PROCEEDINGS ARTICLE | July 24, 2018
Proc. SPIE. 10827, Sixth International Conference on Optical and Photonic Engineering (icOPEN 2018)
KEYWORDS: Monochromatic aberrations, Fringe analysis, Detection and tracking algorithms, Wavefronts, Zernike polynomials, Neural networks, Convolution, Algorithm development, Statistical modeling, Evolutionary algorithms
Deep learning is an extension of machine learning,deep learning uses multilayer neural networks to analyze data[1,2].Convolutional Neural Networks (CNN) is a commonly used neural network structure in deep learning. It is widely used in various fields, especially in the field of machine vision. In the field of optics,Monochromatic aberrations include spherical aberration, coma, astigmatism, defocus and so on, the common way to interpret interferograms is the Zernike polynomials, it generally used to describe the wavefront characteristics.In this paper, the convolutional neural network algorithm is used to identify astigmatism and defocus of the typical monochromatic interferograms, Zernike polynomial is multiplied by the aberration coefficient to represent the wavefront, the wavefront into the light intensity formula to obtain the aberration interferogram, the use of the above method to obtain the defocus interferogram, the result shows that the recognition accuracy is very high.The method of deep learning algorithm used to identify monochrome interferograms is simple and fast, and the training samples do not need manual calibration.
