CN-116699838-B - Telescope aberration correction method based on point spread function

CN116699838BCN 116699838 BCN116699838 BCN 116699838BCN-116699838-B

Abstract

The invention discloses a telescope aberration correction method based on a point spread function. The method comprises the steps of establishing a model and actually using the model, firstly establishing a telescope system, randomly giving respective degree-of-freedom offset of a lens relative to an ideal position, detecting a Zernike polynomial coefficient and a point spread function of the system at the moment, respectively utilizing a deep learning algorithm to establish the two models between the point spread function and the Zernike polynomial coefficient and between the Zernike polynomial coefficient and the degree-of-freedom offset of the lens, and directly inputting a facula image acquired by a CCD camera into a new model for superposing weight parameters of the two models to solve the offset of the lens without using a wavefront sensing device in the actual use process, thereby correcting low-order aberration generated by the system due to the lens position offset. The method is suitable for solving the lens offset of various complex systems, has high precision and certain real-time performance, and has engineering application value.

Inventors

HUANG YONGMEI
TANG WEI
Tian Siheng
WU QIONGYAN
HE DONG
WANG QIANG
Yuan Liangzhu
TU QIONG

Assignees

中国科学院光电技术研究所

Dates

Publication Date: 20260512
Application Date: 20230619

Claims (7)

1. A method for correcting telescope aberration based on a point spread function, the method comprising the steps of: step one, constructing a telescope system: the system comprises a main mirror, a secondary mirror, an actuating mechanism six-degree-of-freedom displacement platform for controlling the secondary mirror, a CCD camera and an interferometer; Constructing a model between a point spread function and a Zernike polynomial coefficient, wherein the modeling adopts a convolutional neural network algorithm, and firstly, constructing a neural network data set, wherein the data set comprises a training set and a testing set, and the method comprises the following steps of: Recording the ideal space position of a secondary mirror and the point spread function under the state when the telescope system is in the completion of adjustment, adding known degree of disorder to the secondary mirror, acquiring the point spread function under the state by using a CCD camera, acquiring Zernike polynomial coefficients representing system aberration by using an interferometer, and taking the point spread function and the Zernike polynomial coefficients at the moment as a group of samples; step 2.2, repeating the step 2.1 until N groups of samples are obtained, namely obtaining a complete data set containing a training set and a testing set; Constructing a mathematical model between the degree offset of each secondary mirror and the zernike polynomial coefficient, wherein the modeling adopts a method of using a fully-connected neural network algorithm, firstly constructing a neural network data set, wherein the data set comprises a training set and a testing set, and the method comprises the following steps: step 3.1, taking the zernike polynomial coefficient and the secondary mirror degree offset of each in the step 2.2 as a group of samples; Step 3.2, obtaining a complete data set containing a training set and a testing set according to the step 2.2 until N groups of samples are obtained; training a neural network model: Step 4.1, firstly selecting a convolutional neural network model, wherein the neural network comprises an input layer, a hidden layer and an output layer, the input is a point spread function of a secondary mirror under the condition that each degree of freedom is offset, and the output is a Zernike polynomial coefficient corresponding to a system; step 4.2, randomly selecting a part of the data set as a training set to train the two neural networks according to a certain proportion, and when the networks are converged, completing the whole training process, and verifying the fitting capacity of the neural networks by using the rest part of the data set as a test set; Step five, solving the offset of the secondary mirror and realizing the low-order aberration correction of the telescope system, comprising the following steps: Step 5.1, superposing the two trained neural network model parameters to construct a new model, firstly, acquiring a point spread function of a system by using a CCD camera in the low-order aberration correction process of an actual telescope system, inputting the point spread function into the new neural network model, then outputting the respective degree of freedom offset of a secondary mirror, and inputting the opposite number of the data into an actuating mechanism six-degree-of-freedom platform for controlling the space position of the secondary mirror to correct the low-order aberration of the system; And 5.2, extracting the system point spread function after correcting the position deviation of the secondary mirror again, comparing the system point spread function with the point spread function at the ideal position of the secondary mirror, if the compared point spread function deviation is within the error allowable range, completing the aberration correction process, otherwise, repeating the step 5.1.
2. The method for correcting aberration of telescope based on point spread function according to claim 1, it is characterized in that the method comprises the steps of, In the second step, the point spread function refers to a system facula image acquired by the CCD camera.
3. The method for correcting aberration of telescope based on point spread function according to claim 1, it is characterized in that the method comprises the steps of, In step 2.1, known offset errors of six degrees of freedom, namely an eccentric error and a tilting error in X, Y and a Z axis, are added to the secondary mirror.
4. The method for correcting aberration of telescope based on point spread function according to claim 1, it is characterized in that the method comprises the steps of, And step two, adding an aberration detection link for solving the offset of the secondary mirror according to the point spread function.
5. The telescope aberration correction method based on point spread function according to claim 1, wherein in the second step, the aberration of any field of view is expressed as Zernike polynomial coefficients: Where M is the total number of terms of the polynomial, A i represents the coefficient of the ith term Z i (ρ, φ), Z i represents the ith Xiang Zeni gram polynomial, ρ represents the normalized value of the exit pupil aperture, φ represents the aperture angle at the exit pupil.
6. The method for correcting aberration of telescope based on point spread function according to claim 1, it is characterized in that the method comprises the steps of, In the second step, the aberration caused by the minor misalignment is a lower order aberration, and the Z 4 ～Z 9 term of the zernike polynomial coefficient detected by the interferometer.
7. The method of claim 1, wherein in the fourth step, the actuator uses a six-degree-of-freedom displacement stage for controlling the spatial position of the secondary mirror to correct the system aberration.

Description

Telescope aberration correction method based on point spread function Technical Field The invention belongs to the field of aberration correction of telescope systems, and mainly aims at the problem of low-order aberration correction generated by the change of the relative positions of optical lenses in an optical system, so that the invention can be applied to static aberration correction during the adjustment of the telescope system and can also dynamically correct the low-order aberration generated during the use of the telescope system in real time. The method is not only suitable for off-axis systems, but also for on-axis systems, and can be used for solving the offset of a complex multi-mirror system. In particular to a telescope aberration correction method based on a point spread function. Background Among the optical systems, the off-axis reflective optical system has the advantages of no central obscuration, high light energy concentration, large field of view, high resolution and the like, and is increasingly widely applied to the fields of space optical communication and remote sensing. On one hand, the off-axis telescope is more sensitive to offset errors in the adjustment process, the non-rotational symmetrical structure of the off-axis telescope is higher in adjustment difficulty compared with the on-axis system, and on the other hand, the telescope is in use, external factors such as temperature change of the environment, air flow disturbance and the like, and internal factors such as self gravity and shake and the like can cause the spatial position of each optical lens to deviate from an adjustment design value, so that the correction of low-order aberration generated by the change of the relative position of each optical lens is a crucial step for guaranteeing the final imaging quality of the telescope system. On one hand, the vector wave aberration method and the sensitivity matrix method based on numerical fitting are both used for detecting the aberration by using an additional wave front sensor in actual use, wave aberration coefficients of a specified field of view are required to be detected, the complexity of a system is obviously increased, the solving precision is greatly dependent on the detecting precision of the wave aberration, the method is difficult to implement under the condition that no wave front sensor is used, and on the other hand, the existing method for skipping an aberration detection link is used, and the nonlinear optimization for solving the offset of each optical lens by directly utilizing the light spot diagram acquired by a CCD camera is required to be iterated for a plurality of times. In practical application scenes, the experimental requirement of taking an interferometer as wavefront sensing is very strict, and finding out a lens deviation for solving an optical system without adopting wavefront sensing is a problem to be solved by a person skilled in the art. Disclosure of Invention The invention aims at providing a telescope aberration correction method based on a point spread function aiming at low-order aberration generated by deviation of the spatial relative position of an optical lens and an ideal position, wherein in the method, in the early modeling stage, compared with the traditional method for solving the lens offset based on the point spread function, a wave aberration (Zernike polynomial coefficient) detection link is added, more characteristics of a facula image are extracted, and therefore, the system offset solving precision is improved; in the actual use process, no extra wavefront sensing equipment is needed, the offset of each element is solved by directly utilizing the facula image acquired by the CCD camera, so that the aberration is corrected, the complexity of an optical system is effectively reduced, and the imaging quality of the system is improved. The system of the invention mainly comprises a primary mirror optical system, a secondary mirror displacement table, a CCD camera, a neural network and an interferometer. The neural network mainly comprises an input layer, a plurality of hidden layers and an output layer. The principle of the invention is that the relative position imbalance of the lens of the optical system can cause the change of the wave aberration of the system, the wave aberration of the system and the point spread function of the optical system have a certain mapping relation, and the Zernike polynomial coefficient can well describe the wave aberration of the system, so the invention solves the spatial position imbalance of the lens of the optical system by extracting the characteristic of the Zernike polynomial coefficient of the point spread function. On the other hand, the mathematical model between the point spread function and the Zernike polynomial coefficient and the degree offset of each secondary mirror is complex, and the neural network has extremely strong nonlinear fitting capability. According to t