,we studied the traditional super resolution reconstruction algorithm and accomplished the experimental simulation.3.We studied the focus prediction method based on neural network.We proposed an improved neural network architecture to predict the focus distance based on the convolutional neural network,which finally reached the accuracy of millimeter level.4.We studied the super-resolution reconstruction algorithm based on neural network,and built an improved super-resolution reconstruction architecture based on neural network,so as to realize the super-resolution reconstruction of images.