Computer Science · Computer Vision and Pattern Recognition
Spherical aberration coefficients identification of spherical aberration interferograms based on deep learning algorithm
See on the mapAbstract
In the field of optical measurement,the spherical aberration always causes a circular speckle which is symmetrical to the optic axis on interferograms and it can even blur the images.In this paper, a deep learning algorithm is put forward to identify the spherical aberration coefficients of monochromatic spherical aberration interferograms.The deep learning algorithm needs sufficient training samples, but it also takes a lot of time to collect the real interferograms as the training samples.The method of zernike polynomial fitting wave surface is much better relative to the least square method.Therefore,this paper adopts the zernike polynomials fitting method,by means of changing aberration coefficients,to obtain a large number of training samples of spherical aberration interferograms by computer simulation.Different from other traditional methods, using deep learning algorithms to identify the information of the interferograms is more efficient, the spherical aberration coefficients of interferograms can be accurately and quickly identified.What is more,this method also simplify the manual processing.