Search

CN-121977471-A - Unknown free-form surface interference detection device for quick recovery of sparse fringes

CN121977471ACN 121977471 ACN121977471 ACN 121977471ACN-121977471-A

Abstract

The invention discloses an unknown free-form surface interference detection device for quickly recovering sparse fringes, which relates to the field of optical detection and comprises a self-adaptive interference detection system and a self-adaptive aberration compensation system based on a polarization principle. The self-adaptive interference detection system integrates a commercial Fizeau interferometer, a deformable reflector (DM) and a partial zero compensation mirror (PNL), and on the basis of ensuring high robustness retrieval of wavefront data, the flexible aberration compensation capability of the interferometer is endowed, and the flexible detection level of the interferometer is remarkably improved; the self-adaptive aberration compensation system provided by the invention consists of the convolutional neural network and the ray tracing technology, the wavefront aberration prediction is carried out on the incomplete interference image through the convolutional neural network, and the required aberration compensation quantity is solved by combining the ray tracing technology, so that the rapid recovery of sparse fringes without iteration is realized, and the limitation problem that the traditional algorithm is low in convergence speed, long in time consumption and high in manual dependency is effectively solved.

Inventors

  • ZHANG LEI
  • FENG ZHAN
  • LIU RENHU

Assignees

  • 安徽大学

Dates

Publication Date
20260505
Application Date
20260116

Claims (5)

  1. 1. The unknown free-form surface interference detection device for quickly recovering sparse fringes is characterized by comprising an adaptive interference detection system (L1) and an adaptive aberration compensation system (L2) based on a polarization principle, wherein the adaptive interference detection system (L1) comprises a Fizeau interferometer (S1), a polaroid (S2), a polarization beam splitter (S3), a quarter wave plate (S4), a deformable mirror (S5), a part of zero compensation mirror (S6) and a measured free-form surface (S7), the Fizeau interferometer (S1), the polaroid (S2), the polarization beam splitter (S3), the quarter wave plate (S4) and the deformable mirror (S5) are in the same horizontal direction, and the part of zero compensation mirror (S6) and the measured free-form surface (S7) are in the same horizontal direction along the propagation direction of a detection beam; The detection light beam emitted by the Fizeau interferometer (S1) is modulated into p-polarized light through a linear polarizer (S2) and then enters a polarization beam splitter (S3), wherein the polarized light transmitted through the polarization beam splitter (S3) is converted into circularly polarized light through a quarter wave plate (S4), the circularly polarized light enters a deformable mirror (S5), then returns to be converted into S-polarized light through the quarter wave plate (S4), then is reflected to a partial zero compensation mirror (S6) through the polarization beam splitter (S3), then enters a tested freeform surface (S7), the reflected detection light beam returns to the original path, and finally returns to the Fizeau interferometer (S1) after passing through the partial zero compensation mirror (S6), the polarization beam splitter (S3), the quarter wave plate (S4), the deformable mirror (S5), the quarter wave plate (S4) and the polarization beam splitter (S3) in sequence, so that an initial incomplete interference pattern is obtained, and aberration compensation is needed through an adaptive aberration compensation system (L2) algorithm to obtain a measurable sparse fringe; The method comprises the steps of obtaining an initial incomplete interference pattern by a self-adaptive interference detection system (L1), inputting the initial incomplete interference pattern into a self-adaptive aberration compensation system (L2), preprocessing the incomplete interference pattern containing a dark area by a non-complete interference pattern preprocessing module through machine vision, inputting the preprocessed interference pattern into a trained convolutional neural network, extracting 2-16 Zernike coefficients corresponding to the interference pattern based on a wavefront aberration prediction module of the neural network to obtain an approximately real detection wavefront phase, performing ray tracing by using a ray tracing module to approximate the surface shape of a measured free-form surface, selecting a driving parameter corresponding to a DM deformation control module according to the aberration compensation quantity, and loading the driving parameter to the deformable mirror (S5) to enable the original incomplete interference pattern containing the dark area to be quickly restored to a measurable sparse fringe.
  2. 2. The unknown free-form surface interference detection device for rapid recovery of sparse fringes according to claim 1, wherein the deformable mirror (S5) is designed by adopting a continuous surface-shaped variable film driven by an electromagnetic field and is used for generating aberration compensation required by rapid recovery of sparse fringes, and after the unknown free-form surface is subjected to aberration compensation, an interference pattern of the measured free-form surface (S7) is enabled to reach the measurement range of a Fizeau interferometer.
  3. 3. An unknown free-form surface interference detection device for rapid recovery of sparse fringes according to claim 1, characterized in that the partial zero compensation mirror (S6) is used to compensate for rotationally symmetric aberrations.
  4. 4. The unknown freeform surface interference detection device for rapid recovery of sparse fringes according to claim 1, wherein preprocessing contents of the incomplete interferogram preprocessing module comprise dark area identification and gray value correction, and the preprocessing operation is completed by finding out the dark area in the initial incomplete interferogram and assigning 1.
  5. 5. The unknown freeform surface interference detection device for rapid recovery of sparse fringes according to claim 1, wherein a neural network in the neural network-based wavefront aberration prediction module is Xception, and input and output dimensions of Xception are respectively modified to 256×256×1 and 15×1.

Description

Unknown free-form surface interference detection device for quick recovery of sparse fringes Technical Field The invention relates to the field of optical detection, in particular to an unknown free-form surface interference detection device for quickly recovering sparse fringes. Background Free-form surfaces are commonly used in the leading edge field as a promising optical element due to their high degree of freedom. With the increasing demand for high performance optical systems, mass production of free-form surfaces is also becoming more and more urgent. High precision measurement of free-form surfaces has become a major bottleneck limiting this link. Interferometry is a widely used surface shape detection method because of the advantages of non-contact, full-field measurement and high accuracy. In freeform surface detection, an aberration compensator is required to tailor the wavefront aberration to match the nominal surface shape of the surface under test, or at least to control the residual wavefront aberration within a dynamic range measurable by the interferometer. A variety of static compensators were first studied and used. However, static compensators are not universal and cannot adapt to free-form surfaces that change continuously during processing. Adaptive optics such as Deformable Mirrors (DM) with programmable control wavefronts can provide flexible aberration compensation in interferometry of unknown free-form surfaces. Dense fringes and even dark areas (incomplete interferograms) due to large deviations are restored to sparse fringes by finding the required aberration compensation using an adaptive search algorithm. A random parallel gradient descent (SPGD) algorithm was first proposed. Monte-Carlo, however, shows that SPGD tends to sink to a local optimum and that the rate of non-convergence is about 21%. Therefore, a global search-based Genetic Algorithm (GA) was proposed to reduce the non-convergence rate to about 0.6%, but it took about 10 times of SPGD. That is, when the face deviation increases, the measurement efficiency may be severely reduced. In 2021, a simulated annealing-hill climbing (SA-HC) algorithm was proposed that alleviates to some extent the contradiction between time consuming and non-converging. However, the speed advantage is still not obvious compared to SPGD. 2022. We have further proposed an adaptive momentum estimation (Adam) based SPGD algorithm with a non-convergence rate of about 1% and achieved an acceleration of about 37% compared to SPGD. Although excellent in performance, the method is inconvenient in the self-adaptation process because performance indexes need to be adjusted at different stages. More importantly, the algorithm is used as a blind optimization scheme, a plurality of internal parameters still need to be set manually before operation, and the time consumption is increased along with the increase of the deviation of the surface shape. Disclosure of Invention Based on the defects of the prior art, the invention provides an unknown free-form surface interference detection device for quickly recovering sparse fringes. The technical scheme adopted by the invention is as follows: An unknown free-form surface interference detection device for quickly recovering sparse fringes comprises an adaptive interference detection system and an adaptive aberration compensation system based on a polarization principle; the adaptive interference detection system comprises a Fizeau interferometer, a polaroid, a polarization beam splitter, a quarter wave plate, a deformable reflecting mirror, a part of zero position compensation mirror and a measured free-form surface, wherein the Fizeau interferometer, the polaroid and the polarization beam splitter are in the same horizontal direction, the part of zero position compensation mirror and the measured free-form surface (S7) are in the same horizontal direction along the propagation direction of a detection light beam, the adaptive aberration compensation system comprises an incomplete interference pattern preprocessing module, a wavefront aberration prediction module based on a neural network, a light tracking module and a DM deformation control module which are sequentially connected, a detection light beam emitted by the Fizeau interferometer is modulated into p-polarized light through the linear polarizing plate and then is incident to the polarization beam splitter, the polarized light transmitted through the polarization beam splitter is converted into circular polarized light through the quarter wave plate and then is returned into S-polarized light, then is reflected to the part of the zero position compensation mirror through the polarization beam splitter and then is incident to the measured free-form surface, the detection light beam after being reflected returns to the original path, the initial path is sequentially returned through the partial zero position compensation, the quarter wave plate, the four