Self-adaptive optical correction method and system based on convolutional neural network

A convolutional neural network and adaptive optics technology, applied in the field of optics, can solve the problems of difficult application of adaptive optics system and slow convergence speed, and achieve the effect of low cost and high system bandwidth

Active Publication Date: 2018-12-18
ANHUI AGRICULTURAL UNIVERSITY
View PDF5 Cites 25 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to propose a structure for the problems of complex structure, high cost and difficult application of conventional adaptive optics systems in some occasions in the prior art, as well as the slow convergence speed of existing adaptive optics sy

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Self-adaptive optical correction method and system based on convolutional neural network
  • Self-adaptive optical correction method and system based on convolutional neural network
  • Self-adaptive optical correction method and system based on convolutional neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0038] see figure 1, in an embodiment of the present invention, an adaptive optics correction system based on a convolutional neural network, including a wavefront corrector 1, a wavefront corrector drive circuit 2, a main control computer 3, an optical beam splitter 4 and an imaging lens 5 , the function of the deformable mirror 8 here is to reflect the distorted wavefront to be corrected, which actually belongs to a part of the wavefront corrector 1, and an imaging lens 5 is arranged at one side of the wavefront corrector 1, and the The distorted wavefront to be corrected reflected by the front corrector 1 is incident into the imaging lens 5, and one side of the imaging lens 5 is provided with a focal plane CCD6 and a defocused plane CCD7. At the focus position, the distorted wavefront to be corrected by the imaging lens 5 is divided into two paths by the beam splitter 4, one of which is focused on the focal plane CCD6 by the focusing lens, and the other is focused by the fo...

Embodiment 2

[0040] see Figure 2-3 , Embodiment 1 has described the adaptive optics correction system based on convolutional neural network in detail. This embodiment mainly describes the adaptive optics correction method based on convolutional neural network. Specifically, the method includes the following steps :

[0041] S1, based on the convolutional neural network to train the convolutional neural network model driven by the distorted far-field light intensity and the wavefront corrector, use the convolutional neural network model to directly obtain the driving signal of the wavefront corrector according to the input light intensity, and use the drive The signal controls the wavefront corrector to generate a deformation conjugate to the wavefront to be corrected to correct the aberration of the incident wavefront;

[0042] Specifically, the construction method of the convolutional neural network model includes the following steps:

[0043] (1) Using the Zernike coefficient {a 1 ,a...

Embodiment 3

[0076] On the basis of Embodiment 2, this embodiment also adopts the stochastic gradient descent algorithm in the training of the convolutional neural network model, and learns the parameters of the established convolutional neural network by continuously reducing the function value of the loss function, specifically of:

[0077] The loss function is expressed as:

[0078]

[0079] In the formula, y i For the real Zernike coefficient, y_predicted i is the predicted Zernike coefficient.

[0080] The stochastic gradient descent method used in the training of the convolutional neural network model refers to the use of the samples in each iteration to learn parameters and updates, and the learning parameters and updates of each generation can be expressed as:

[0081] V i+1 =μV t -α▽loss(W i );

[0082] W t+1 =W t +V t+1 ;

[0083] In the formula, t is the number of iterations, W t is the parameter at time t, V t is the increment at time t, α is the learning rate, μ ...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses a self-adaptive optical correction method based on a convolutional neural network. The method comprises the following steps of S1, based on the convolutional neural network, training a convolutional neural network model of a distorted far-field light intensity image and a wavefront corrector driving signal; and S2, after the convolutional neural network model is built, dividing a to-be-corrected distorted wavefront into two parts after reflection by a wavefront corrector and beam splitting of a light path, and performing imaging on a focus plane CCD and an out-of-focusplane CCD. The method has the beneficial effects that the wavefront corrector driving signal is obtained directly according to the input light intensity by utilizing the convolutional neural network model; the driving signal is used for controlling the wavefront corrector to generate a deformation amount which is conjugated with the to-be-corrected wavefront so as to correct an aberration of an incident wavefront; the wavefront detection and the corresponding reconstruction calculation do not need to be carried out, and the iteration optimization does not need to be carried out; and a system is simple in structure, easy to implement, low in cost and wide in bandwidth.

Description

technical field [0001] The invention relates to the field of optical technology, in particular to an adaptive optical correction method and system based on a convolutional neural network. Background technique [0002] When light is transmitted in the atmosphere, it is inevitably affected by the atmospheric turbulence effect. The turbulence effect will cause the deterioration of the phase coherence of the light field, resulting in fluctuations in light intensity, spot drift, expansion and fragmentation. These effects have varying degrees of influence on different laser engineering applications. For example, the degradation of the phase coherence of the light field will reduce the image resolution of the imaging system; the fluctuation of light intensity and the drift of the light spot will cause the increase of the bit error rate of the laser communication system; the expansion and fragmentation of the light spot will lead to the decrease of the laser transmission energy conc...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
IPC IPC(8): G02B26/06G02B27/00G06N3/04G06N3/08
CPCG02B26/06G02B27/0025G02B27/0068G06N3/08G06N3/045
Inventor 马慧敏张武
Owner ANHUI AGRICULTURAL UNIVERSITY
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap