Multi-surface phase recovery method and system based on convolutional neural network

Abstract

The invention relates to the technical field of interference phase shift, and discloses a multi-surface phase recovery method and system based on a convolutional neural network. According to the method, four frames of simulated multi-surface phase-shift interferograms are used as input data, corresponding separated single-surface fringe interferograms are used as labels, and a multi-surface phase-shift interferogram fringe separation model based on a convolutional neural network is trained. And performing fringe separation on the multi-surface phase shift interferogram by using the trained optimal separation model to obtain a plurality of corresponding single-surface fringe prediction interferograms. And based on a plurality of single-surface fringe prediction interferograms, a phase distribution result of a corresponding single surface is recovered and obtained by using a four-step phase shift method. According to the method, only four frames of multi-surface phase shift interferograms need to be collected, and a four-step phase shift phase method is adopted for phase recovery, so that the wavelength tuning range required during measurement of the measured piece is reduced, and the thickness measurement range of the measured piece is enlarged. The accuracy of multi-surface phase shift interferogram separation is improved by using the convolutional neural network, and the anti-noise capability is high.

Description

Technical Field

[0001] The present application relates to the field of interference phase shift technology, and in particular to a multi-surface phase recovery method and system based on convolutional neural network. Background Art

[0002] In optical multi-surface interferometry, accurate extraction of the phase information of the front and back surfaces and thickness is crucial. Phase-shift interferometry is a widely used and effective phase extraction method in interferometry, which achieves high-precision phase recovery by recording multiple frames of phase-shift interferograms.

[0003] At present, in order to achieve high-precision multi-surface interferometry, there are many classic multi-surface phase recovery algorithms, such as the 19-frame phase recovery algorithm, the multi-surface iterative algorithm based on least squares, the weighted multi-step phase shift recovery algorithm, etc. However, these multi-surface phase recovery algorithms require a large number of multi-surface phase shift interferograms to recover the phase information more accurately.

[0004] And when actually collecting multi-surface phase-shift interferograms, wavelength-tuned interferometers are often used to collect the required multi-frame multi-surface phase-shift interferograms. The thinner the thickness of the measured component, the more interferogram frames collected, and the smaller the interference cavity length, the larger the required wavelength tuning range. However, since the output wavelength of the wavelength-tuned laser has a range limitation (about 0.2633nm), when the component is too thin or the number of collected frames is too large, it will not be possible to obtain a sufficient number of multi-surface phase-shift interferograms, and thus the measurement cannot be completed. Summary of the invention

[0005] The present application provides a multi-surface phase recovery method based on a convolutional neural network to solve the problem in the prior art that the existing multi-surface phase recovery algorithm introduces a large environmental error due to the excessive number of interference pattern frames required, and when the test piece is too thin, it is difficult to obtain a sufficient number of interference patterns within a limited wavelength range.

[0006] Correspondingly, the present application also provides a multi-surface phase recovery system based on convolutional neural network, an electronic device, and a computer-readable storage medium to ensure the implementation and application of the above method.

[0007] In order to solve the above technical problems, the present application discloses a multi-surface phase recovery method based on a convolutional neural network, the method comprising:

[0008] Generate four frames of simulated multi-surface phase-shift interferograms in four different phase states, and multiple separated single-surface fringe interferograms corresponding to each frame of simulated multi-surface phase-shift interferogram; the multiple single-surface fringe interferograms include a front surface fringe interferogram, a rear surface fringe interferogram, and a thickness fringe interferogram;

[0009] Construct a multi-surface phase-shift interference pattern fringe separation model based on convolutional neural network;

[0010] Four frames of simulated multi-surface phase-shift interferograms are used as input data, and the corresponding separated multiple single-surface fringe interferograms are used as labels to train the multi-surface phase-shift interferogram fringe separation model and obtain the optimal separation model.

[0011] The optimal separation model is used to separate the fringes of four frames of multi-surface phase-shift interferograms in four different phase states, and multiple single-surface fringe prediction interferograms in the corresponding phase states are obtained; the multiple single-surface fringe prediction interferograms include the front surface fringe prediction interferogram, the back surface fringe prediction interferogram and the thickness fringe prediction interferogram;

[0012] Based on multiple single-surface fringe prediction interferograms under four different phase states, the four-step phase shift method is used to recover the phase distribution results of the corresponding single surface.

[0013] The present application also discloses a multi-surface phase recovery system based on a convolutional neural network, the system comprising:

[0014] A data generation module is used to generate four frames of simulated multi-surface phase-shift interferograms in four different phase states, and a plurality of separated single-surface fringe interferograms corresponding to each frame of simulated multi-surface phase-shift interferogram; the plurality of single-surface fringe interferograms include a front surface fringe interferogram, a rear surface fringe interferogram, and a thickness fringe interferogram;

[0015] Model building module, used to build a multi-surface phase-shift interferogram fringe separation model based on convolutional neural network;

[0016] The model optimization module is used to take four frames of simulated multi-surface phase-shift interferograms as input data and the corresponding separated multiple single-surface fringe interferograms as labels, train the multi-surface phase-shift interferogram fringe separation model, and obtain the optimal separation model;

[0017] A separation prediction module is used to perform fringe separation on four frames of multi-surface phase-shift interferograms in four different phase states using an optimal separation model to obtain multiple single-surface fringe prediction interferograms in corresponding phase states; the multiple single-surface fringe prediction interferograms include a front surface fringe prediction interferogram, a rear surface fringe prediction interferogram, and a thickness fringe prediction interferogram;

[0018] The phase recovery module is used to predict the interference pattern based on multiple single-surface fringes in four different phase states, and to recover the phase distribution result of the corresponding single surface using the four-step phase shift method.

[0019] The present application also discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, one or more methods described in the present application are implemented.

[0020] The present application also discloses a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, one or more methods described in the present application are implemented.

[0021] In the present application, four frames of simulated multi-surface phase-shift interferograms in four different phase states, and the corresponding separated front surface fringe interferograms, rear surface fringe interferograms and thickness fringe interferograms are generated. A multi-surface phase-shift interferogram fringe separation model based on a convolutional neural network is constructed; the four frames of simulated multi-surface phase-shift interferograms are used as input data, and the corresponding separated multiple single-surface fringe interferograms are used as labels to train the multi-surface phase-shift interferogram fringe separation model to obtain the optimal separation model. The four frames of multi-surface phase-shift interferograms in four different phase states are separated by the optimal separation model to obtain the front surface fringe predicted interferogram, the rear surface fringe predicted interferogram and the thickness fringe predicted interferogram in the corresponding phase state. Based on the multiple single-surface fringe predicted interferograms in four different phase states, the phase distribution result of the corresponding single surface is recovered by the four-step phase shift method. In the present application, only four frames of multi-surface phase-shift interferograms need to be collected, and a more mature and accurate four-step phase-shift phase recovery algorithm is used, which greatly reduces the wavelength tuning range required for measuring the measured object, and further increases the thickness measurement range of the measured object. The method in this application is an end-to-end processing, which uses a model based on a convolutional neural network to extract interference pattern fringe features, can effectively capture the complex features in multi-surface interference patterns, improve the accuracy of multi-surface phase-shift interference pattern separation, and has strong noise resistance.

[0022] Additional aspects and advantages of the present application will be given in the following description, which will become apparent from the following description, or will be understood through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0024] Figure 1 Flow chart of multi-surface phase recovery method based on convolutional neural network;

[0025] Figure 2 It is a structural schematic diagram of the multi-surface phase-shift interference pattern fringe separation model;

[0026] Figure 3 It is a schematic diagram of the training phase of the multi-surface phase-shift interference pattern fringe separation model;

[0027] Figure 4 (a) is the input data for the first set of model tests;

[0028] Figure 4 (b) is the front surface stripe label data for the first set of model tests;

[0029] Figure 4 (c) Output data of the front surface stripes for the first set of model tests;

[0030] Figure 4 (d) is the thickness fringe label data of the first set of model tests;

[0031] Figure 4 (e) Thickness fringe output data for the first set of model tests;

[0032] Figure 4 (f) is the rear surface stripe label data of the first group of model tests;

[0033] Figure 4 (g) is the rear surface stripe output data of the first set of model tests;

[0034] Figure 5 (a) The front surface phase image obtained by restoring the front surface fringe output data of the first group of model tests using the four-step phase shift method;

[0035] Figure 5 (b) The error between the front surface phase image obtained by restoring the front surface fringe output data of the first group of model tests using the four-step phase shift method and the true phase image;

[0036] Figure 5 (c) is the thickness phase map obtained after the thickness fringe output data of the first group of model tests was restored using the four-step phase shift method;

[0037] Figure 5 (d) The error between the thickness phase image and the true phase image after the thickness fringe output data of the first group of model tests was restored using the four-step phase shift method;

[0038] Figure 5 (e) The rear surface phase image obtained after the rear surface fringe output data of the first group of model tests was restored using the four-step phase shifting method;

[0039] Figure 5 (f) The error between the rear surface phase image and the real phase image after the rear surface fringe output data of the first group of model tests was restored using the four-step phase shift method;

[0040] Figure 6 (a) is the input data for the second set of model tests;

[0041] Figure 6 (b) is the front surface stripe label data for the second set of model tests;

[0042] Figure 6 (c) front surface fringe output data for the second set of model tests;

[0043] Figure 6 (d) is the thickness fringe label data of the second set of model tests;

[0044] Figure 6 (e) Thickness fringe output data for the second set of model tests;

[0045] Figure 6 (f) is the rear surface stripe label data of the second group of model tests;

[0046] Figure 6 (g) is the rear surface stripe output data of the second set of model tests;

[0047] Figure 7 (a) The front surface phase image obtained by restoring the front surface fringe output data of the second group of model tests using the four-step phase shifting method;

[0048] Figure 7 (b) The error between the front surface phase image and the real phase image after the front surface fringe output data of the second group of model tests were restored using the four-step phase shift method;

[0049] Figure 7 (c) is the thickness phase map obtained after the thickness fringe output data of the second group of model tests was restored using the four-step phase shift method;

[0050] Figure 7 (d) The error between the thickness phase image and the true phase image after the thickness fringe output data of the second group of model tests was restored using the four-step phase shift method;

[0051] Figure 7 (e) The rear surface phase image obtained after the rear surface fringe output data of the second group of model tests was restored using the four-step phase shifting method;

[0052] Figure 7 (f) is the error between the rear surface phase image obtained after the rear surface fringe output data of the second group of model tests was restored using the four-step phase shift method and the true phase image;

[0053] Figure 8 (a) Input data for the third set of model tests;

[0054] Figure 8 (b) is the front surface stripe label data for the third group of model tests;

[0055] Figure 8 (c) Output data of the front surface stripes for the third set of model tests;

[0056] Figure 8 (d) is the thickness fringe label data of the third group of model tests;

[0057] Figure 8 (e) Thickness fringe output data for the third set of tests;

[0058] Figure 8 (f) is the rear surface stripe label data of the third group of model tests;

[0059] Figure 8 (g) is the rear surface stripe output data of the third group of model tests;

[0060] Fig. 9 (a) The front surface phase image obtained by restoring the front surface fringe output data of the third group of model tests using the four-step phase shifting method;

[0061] Fig. 9 (b) The error between the front surface phase image and the true phase image after the front surface fringe output data of the third group of model tests was restored using the four-step phase shift method;

[0062] Fig. 9 (c) is the thickness phase map obtained after the thickness fringe output data of the third group of model tests was restored using the four-step phase shift method;

[0063] Fig. 9 (d) The error between the thickness phase image and the true phase image after the thickness fringe output data of the third group of model tests was restored using the four-step phase shift method;

[0064] Fig. 9 (e) The rear surface phase image obtained after the rear surface fringe output data of the third group of model tests was restored using the four-step phase shifting method;

[0065] Fig. 9 (f) is the error between the rear surface phase image obtained by restoring the rear surface fringe output data of the third group of model tests using the four-step phase shift method and the true phase image;

[0066] Fig.10 A schematic diagram of the structure of a vehicle speed measurement system based on road monitoring video provided in an embodiment of the present application;

[0067] Fig.11 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0068] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be interpreted as limiting the present application.

[0069] It will be understood by those skilled in the art that, unless expressly stated, the singular forms "one", "said", and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present application refers to the presence of features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof. It should be understood that when we refer to an element as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, or there may be an intermediate element. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The term "and / or" used herein includes all or any unit and all combinations of one or more associated listed items.

[0070] Those skilled in the art will appreciate that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as those generally understood by those skilled in the art in the field to which the present invention belongs. It should also be understood that terms such as those defined in general dictionaries should be understood to have meanings consistent with the meanings in the context of the prior art, and will not be interpreted with idealized or overly formal meanings unless specifically defined as here.

[0071] The solution provided in the embodiments of the present application can be executed by any electronic device, such as a terminal device or a server, wherein the server can be an independent physical server, a server cluster or a distributed system composed of multiple physical servers, or a cloud server that provides cloud computing services. The terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a smart watch, etc., but is not limited to this. The terminal and the server can be directly or indirectly connected via wired or wireless communication, and the present application does not limit this. With respect to the technical problems existing in the prior art, the multi-surface phase recovery method and system based on convolutional neural network provided in the present application is intended to solve at least one of the technical problems of the prior art.

[0072] The technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0073] Existing multi-surface phase recovery algorithms require a large number of multi-surface phase shift interferograms to recover phase information more accurately. In contrast, single-surface phase recovery algorithms require fewer interferograms and are more mature. For example, the four-step phase shift method has the advantages of high precision, strong noise resistance, wide application range, and simple and easy algorithm implementation. It has good applicability and stability in the field of interferometric measurement. Based on this, a possible implementation method is provided in the embodiment of the present application, such as Figure 1 As shown, a flowchart of a multi-surface phase recovery method based on a convolutional neural network is provided, which can be executed by any electronic device, and optionally, can be executed on a server or terminal device.

[0074] like Figure 1 As shown in , the method may include the following steps:

[0075] Step 101, generating four frames of simulated multi-surface phase-shift interferograms in four different phase states, and a plurality of separated single-surface fringe interferograms corresponding to each frame of simulated multi-surface phase-shift interferogram.

[0076] The multiple single-surface fringe interference patterns include a front surface fringe interference pattern, a back surface fringe interference pattern and a thickness fringe interference pattern.

[0077] In the embodiment of the present application, four frames of simulated multi-surface phase-shift interference patterns and corresponding separated front surface fringe interference patterns, rear surface fringe interference patterns and thickness fringe interference patterns are generated based on Zernike polynomials, wavelength tuning principles and interference light intensity superposition theoretical models. Among them, the four frames of simulated multi-surface phase-shift interference patterns are in different phase conditions, and each frame increases the phase shift value of π / 2 on the basis of the previous frame.

[0078] Step 102, constructing a multi-surface phase-shift interference pattern fringe separation model based on a convolutional neural network.

[0079] Step 103 , using four frames of simulated multi-surface phase-shift interferograms as input data and corresponding separated multiple single-surface fringe interferograms as labels, training a multi-surface phase-shift interferogram fringe separation model to obtain an optimal separation model.

[0080] The input data is divided into four channels and input into the multi-surface phase-shift interferogram fringe separation model to learn the front surface fringe features, the back surface fringe features and the thickness fringe features, and output the corresponding front surface fringe predicted interferogram, the back surface fringe predicted interferogram and the thickness fringe predicted interferogram. In the iterative training process, the front surface fringe predicted interferogram, the back surface fringe predicted interferogram and the thickness fringe predicted interferogram are compared with the corresponding labels (front surface fringe interferogram, back surface fringe interferogram and thickness fringe interferogram), the loss is calculated, and the multi-surface phase-shift interferogram fringe separation model is updated until the optimal separation model is obtained after the preset conditions are met.

[0081] The method in the embodiment of the present application is to separate the front and surface stripes, the rear surface stripes and the thickness stripes from the original cross stripes by inputting 4 frames of multi-surface phase-shift interferograms through a convolutional neural network. Therefore, the accuracy of the training data is crucial. At the same time, the input data and the label data are in a one-to-one correspondence to ensure that the model can correctly learn the corresponding relationship between the input and the output during the training process.

[0082] Step 104, using the optimal separation model to perform fringe separation on four frames of multi-surface phase-shift interferograms in four different phase states, to obtain multiple single-surface fringe prediction interferograms in corresponding phase states.

[0083] The multiple single-surface fringe prediction interference patterns include a front-surface fringe prediction interference pattern, a rear-surface fringe prediction interference pattern and a thickness fringe prediction interference pattern.

[0084] Step 105 , based on the multiple single-surface fringe prediction interference patterns in four different phase states, a four-step phase shift method is used to restore the phase distribution result of the corresponding single surface.

[0085] In the embodiment of the present application, four frames of simulated multi-surface phase-shift interferograms in four different phase states, and the corresponding separated front surface fringe interferograms, rear surface fringe interferograms and thickness fringe interferograms are generated. A multi-surface phase-shift interferogram fringe separation model based on a convolutional neural network is constructed; the four frames of simulated multi-surface phase-shift interferograms are used as input data, and the corresponding separated multiple single-surface fringe interferograms are used as labels to train the multi-surface phase-shift interferogram fringe separation model to obtain the optimal separation model. The four frames of multi-surface phase-shift interferograms in four different phase states are separated by the optimal separation model to obtain the front surface fringe predicted interferogram, the rear surface fringe predicted interferogram and the thickness fringe predicted interferogram in the corresponding phase state. Based on the multiple single-surface fringe predicted interferograms in four different phase states, the phase distribution result of the corresponding single surface is recovered by the four-step phase shift method. In the embodiment of the present application, only four frames of multi-surface phase-shift interferograms need to be collected, and a more mature and accurate four-step phase-shift phase recovery algorithm is used, which greatly reduces the wavelength tuning range required for measuring the measured object, and further increases the thickness measurement range of the measured object. The method in the embodiment of the present application is an end-to-end processing, which uses a model based on a convolutional neural network to extract interference pattern fringe features, can effectively capture the complex features in multi-surface interference patterns, improve the accuracy of multi-surface phase shift interference pattern separation, and has strong noise resistance.

[0086] As a first example, the mathematical model of the generated simulated multi-surface phase-shift interferogram and the corresponding separated front surface fringe interferogram, rear surface fringe interferogram and thickness fringe interferogram is expressed as follows:

[0087]

[0088] In the formula, I k (x, y) is the simulated multi-surface phase-shift interference pattern; I1 k (x, y) is the corresponding front surface fringe interference pattern; I2 k (x, y) is the corresponding rear surface fringe interference pattern; I3 k (x, y) is the corresponding thickness fringe interference pattern; k is the number of interference pattern frames (k = 1 to 4), and k is 1 in the initial interference pattern; a(x, y) is the background light intensity, and a(x, y) = 0.5; b1(x, y), b2(x, y), and b3(x, y) are the modulation degrees of the front surface, the back surface, and the thickness, respectively, and b1(x, y) = b2(x, y) = 0.3, and b3(x, y) = 0.2; are the front surface phase information, back surface phase information and thickness information to be restored respectively, where represented by Zernike polynomials); θ 1k (x, y), θ 2k (x, y), θ3k (x, y) are the phase shift values ​​of the front surface, the back surface and the thickness respectively (θ 11 (x, y) = θ 21 (x, y) = θ 31 (x, y) = 0), where the front surface phase shift value is π / 4; η k (x, y) is the noise added to the interference pattern.

[0089] In the above multi-surface phase-shift interferogram, the Zernike polynomials involved can represent optical aberrations, theoretical wavefronts, and fitted wavefronts. The process of generating simulated interferograms involves using Zernike polynomials to simulate different aberration effects, superimposing them together to generate phase distribution, and then generating 4 frames of multi-surface phase-shift interferograms.

[0090] The mathematical model expression of phase distribution based on Zernike polynomials is as follows:

[0091]

[0092] In the formula, is the front surface phase distribution, is the phase distribution on the rear surface, c 1i 、c 2i They are and The coefficient of the i-th term of the Zernike polynomial; Z i (x, y) is the i-th term of the Zernike polynomial, n is the total number of terms of the Zernike polynomial used, and in the embodiment of the present application, n=20.

[0093] In an embodiment of the present application, when collecting input data and corresponding labels, the first twenty terms of the Zernike polynomial are used to generate a phase distribution to meet the diversity of training data; random noise with different signal-to-noise ratios (20 to 100 dB) is added to make the interference pattern collected in a real environment as close as possible.

[0094] The method in the embodiment of the present application is applicable to the processing of multi-surface phase shift interference patterns with a front surface phase shift value of π / 4.

[0095] In an optional embodiment, the multi-surface phase-shift interference pattern fringe separation model includes a down-sampling module, an up-sampling module and an output layer;

[0096] The downsampling module is used to extract multiple single surface stripe features corresponding to the input data by reducing the size of the feature map; the multiple single surface stripe features include front surface stripe features, rear surface stripe features and thickness stripe features;

[0097] The upsampling module includes a plurality of independent upsampling parts, and the upsampling parts are used to restore the corresponding single surface stripe features to the size of the input data; wherein the plurality of upsampling parts correspond to the plurality of single surface stripe features one by one;

[0098] There are multiple output layers, which correspond to the up-sampling parts one by one, and are used to convert the channel features of the output of the corresponding up-sampling part into the corresponding separated single surface fringe prediction interference map.

[0099] In an optional embodiment, the down-sampling module includes a first down-sampling part and a second down-sampling part;

[0100] The first downsampling part includes a plurality of convolutional layers; the first sampling part is used to reduce the size of the feature map by a step size specified by the convolutional layer;

[0101] The second downsampling part includes a plurality of alternating convolutional layers and maximum pooling layers, and the second downsampling part is used to reduce the size of the feature map by using the maximum pooling layer.

[0102] As a second example, the input data is four frames of simulated multi-surface phase-shift interference patterns, the image size is 256*256 pixels, divided into four channels, and input into the multi-surface phase-shift interference pattern fringe separation model.

[0103] like Figure 2 As shown, the multi-surface phase-shifting interference pattern fringe separation model includes a shared first down-sampling part, a second down-sampling part, three independent up-sampling parts and three independent output layers.

[0104] The first downsampling part consists of 7 "convolution layers + activation functions", from DownA_1 to DownA_7. In DownA_1, the input is a 256*256 4-channel simulated multi-surface phase shift interference pattern, and 16 convolution kernels with a step size of 2 and a size of 3*3 are used to extract low-level features, and then through the ReLU activation function, a 16-channel feature map with a size of 128*128 is generated. Similarly, 32, 64, 128, 256, 512 and 1024 channel feature maps are obtained in subsequent convolution layers (DownA_2 to DownA_7), and the feature map sizes are 64*64, 32*32, 16*16, 8*8, 4*4 and 2*2 respectively. The first downsampling part reduces the size of the feature map by a step size of 2, which can pay more attention to the overall feature expression, which is conducive to the network model learning the distribution of background light intensity.

[0105] The second downsampling part consists of 7 "convolution layers + activation functions" and 7 maximum pooling layers, from DownB_1 to DownB_7. In DownB_1, the input is a 256*256 4-channel simulated multi-surface phase shift interference pattern, and 16 convolution kernels with a step size of 1 and a size of 3*3 are used to extract the underlying features, and then through the ReLU activation function, a 16-channel feature map of size 256*256 is obtained; then a filter with a step size of 2 and a size of 2*2 is used for maximum pooling, and a 16-channel feature map of size 128*128 is obtained. Similarly, in the subsequent parts (DownB_2 to DownB_7), 32, 64, 128, 256, 512 and 1024 channel feature maps are obtained, and their sizes are 64*64, 32*32, 16*16, 8*8, 4*4 and 2*2 respectively. The second downsampling part reduces the size of the feature map through the maximum pooling layer, which can retain more feature information and is more sensitive to feature details, which is conducive to the learning of the front surface fringe features, the back surface fringe features and the thickness fringe features by the multi-surface phase-shift interferogram fringe separation model.

[0106] The three independent upsampling parts are composed of 7 "convolution layers + activation functions" and 7 bilinear interpolation upsampling layers, from UpA_1 to UpA_7, UpB_1 to UpB_7, and UpC_1 to UpC_7. Since the structures of the three independent upsampling parts are the same, the embodiment of the present application takes the UpA part as an example for explanation. In UpA_1, the input is a 2048-channel feature map of size 2*2 obtained by splicing the 1024-channel feature map of size 2*2 output by DownA_7 and the 1024-channel feature map of size 2*2 output by DownB_7. 1024 convolution kernels with a step size of 1 and a size of 3*3 are used to extract deep features, and then the Relu activation function is used to generate a 1024-channel feature map of size 2*2; then the feature map size is restored through the bilinear interpolation upsampling layer, thereby obtaining a 1024-channel feature map of size 4*4. In UpA_2, the input is a 4*4 2048-channel feature map obtained by concatenating the 4*4 512-channel feature map output by DownB_6 and the 4*4 1024-channel feature map output by UpA_1, and then a 8*8 1024-channel feature map is obtained through a convolutional layer and a bilinear interpolation upsampling layer. Similarly, 768, 512, 320, 192, and 112-channel feature maps are obtained in the subsequent upsampling parts (UpA_3 to UpA_7), and the feature map sizes are 16*16, 32*32, 64*64, 128*128, and 256*256, respectively. Three independent upsampling parts are used to restore the feature images extracted by the downsampling part to the original input image size and combine features of different resolutions.

[0107] Finally, three independent output layers are used. Each of the three independent output layers is composed of three "convolutional layers + activation functions" and one convolutional layer, namely OutA, OutB and OutC. Since the structures of the three independent output layers are the same, the embodiment of the present application takes the OutA part as an example for explanation. In OutA, the input is a 112-channel feature map of size 256*256 output by UpA_7. Through three "convolutional layers + ReLU activation functions" with a step size of 1, a convolution kernel size of 3*3, and the number of convolution kernels of 56, 28 and 14 respectively, a 14-channel feature map of size 256*256 is obtained; then, four convolution kernels with a step size of 1 and a size of 1*1 are used to convert the 14-channel feature map of size 256*256 into a 4-channel feature map of size 256*256 as the final output.

[0108] In general, the embodiment of the present application adopts a multi-task learning method. The multi-surface phase-shift interferogram fringe separation model extracts the feature information from the input 4-channel simulated multi-surface phase-shift interferogram through a downsampling module to obtain a feature map with richer feature information. The obtained feature information is then utilized through a jump connection (splicing) and an upsampling module, and finally the corresponding front surface fringe prediction interferogram, rear surface fringe prediction interferogram, and thickness fringe prediction interferogram (3 groups of independent fringes) are separated from the multi-surface phase-shift interferogram (cross fringes).

[0109] In an optional embodiment, four frames of simulated multi-surface phase-shift interferograms are used as input data, and the corresponding separated multiple single-surface fringe interferograms are used as labels to train a multi-surface phase-shift interferogram fringe separation model to obtain an optimal separation model, including:

[0110] Acquire multiple sets of input data and corresponding labels; each set of input data includes four frames of simulated multi-surface phase-shift interferograms;

[0111] Input multiple sets of input data into the multi-surface phase-shift interference pattern fringe separation model to obtain multiple sets of prediction results; each set of prediction results includes multiple single-surface fringe prediction interference patterns;

[0112] Use the loss function to calculate the difference between each set of prediction results and the corresponding labels, and calculate the gradient of the loss function to the model;

[0113] The parameters of the fringe separation model are determined based on the gradient multi-surface phase-shift interference pattern, and the optimal separation model is obtained when the preset number of iterations is reached.

[0114] In an optional embodiment, multiple sets of input data are input into a multi-surface phase-shift interference pattern fringe separation model to obtain multiple sets of prediction results, including:

[0115] The multiple sets of input data are shuffled through DataLoader, the number of data transmitted in a single time is increased, and then input into the multi-surface phase-shift interferogram fringe separation model;

[0116] Multiple groups of prediction results are obtained through the multi-surface phase-shift interference pattern fringe separation model prediction.

[0117] In an optional embodiment, the loss function is expressed as:

[0118] L=L1+L2+L3

[0119] Among them, L1 is the front surface fringe loss function, which is used to calculate the difference between the front surface fringe predicted interference pattern and the corresponding label, L2 is the back surface fringe loss function, which is used to calculate the difference between the back surface fringe predicted interference pattern and the corresponding label, and L3 is the thickness fringe loss function, which is used to calculate the difference between the thickness fringe predicted interference pattern and the corresponding label.

[0120] In an optional embodiment, the front surface fringe loss function, the rear surface fringe loss function and the thickness fringe loss function are all mean absolute error functions.

[0121] As a third example, 5000 sets of input data (four frames of simulated multi-surface phase-shift interferograms) are shuffled through DataLoader, Batch_Size (the number of data passed to the program for training at a single time) is increased, and then input into the multi-surface phase-shift interferogram fringe separation model. Through the downsampling module and the upsampling module, layer-by-layer calculations are performed to generate three independent estimated results (front surface fringe predicted interferogram, rear surface fringe predicted interferogram, and thickness fringe predicted interferogram). The three estimated results are compared with their corresponding labels (true values), and the mean absolute error function (MAE) is used to measure the difference between the estimated results and the labels, and the sum of the three MAE loss functions is used as the total loss function.

[0122] Set the learning rate to 0.001 and use the Adam optimization algorithm to find the network weights that minimize the loss function.

[0123] like Figure 3 As shown in the figure, 4 frames of simulated multi-surface phase-shift interferograms (input data) are separated into 4 frames of front surface fringe prediction interferograms, 4 frames of rear surface fringe prediction interferograms and 4 frames of front surface fringe prediction interferograms (estimated results) after passing through the multi-surface phase-shift interferogram fringe separation model. The corresponding label data are 4 frames of front surface fringe interferograms, 4 frames of rear surface fringe interferograms and 4 frames of thickness fringe interferograms. The multi-surface phase-shift interferogram fringe separation model is trained by the loss function (MAE) and the optimizer (Adam).

[0124] Specifically, the number of iterations is set to 100 times. In each iteration, the following steps are performed: First, forward propagation is performed to input the input data into the model and calculate the estimated result. Then, the difference between the estimated result and the labeled data is measured by the loss function. Next, back propagation is performed to calculate the gradient of the loss function to the model parameters. Finally, the Adam optimization algorithm is used to update the weights and biases of the multi-surface phase-shift interferogram fringe separation model according to the gradient information, so that the loss function gradually decreases.

[0125] Repeat these steps until the predetermined number of iterations is reached. Each iteration trains the entire training data set (5000 sets of input data) to gradually optimize the performance of the model. Finally, after 100 iterations of training, the optimal separation model is obtained, which can accurately separate the stripes of the input data.

[0126] The specific process is as follows:

[0127] Step 1, data reading: Use DataLoader to load 5000 sets of input data (4 frames of simulated multi-surface phase-shift interferograms) and generate estimated results through calculation of the multi-surface phase-shift interferogram fringe separation model. Its main function is to disrupt the data order (random reading) and increase Batch_size (the number of data passed to the program for training at a single time).

[0128] Step 2, model definition: define the network structure of the multi-surface phase-shift interference pattern fringe separation model and define the weight initialization function.

[0129] Step 3, loss and optimizer: Compare the estimated results with the corresponding labeled (true value) data, and use the mean absolute error function (MAE) to measure the difference between the separated front surface fringe predicted interferogram, back surface fringe predicted interferogram and thickness fringe predicted interferogram and the corresponding labeled data.

[0130] Define the loss function (MAE) and optimizer (Adam) for the training model. By setting the loss function, the mean absolute error is calculated to measure the difference between the separated front surface fringe prediction interferogram, the back surface fringe prediction interferogram and the thickness fringe prediction interferogram and the label data. The mathematical model of the total loss function and the three independent MAE loss functions is expressed as follows:

[0131] L=L1+L2+L3

[0132]

[0133] In the formula, y 1i Predict the interference pattern for the front surface fringes, is the front surface stripe label data, y 2i Predict the interference pattern for the back surface fringes, is the rear surface stripe label data, y 3i Predict the interference pattern for the thickness fringes, is the thickness stripe label data, and n is the number of training samples.

[0134] Among them, the front surface fringe interference pattern, the back surface fringe interference pattern and the thickness fringe interference pattern after separation Figure 3 Each labeled data corresponds to three independent MAE loss functions, which can maintain the accuracy and correlation between the three outputs while reducing the training time.

[0135] In this step, the Adam optimizer is used to adjust the parameters of the neural network to the optimal state, and the learning rate is adjusted at the specified round. The optimizer is updated with the new learning rate to minimize the loss function. The MAE loss function and the Adam optimizer are key components in the training process of the neural network. The model parameters are updated through back propagation and gradient descent to minimize the loss between the prediction results and the label data, thereby improving network performance.

[0136] Step 4, training loop: Iterate the dataset for a certain number of rounds, set the number of iterations to 100, and perform forward propagation, loss calculation, back propagation, and parameter update. Each iteration will train the entire training data once, that is, in each round, traverse the data batch (Batch) and perform a training loop.

[0137] Step 5, save the model: After each round, save the trained model file to the specified folder, and save the learning rate and loss value corresponding to each round in an Excel table to facilitate the analysis of model performance after training.

[0138] In an embodiment of the present application, four frames of re-simulated multi-surface phase-shift interferogram data are used as a test set, the test data is input into the optimal separation model, and the separated front surface fringe predicted interferogram, rear surface fringe predicted interferogram and thickness fringe predicted interferogram are output; in Matlab, the separated front surface fringe predicted interferogram, rear surface fringe predicted interferogram and thickness fringe predicted interferogram are compared with the label data to further evaluate the accuracy of the model.

[0139] The specific process is as follows:

[0140] Set up the GPU device and load the pre-test model, and use the best separation model to load into the test.

[0141] By inputting the test data set into the optimal separation model, the fringe separation results (predicted interference map of front surface fringe, predicted interference map of back surface fringe and predicted interference map of thickness fringe) can be obtained, and the final fringe separation results are compared and evaluated with the label data. In theory, the fringe separation results output by the neural network model are similar to the label data with very small differences.

[0142] The four-step phase shift method is used to recover the phase of the stripe separation results to verify its effectiveness. The performance of the optimal separation model is evaluated by calculating the peak-to-valley (PV) value and root mean square (RMS) value of the recovered phase. Finally, three sets of data (accurate to 4 decimal places) are selected from the test results. The test input data, label data, separation output results, and phase recovery results are shown in Figure 2. Figure 4-Figure 9 The test results are compared in Table 1.

[0143] Table 1 Test results comparison

[0144]

[0145]

[0146] According to the above test results, the error between the actual multi-surface separation result and the predicted separation result of the multi-surface phase-shift interference pattern fringe separation model is extremely small, that is, the fitting effect of the multi-surface phase-shift interference pattern fringe separation model is good, and the test results in the embodiments of the present application are within ideal expectations.

[0147] Based on the same principle as the method provided in the embodiment of the present application, the embodiment of the present application also provides a multi-surface phase recovery system based on a convolutional neural network, such as Fig.10 As shown, the system comprises:

[0148] The data generation module 1001 is used to generate four frames of simulated multi-surface phase-shift interference patterns in four different phase states, and a plurality of separated single-surface fringe interference patterns corresponding to each frame of simulated multi-surface phase-shift interference patterns; the plurality of single-surface fringe interference patterns include a front surface fringe interference pattern, a rear surface fringe interference pattern, and a thickness fringe interference pattern;

[0149] A model building module 1002 is used to build a multi-surface phase-shift interference pattern fringe separation model based on a convolutional neural network;

[0150] The model optimization module 1003 is used to use four frames of simulated multi-surface phase-shift interferograms as input data and the corresponding separated multiple single-surface fringe interferograms as labels to train the multi-surface phase-shift interferogram fringe separation model and obtain the optimal separation model;

[0151] The separation prediction module 1004 is used to perform fringe separation on four frames of multi-surface phase-shift interference images in four different phase states using an optimal separation model to obtain multiple single-surface fringe prediction interference images in corresponding phase states; the multiple single-surface fringe prediction interference images include a front surface fringe prediction interference image, a rear surface fringe prediction interference image, and a thickness fringe prediction interference image;

[0152] The phase recovery module 1005 is used to predict the interference pattern based on multiple single-surface fringes in four different phase states, and to recover the phase distribution result of the corresponding single surface using a four-step phase shift method.

[0153] In the embodiment of the present application, four frames of simulated multi-surface phase-shift interferograms in four different phase states, and the corresponding separated front surface fringe interferograms, rear surface fringe interferograms and thickness fringe interferograms are generated. A multi-surface phase-shift interferogram fringe separation model based on a convolutional neural network is constructed; the four frames of simulated multi-surface phase-shift interferograms are used as input data, and the corresponding separated multiple single-surface fringe interferograms are used as labels to train the multi-surface phase-shift interferogram fringe separation model to obtain the optimal separation model. The four frames of multi-surface phase-shift interferograms in four different phase states are separated by the optimal separation model to obtain the front surface fringe predicted interferogram, the rear surface fringe predicted interferogram and the thickness fringe predicted interferogram in the corresponding phase state. Based on the multiple single-surface fringe predicted interferograms in four different phase states, the phase distribution result of the corresponding single surface is recovered by the four-step phase shift method. In the embodiment of the present application, only four frames of multi-surface phase-shift interferograms need to be collected, and a more mature and accurate four-step phase-shift phase recovery algorithm is used, which greatly reduces the wavelength tuning range required for measuring the measured object, and further increases the thickness measurement range of the measured object. The method in the embodiment of the present application is an end-to-end processing, which uses a model based on a convolutional neural network to extract interference pattern fringe features, can effectively capture the complex features in multi-surface interference patterns, improve the accuracy of multi-surface phase shift interference pattern separation, and has strong noise resistance.

[0154] The multi-surface phase recovery system based on convolutional neural network provided in the embodiment of the present application can achieve Figures 1 to 9 To avoid repetition, the various processes implemented in the method embodiment are not described here.

[0155] The multi-surface phase recovery system based on convolutional neural network in the embodiment of the present application can execute the multi-surface phase recovery method based on convolutional neural network provided in the embodiment of the present application, and the implementation principle is similar. The actions performed by each module and unit in the multi-surface phase recovery system based on convolutional neural network in each embodiment of the present application correspond to the steps in the multi-surface phase recovery method based on convolutional neural network in each embodiment of the present application. For the detailed functional description of each module of the multi-surface phase recovery system based on convolutional neural network, please refer to the description of the corresponding multi-surface phase recovery method based on convolutional neural network shown in the previous text, which will not be repeated here.

[0156] Based on the same principle as the method shown in the embodiment of the present application, the embodiment of the present application also provides an electronic device, which may include but is not limited to: a processor and a memory; a memory for storing a computer program; a processor for executing the multi-surface phase recovery method based on a convolutional neural network shown in any optional embodiment of the present application by calling a computer program. Compared with the prior art, the multi-surface phase recovery method based on a convolutional neural network provided by the present application only needs to collect four frames of multi-surface phase-shift interference patterns, and adopts a more mature and accurate four-step phase-shift phase recovery algorithm, which greatly reduces the wavelength tuning range required for measuring the measured object, and further increases the thickness measurement range of the measured object. The method in the present application is an end-to-end process, which uses a model based on a convolutional neural network to extract interference pattern fringe features, can effectively capture the complex features in the multi-surface interference pattern, improves the accuracy of multi-surface phase-shift interference pattern separation, and has strong anti-noise ability.

[0157] In an optional embodiment, an electronic device is also provided, such as Fig.11 As shown, Fig.11 The electronic device 1100 shown may be a server, including: a processor 1101 and a memory 1103. The processor 1101 and the memory 1103 are connected, such as through a bus 1102. Optionally, the electronic device 1100 may further include a transceiver 1104. It should be noted that in actual applications, the transceiver 1104 is not limited to one, and the structure of the electronic device 1100 does not constitute a limitation on the embodiments of the present application.

[0158] Processor 1101 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It may implement or execute various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 1101 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, and the like.

[0159] The bus 1102 may include a path to transmit information between the above components. The bus 1102 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The bus 1102 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Fig.11 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0160] The memory 1103 can be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disk storage (including compressed optical disk, laser disk, optical disk, digital versatile disk, Blu-ray disk, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited to these.

[0161] The memory 1103 is used to store the application code for executing the solution of the present application, and the execution is controlled by the processor 1101. The processor 1101 is used to execute the application code stored in the memory 1103 to implement the contents shown in the above method embodiment.

[0162] Among them, electronic devices include but are not limited to: mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), and fixed terminals such as digital TVs and desktop computers. Fig.11 The electronic device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0163] The server provided in this application can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The terminal can be a smart phone, tablet computer, laptop computer, desktop computer, smart speaker, smart watch, etc., but is not limited to this. The terminal and the server can be directly or indirectly connected by wired or wireless communication, and this application does not limit this.

[0164] An embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer-readable storage medium is run on a computer, the computer can execute the corresponding content in the aforementioned method embodiment.

[0165] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0166] It should be noted that the computer-readable storage medium mentioned above in the present application can also be a computer-readable signal medium or a combination of a computer-readable storage medium and a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal can take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer readable signal medium may also be any computer readable medium other than a computer readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0167] The computer-readable medium may be included in the electronic device, or may exist independently without being installed in the electronic device.

[0168] The computer-readable medium carries one or more programs. When the one or more programs are executed by the electronic device, the electronic device executes the method shown in the above embodiment.

[0169] According to one aspect of the present application, a computer program product or a computer program is provided, the computer program product or the computer program comprising computer instructions, the computer instructions being stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the multi-surface phase recovery method and system based on a convolutional neural network provided in the above-mentioned various optional implementations.

[0170] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0171] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0172] The modules involved in the embodiments described in this application can be implemented by software or hardware. The name of the module does not limit the module itself in some cases. For example, the model building module can also be described as "a model building module for building a multi-surface phase-shift interference pattern fringe separation model based on a convolutional neural network".

[0173] The above description is only a preferred embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in this application (but not limited to) by each other to form a technical solution.

Claims

1. A multi-surface phase recovery method based on convolutional neural network, characterized in that: The method comprises: Generate four frames of simulated multi-surface phase-shift interference patterns in four different phase states, and a plurality of separated single-surface fringe interference patterns corresponding to each frame of simulated multi-surface phase-shift interference patterns; the plurality of single-surface fringe interference patterns include a front surface fringe interference pattern, a rear surface fringe interference pattern, and a thickness fringe interference pattern; Construct a multi-surface phase-shift interference pattern fringe separation model based on convolutional neural network; Using four frames of simulated multi-surface phase-shift interferograms as input data and corresponding separated multiple single-surface fringe interferograms as labels, training the multi-surface phase-shift interferogram fringe separation model to obtain an optimal separation model; Using the optimal separation model, four frames of multi-surface phase-shift interferograms in four different phase states are subjected to fringe separation to obtain a plurality of single-surface fringe prediction interferograms in corresponding phase states; the plurality of single-surface fringe prediction interferograms include a front surface fringe prediction interferogram, a rear surface fringe prediction interferogram, and a thickness fringe prediction interferogram; Based on multiple single-surface fringe prediction interferograms under four different phase states, the four-step phase shift method is used to recover the phase distribution results of the corresponding single surface.

2. The multi-surface phase recovery method based on convolutional neural network according to claim 1, characterized in that: The multi-surface phase-shift interference pattern fringe separation model includes a down-sampling module, an up-sampling module and an output layer; The downsampling module is used to extract multiple single surface stripe features corresponding to the input data by reducing the size of the feature map; the multiple single surface stripe features include front surface stripe features, rear surface stripe features and thickness stripe features; The up-sampling module comprises a plurality of independent up-sampling parts, and the up-sampling parts are used to restore the corresponding single surface stripe features to the size of the input data; wherein the plurality of up-sampling parts correspond one-to-one to the plurality of single surface stripe features; There are multiple output layers, which correspond to the up-sampling parts one by one, and are used to convert the channel features of the output of the corresponding up-sampling part into the corresponding separated single surface fringe prediction interference map.

3. The multi-surface phase recovery method based on convolutional neural network according to claim 2, characterized in that: The down-sampling module comprises a first down-sampling part and a second down-sampling part; The first downsampling part includes a plurality of convolutional layers; the first sampling part is used to reduce the size of the feature map by a step size specified by the convolutional layer; The second down-sampling part includes a plurality of alternating convolutional layers and maximum pooling layers, and the second down-sampling part is used to reduce the size of the feature map by using the maximum pooling layer.

4. The multi-surface phase recovery method based on convolutional neural network according to claim 2, characterized in that: The method uses four frames of simulated multi-surface phase-shift interferograms as input data and the corresponding separated multiple single-surface fringe interferograms as labels to train the multi-surface phase-shift interferogram fringe separation model to obtain the optimal separation model, including: Acquire multiple sets of input data and corresponding labels; each set of input data includes four frames of simulated multi-surface phase-shift interferograms; Inputting multiple sets of input data into the multi-surface phase-shift interference pattern fringe separation model to obtain multiple sets of prediction results; each set of prediction results includes multiple single-surface fringe prediction interference patterns; Use the loss function to calculate the difference between each set of prediction results and the corresponding labels, and calculate the gradient of the loss function to the model; The parameters of the multi-surface phase-shift interference pattern fringe separation model are determined based on the gradient, and the optimal separation model is obtained when a preset number of iterations is reached.

5. The multi-surface phase recovery method based on convolutional neural network according to claim 4, characterized in that: The expression of the loss function is: L=L1+L2+L3 Among them, L1 is the front surface fringe loss function, which is used to calculate the difference between the front surface fringe predicted interference pattern and the corresponding label, L2 is the back surface fringe loss function, which is used to calculate the difference between the back surface fringe predicted interference pattern and the corresponding label, and L3 is the thickness fringe loss function, which is used to calculate the difference between the thickness fringe predicted interference pattern and the corresponding label.

6. The multi-surface phase recovery method based on convolutional neural network according to claim 5, characterized in that: The front surface fringe loss function, the rear surface fringe loss function and the thickness fringe loss function are all mean absolute error functions.

7. The multi-surface phase recovery method based on convolutional neural network according to claim 4, characterized in that: The step of inputting multiple sets of input data into the multi-surface phase-shift interference pattern fringe separation model to obtain multiple sets of prediction results includes: The plurality of input data are shuffled through DataLoader, the number of data transmitted in a single time is increased, and then input into the multi-surface phase-shift interferogram fringe separation model; Multiple groups of prediction results are obtained through the multi-surface phase-shift interference pattern fringe separation model prediction.

8. A multi-surface phase retrieval system based on convolutional neural network, characterized in that: The system comprises: A data generation module, used to generate four frames of simulated multi-surface phase-shift interference patterns in four different phase states, and a plurality of separated single-surface fringe interference patterns corresponding to each frame of simulated multi-surface phase-shift interference patterns; the plurality of single-surface fringe interference patterns include a front surface fringe interference pattern, a rear surface fringe interference pattern, and a thickness fringe interference pattern; Model building module, used to build a multi-surface phase-shift interferogram fringe separation model based on convolutional neural network; A model optimization module is used to use four frames of simulated multi-surface phase-shift interferograms as input data and the corresponding separated multiple single-surface fringe interferograms as labels to train the multi-surface phase-shift interferogram fringe separation model to obtain an optimal separation model; A separation prediction module is used to perform fringe separation on four frames of multi-surface phase-shift interference images in four different phase states using an optimal separation model to obtain a plurality of single-surface fringe prediction interference images in corresponding phase states; the plurality of single-surface fringe prediction interference images include a front surface fringe prediction interference image, a rear surface fringe prediction interference image, and a thickness fringe prediction interference image; The phase recovery module is used to predict the interference pattern based on multiple single-surface fringes in four different phase states, and to recover the phase distribution result of the corresponding single surface using the four-step phase shift method.

9. An electronic device, characterized in that: The invention comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images