Method and apparatus for three-dimensional reconstruction of a double-layer transparent object

Abstract

The application provides a kind of double-layer transparent object three-dimensional reconstruction method and device, comprising: obtaining the double-frequency phase shift reflection fringe image formed by the reflection of the object to be measured;The double-frequency phase shift reflection fringe image is preprocessed, and the first phase initial value of the front surface of the object to be measured and the second phase initial value of the back surface of the object to be measured are obtained respectively;Based on the first phase initial value, the second phase initial value and the first modulation degree initial value and the second modulation degree initial value set in advance, the modulation degree and phase of the front and back surfaces of the object to be measured are globally optimized by using pattern reconstruction method, and the phase solution result of the front and back surfaces of the object to be measured is obtained;Based on the phase solution result, the object to be measured is three-dimensionally reconstructed.The application can realize the phase decoupling of the front and back surfaces of the object to be measured, has the characteristics of high robustness, accuracy and fast decoupling, can be used for high-precision, high-speed and high-dynamic-range double-layer lens deflection meter system, and has far-reaching significance in the field of precision freeform surface processing and detection.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of precision metrology surface shape detection, and in particular to a three-dimensional reconstruction method and device for a double-layer transparent object. BACKGROUND

[0002] The traditional method faces three main challenges in detecting the surface shape of a transparent object: 1) the sparseness of the texture or features on the surface hinders the effectiveness of texture-based visual detection methods; 2) the non-lambertian nature of the surface of the transparent object is not an ideal lambertian model, and cannot be detected using the general method for diffuse reflection materials; 3) the complex interaction between the incident light and the transparent object, such as reflection, refraction, scattering and absorption, exacerbates the complexity of the light path analysis of the transparent object.

[0003] At present, phase measurement deflection can be used to complete the reconstruction of a transparent object that also has a specular reflection characteristic.

[0004] However, when phase measurement deflection is used to reconstruct a transparent object, the front and back surfaces of the transparent object will reflect light from the display screen, causing the camera to capture stripes that are overlapped, and the superposition of the light intensity of the stripes is fatal to pixel matching, so the above method cannot complete the reconstruction of the front and back surfaces at the same time, and the traditional phase demodulation method used by phase measurement deflection often faces problems such as phase overlap, phase order jump and slow demodulation speed of the front and back surfaces. SUMMARY

[0005] The present application provides a three-dimensional reconstruction method and device for a double-layer transparent object, which solves the problems of stripe overlap of the front and back surfaces and defects such as phase overlap, phase order jump and slow demodulation speed of the front and back surfaces when the traditional phase demodulation method is used in the phase measurement deflection, and realizes high-precision three-dimensional reconstruction of the double-layer transparent object.

[0006] The present application provides a three-dimensional reconstruction method for a double-layer transparent object, which is applied to a transparent object deflection meter system, the system comprising a display screen, a to-be-measured object and a camera installed in sequence, wherein the to-be-measured object is a double-layer transparent object; the display screen is used to display a pre-generated double-frequency phase shift initial fringe image and project the double-frequency phase shift initial fringe image on the to-be-measured object; the camera is used to capture a double-frequency phase shift reflection fringe image formed by the double-frequency phase shift initial fringe image reflected by the front surface and the back surface of the to-be-measured object, wherein the front surface and the back surface respectively represent the surface of the to-be-measured object that first contacts the projected light and the surface of the to-be-measured object that later contacts the projected light.

[0007] The three-dimensional reconstruction method for a double-layer transparent object comprises the following steps.

[0008] acquire a double-frequency phase shift initial fringe image reflected by a to-be-measured object to form a double-frequency phase shift reflected fringe image;

[0009] perform preprocessing on the double-frequency phase shift reflected fringe image to obtain a first phase initial value of a front surface of the to-be-measured object and a second phase initial value of a rear surface of the to-be-measured object respectively;

[0010] based on the first phase initial value, the second phase initial value, a first modulation initial value of the front surface of the to-be-measured object and a second modulation initial value of the rear surface of the to-be-measured object set in advance, and based on a mode reconstruction method, globally optimize modulation and phase of the front surface and modulation and phase of the rear surface of the to-be-measured object to obtain modulation calculation results and phase calculation results of the front surface and the rear surface respectively;

[0011] based on the modulation calculation results and the phase calculation results, perform three-dimensional reconstruction on the to-be-measured object.

[0012] According to the three-dimensional reconstruction method of the double-layer transparent object, the double-frequency phase shift initial fringe image reflected by the front surface and the rear surface of the to-be-measured object is continuously captured by a camera to generate the double-frequency phase shift reflected fringe image;

[0013] the light intensity distribution I of the reflected fringe captured by the nth camera n (x, y) is represented as:

[0014] ,

[0015] wherein, is the ambient background light, and are modulation intensities of the front surface and the rear surface respectively, and represent phases of the front surface and the rear surface respectively, , , represents the phase shift amount of the display screen, represents the total phase shift amount, represents the frequency ratio in the case of multi-frequency fringe.

[0016] According to the three-dimensional reconstruction method of the double-layer transparent object, the step of globally optimizing modulation and phase of the front surface and modulation and phase of the rear surface of the to-be-measured double-layer transparent object comprises:

[0017] setting the modulation of the front surface of the to-be-measured double-layer transparent object and the phase of the front surface of the to-be-measured double-layer transparent object , the modulation of the rear surface of the to-be-measured double-layer transparent object and the phase of the rear surface of the to-be-measured double-layer transparent object as initial iteration parameters;

[0018] The initial iteration parameters are decomposed by using a mode reconstruction method to obtain decomposition parameters J; 、 、 and

[0019] Set represents a residual error, and when i is respectively 1, 2, 3 and 4, is respectively represented by 、 、 and , and a global optimization objective function is set as :

[0020] ,

[0021] wherein is a parameter representation of a front surface modulation degree of the object to be measured, is a parameter representation of a rear surface modulation degree of the object to be measured, is a parameter representation of a front surface phase of the object to be measured, is a parameter representation of a rear surface phase of the object to be measured;

[0022] Based on the first phase initial value, the second phase initial value, and the first modulation degree initial value of the front surface of the double-layer transparent object to be measured and the second modulation degree initial value of the rear surface of the double-layer transparent object to be measured which are set in advance, the global optimization objective function is iteratively solved until a preset iteration number is reached or a preset step threshold is reached.

[0023] The front surface modulation degree , the rear surface modulation degree , the front surface phase and the rear surface phase outputted by solving are outputted.

[0024] According to the three-dimensional reconstruction method of the double-layer transparent object provided by the application, the step of pre-processing the double-frequency phase shift reflection fringe image specifically comprises:

[0025] Performing fast Fourier transform on the double-frequency phase shift reflection fringe image to generate a Fourier transform image;

[0026] Filtering the Fourier transform image to obtain fringe information of the front and rear surfaces of the object to be measured;

[0027] Based on the fringe information of the front and rear surfaces of the object to be measured, the truncated phases of the front and rear surfaces of the object to be measured are calculated;

[0028] ​Based on the cut-off phase of the front and back surfaces of the object to be measured, the phase order and the unwrapped phase of the front and back surfaces of the object to be measured are calculated.

[0029] The unwrapped phase of the front surface of the object to be measured is taken as the first phase initial value, and the unwrapped phase of the back surface of the object to be measured is taken as the second phase initial value.

[0030] According to the three-dimensional reconstruction method of the double-layer transparent object provided by the application, the step of filtering the Fourier transform image specifically comprises:

[0031] The log spectrum of the Fourier transform image is calculated, and the calculation formula is:

[0032] ;

[0033] Wherein, respectively represent the coordinates of the pixel points in the x and y directions; represent the Fourier transform image, represent the result of taking the logarithm of the Fourier transform image;

[0034] The position coordinates of the center peak on the frequency spectrum plane formed by the log spectrum are obtained.

[0035] Based on the position coordinates of the center peak, the fringe information of the front and back surfaces of the object to be measured is separated.

[0036] According to the three-dimensional reconstruction method of the double-layer transparent object provided by the application, based on the fringe information of the front and back surfaces of the object to be measured, the cut-off phase of the front and back surfaces of the object to be measured is calculated by using the four-step phase shift method.

[0037] According to the three-dimensional reconstruction method of the double-layer transparent object provided by the application, based on the cut-off phase of the front and back surfaces of the object to be measured, the phase order and the unwrapped phase of the front and back surfaces of the object to be measured are calculated.

[0038] Based on the cut-off phase of the front and back surfaces of the object to be measured, the phase order of the front and back surfaces of the object to be measured is calculated.

[0039] Based on the cut-off phase and the phase order of the front and back surfaces of the object to be measured, the unwrapped phase of the front and back surfaces of the object to be measured is calculated by using the time phase unwrapping technology of the double-frequency fringe.

[0040] According to the three-dimensional reconstruction method of the double-layer transparent object provided by the application, based on the modulation solution result and the phase solution result, the front surface of the object to be measured is reconstructed by using the method based on the pre-known surface shape iteration, and the back surface of the object to be measured is reconstructed by using the method based on the light beam tracing iteration reflection point. ​

[0041] The application further provides a three-dimensional reconstruction device for a double-layer transparent object, which is applied to a transparent object deflection meter system, and the system comprises a display screen, a to-be-measured object and a camera which are sequentially arranged, wherein the to-be-measured object is a double-layer transparent object; the display screen is used for displaying a pre-generated double-frequency phase shift initial fringe image and projecting the double-frequency phase shift initial fringe image on the to-be-measured object; and the camera is used for capturing a double-frequency phase shift reflection fringe image formed by reflection of the double-frequency phase shift initial fringe image on the front surface and the rear surface of the to-be-measured object, wherein the front surface and the rear surface respectively represent a to-be-measured object surface first contacted with the projected light and a to-be-measured object surface last contacted with the projected light.

[0042] The three-dimensional reconstruction device for the double-layer transparent object comprises the following modules:

[0043] A reflection fringe image acquisition module is configured to acquire a double-frequency phase shift reflection fringe image formed by reflection of a double-frequency phase shift initial fringe image on the to-be-measured object.

[0044] A preprocessing module is configured to pre-process the double-frequency phase shift reflection fringe image to obtain a first phase initial value of the front surface of the to-be-measured object and a second phase initial value of the rear surface of the to-be-measured object.

[0045] A global optimization module is configured to perform global optimization on the modulation and phase of the front surface and the modulation and phase of the rear surface of the to-be-measured object based on a pattern reconstruction method based on the first phase initial value, the second phase initial value, a first modulation initial value of the front surface of the to-be-measured object and a second modulation initial value of the rear surface of the to-be-measured object which are pre-set, to obtain modulation calculation results and phase calculation results of the front surface and the rear surface.

[0046] A reconstruction module is configured to perform three-dimensional reconstruction on the to-be-measured object based on the modulation calculation results and the phase calculation results.

[0047] The application further provides an electronic device comprising a memory, a processor and a computer program stored in the memory and capable of running on the processor, and the processor implements the three-dimensional reconstruction method for the double-layer transparent object according to any one of the above-mentioned methods when executing the computer program.

[0048] The application further provides a non-transitory computer readable storage medium having a computer program stored thereon, and the computer program is executed by a processor to implement the three-dimensional reconstruction method for the double-layer transparent object according to any one of the above-mentioned methods.

[0049] The application further provides a computer program product comprising a computer program, and the computer program is executed by a processor to implement the three-dimensional reconstruction method for the double-layer transparent object according to any one of the above-mentioned methods.

[0050] The application provides a three-dimensional reconstruction method and device for a double-layer transparent object. BRIEF DESCRIPTION OF DRAWINGS

[0051] In order to more clearly illustrate the technical solutions in the application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the application, and other drawings can be obtained by those skilled in the art without any creative effort on the basis of these drawings.

[0052] Figure 1 FIG. 1 is a flowchart of the three-dimensional reconstruction method for the double-layer transparent object provided by the application.

[0053] Figure 2 FIG. 2 is a structural diagram of the transparent object refractometer system provided by the application.

[0054] Figure 3 FIG. 3 is a flowchart of the preprocessing of the double-frequency phase shift reflection fringe image provided by the application.

[0055] Figure 4 FIG. 4 is a flowchart of the global optimization of the modulation and phase provided by the application.

[0056] Figure 5It is the reconstruction result of the plano-convex lens by the three-dimensional reconstruction method of the double-layer transparent object provided by the application.

[0057] Figure 6 It is the reconstruction result of the astigmatic lens by the three-dimensional reconstruction method of the double-layer transparent object provided by the application.

[0058] Figure 7 It is the structural schematic diagram of the three-dimensional reconstruction device of the double-layer transparent object provided by the application.

[0059] Figure 8 It is the structural schematic diagram of the electronic device provided by the application. DETAILED DESCRIPTION

[0060] In order to make the objects, technical solutions and advantages of the present application clearer, the technical solutions in the present application will be described clearly and completely below in combination with the drawings in the present application. Obviously, the described embodiments are a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all the other embodiments obtained by those skilled in the art without creative labor fall within the protection scope of the present application.

[0061] The present application will be described in detail below in combination with the drawings in the specification. The specific operation method in the method embodiment can also be applied to the device embodiment or the system embodiment. In the description of the present application, unless otherwise specified, "at least one" includes one or more. "Multiple" refers to two or more. For example, at least one of A, B and C includes: A alone, B alone, A and B together, A and C together, B and C together, and A, B and C together. In the present application, " / " represents the meaning of or, for example, A / B can represent A or B; "and / or" in the present application only describes the association relationship of the associated objects, which means that there can be three kinds of relationships, for example, A and / or B can represent: A alone, A and B together, B alone, and the three cases.

[0062] Transparent objects have a wide range of applications in various fields such as optical elements, virtual reality devices and glasses, heads-up displays (HUD) and windshields in automobiles, screen protection glasses for 3C devices, etc. The accuracy of their three-dimensional morphology significantly affects the accuracy of the application system, the clarity of the image, the durability and rigidity. Therefore, it is crucial to evaluate the surface morphology on site during the manufacturing process to improve the production standards. However, there are three main challenges in detecting the surface morphology of transparent objects: 1) the sparseness of textures or features on the surface hinders the effectiveness of texture-based visual detection methods, 2) the non-lambertian nature of the surface of transparent objects cannot be detected using the general method for diffuse materials, and 3) the complex interactions between incident light and transparent objects, such as reflection, refraction, scattering and absorption, exacerbate the complexity of the analysis of the light path of transparent objects.

[0063] Deflectometry has developed rapidly in recent years, with characteristics such as simple system, wide dynamic range, fast speed, high precision, and easy implementation of in-situ measurement of arbitrary free-form surfaces. PMD systems are based on Snell's law of reflection and typically consist of a screen (usually a liquid crystal display or LCD), a camera, and the surface of the object to be measured. The structured light pattern displayed on the LCD is reflected onto the surface of the object to be measured and then captured by the camera. By matching the corresponding pixels on the screen and the camera, the gradient of the surface to be measured is obtained through ray tracing. Finally, the three-dimensional shape of any free-form surface can be reconstructed by integration. Phase measurement deflectometry can also be used to reconstruct transparent objects with specular reflection characteristics, but it faces the problem of front and back surface stripe aliasing: both the front and back surfaces reflect light from the display screen, resulting in captured stripes that are aliased, and the superposition of stripe light intensity is fatal to pixel matching, which will result in the inability to separately reconstruct the front and back surfaces. At the same time, traditional phase demodulation methods often face problems such as phase aliasing, phase level jump, and slow demodulation speed. Phase aliasing refers to the situation where the front and back surface phase demodulation alternates when the front and back surfaces are close together. Phase level jump: due to the periodic relationship between phase and intensity, local optimal rather than global optimal solutions are easily produced during phase demodulation, leading to phase level jump. Slow demodulation speed: the pixel-by-pixel demodulation method results in slow demodulation speed.

[0064] Phase Measuring Deflectometry (PMD) is a triangulation-based 3D reconstruction technique for specular objects, which has been widely applied in microelectronics, automotive manufacturing, aerospace, and other fields. By simply capturing the deformed fringe pattern reflected by the measured specular surface, the integral reconstruction can be used to recover the arbitrary free-form specular surface. However, transparent objects such as lenses, screens and glasses cause the challenge of phase decoupling due to the front and back surface reflection, which leads to the overlapping of the fringe pattern. The traditional method relies on the nonlinear equation of each pixel, which has the problems of phase ambiguity, phase step and low demodulation efficiency, resulting in poor robustness, high dependence on initial conditions and low efficiency.

[0065] To solve the above problems, the present application provides a double-layer transparent object three-dimensional reconstruction method and device, which is applied to a transparent object deflectometry system, fully utilizes the physical space continuity of the front and back surface modulation and phase of the measured object, decouples the mixed fringe of the double-layer transparent object through a global optimization iterative algorithm, avoids the problems of phase ambiguity, phase step and low calculation efficiency, provides an effective solution for high-precision three-dimensional reconstruction of transparent objects, and has important industrial application value.

[0066] The present application will be described in detail below with reference to specific embodiments.

[0067] In some embodiments of the present application, as shown in Figure 1 The present application provides a double-layer transparent object three-dimensional reconstruction method, which is applied to a transparent object deflectometry system. The system includes a display screen, a measured object and a camera installed in sequence. The measured object is a double-layer transparent object. The display screen is used to display a pre-generated double-frequency phase shift initial fringe image and project the double-frequency phase shift initial fringe image on the measured object. The camera is used to capture a double-frequency phase shift reflection fringe image formed by the double-frequency phase shift initial fringe image reflected by the front and back surfaces of the measured object. The front surface and the back surface represent the measured object surface first contacted with the projected light and the measured object surface contacted with the projected light, respectively.

[0068] The method includes the following steps:

[0069] Step 100: obtaining a double-frequency phase shift reflection fringe image formed by the double-frequency phase shift initial fringe image reflected by the measured object;

[0070] Step 200: pre-processing the double-frequency phase shift reflection fringe image to obtain a first phase initial value of the front surface of the measured object and a second phase initial value of the back surface of the measured object, respectively.

[0071] Step 300, based on the first phase initial value, the second phase initial value and the first modulation degree initial value of the front surface of the object to be measured and the second modulation degree initial value of the rear surface of the object to be measured, the modulation degree and phase of the front surface and the modulation degree and phase of the rear surface of the object to be measured are globally optimized based on a pattern reconstruction method, and the modulation degree solution and the phase solution of the front surface and the rear surface are obtained respectively.

[0072] Step 400, based on the modulation degree solution and the phase solution, three-dimensional reconstruction is performed on the object to be measured.

[0073] It should be noted that the existing three-dimensional reconstruction scheme relies on a pixel-by-pixel nonlinear equation, which is difficult to decouple overlapping stripes, and has problems of phase ambiguity, phase step and low demodulation efficiency.

[0074] Therefore, the present application obtains the phase initial value of the front and rear surfaces of the object to be measured by preprocessing the double-frequency phase shift reflection stripe image formed by the reflection of the object to be measured, globally optimizes the phase and modulation degree of the front and rear surfaces of the object to be measured based on a pattern reconstruction method according to the phase initial value of the front and rear surfaces of the object to be measured and the preset modulation degree initial value, and performs three-dimensional reconstruction on the object to be measured according to the optimization result. The method is a high-robustness, high-precision and fast double-layer transparent object phase demodulation method based on a double-layer transparent object, which realizes phase measurement deflectometry reconstruction of the transparent object, and has very important significance for the field of free-form surface precision machining and detection.

[0075] It can be understood that the existing pixel-by-pixel solution ignores the continuity of the phase between the pixels, while in the actual physical world, the front and rear surfaces of the transparent object are mostly continuous, so the corresponding phase is also continuous, and the phase of the front and rear surfaces can be solved globally.

[0076] In some possible embodiments of the present application, as shown in Figure 2 The transparent object deflectometry system 10 includes a display screen 101, a measured mirror 102 and a camera 103 installed in sequence, and the measured mirror is a double-layer transparent object. The overall idea of the present application is that the display screen 101 generates a double-frequency phase shift initial stripe image, which is reflected on the front and rear surfaces of the measured mirror 102, and the double-frequency phase shift reflection stripe 104 is captured by the camera 103.

[0077] Further, as shown in Figure 3 In this embodiment, it is assumed that the camera captures 8 reflection stripe images, and the 8 double-frequency phase shift reflection stripe images 201 captured by the camera are preprocessed 20, which includes spatial frequency filtering 202, truncated phase extraction 203 and unfolded phase acquisition 204 to obtain the initial value of the front and rear surface phase.

[0078] On this basis, the global optimization method 30 based on pattern reconstruction is reused, as shown in Figure 3 The modulation and phase of the front and back surfaces are globally optimized; finally, the phase matching of the display screen and the camera is realized, and the three-dimensional reconstruction 40 of the double-layer surface is realized through light path tracing. Figure 4

[0079] Specifically, the transparent object deflection meter system 10 also includes a computer, and the two different frequency phase shift fringes in the initial double-frequency phase shift fringe image are generated by the computer, displayed by the display screen, deformed after being reflected by the to-be-measured mirror, and captured by the camera to obtain the deformed double-frequency phase shift fringe.

[0080] In possible embodiments, a standard phase measurement deflection system mainly includes a camera (c), a to-be-measured mirror (m), and a display screen (s). The display screen displays sinusoidal variation grating fringes with phase shift, and the fringe image is deformed after being reflected by the to-be-measured mirror, and then captured by the camera. The deformed fringe pattern captured by the camera contains the topographic information of the surface of the to-be-measured object. After phase demodulation and phase order extraction, the unwrapped phase is obtained, and then combined with the spatial position relationship of the camera, the display screen, and the to-be-measured mirror provided by system calibration. The object surface normal vector is obtained by inverse ray tracing, and then the to-be-measured object surface gradient is calculated, as shown in the formula. Finally, the three-dimensional surface profile of the to-be-measured object is reconstructed.

[0081] (1),

[0082] (2),

[0083] wherein, respectively represent the corresponding points of the to-be-measured mirror, the camera, and the display screen in ray tracing, , respectively represent the distances from M to P and M to Q .

[0084] In some possible embodiments of the present application, the double-frequency phase shift initial fringe image reflected by the front surface and the back surface of the to-be-measured object is captured by the camera to generate the double-frequency phase shift reflected fringe image; the light intensity distribution I n (x,y) of the reflected fringe captured by the camera for the nth time is represented as:

[0085] (3),

[0086] wherein, A(x, y) is the environmental background light, B f (x,y) and Br(x,y) are the modulation intensities of the front surface and the back surface, respectively,​ and φfand φbdenote the phase of the front and back surfaces respectively, , n=0,1,2…N-1 represent the phase shift of the display screen, N represents the total phase shift, and k represents the frequency ratio in the case of multi-frequency fringes.

[0087] Specifically, the embodiment provides an implementation of capturing a double-frequency phase shift reflection fringe image by a camera, and the light intensity distribution of the reflection fringe captured by the camera each time is constructed.

[0088] Specifically, for a transparent object such as a spectacle lens, an optical lens or the like, the front and back surfaces will both reflect, and the same camera pixel point C corresponds to two different display screen pixel points S1 and S2; this indicates that the phases corresponding to the front and back surfaces are different, and the simultaneous capture by the camera will cause the fringe intensity to be superimposed; in formula (3), I n (x,y) represents the light intensity distribution captured by the camera at the nth phase shift, A(x, y) is the environmental background light, B f (x, y) and Br(x, y) are the modulation intensities of the front and back surfaces respectively, and φfand φbdenote the phase of the front and back surfaces respectively, , n=0,1,2…N-1 represent the phase shift of the display screen, N represents the total phase shift, and k represents the frequency ratio in the case of multi-frequency fringes.

[0089] In a specific embodiment, as shown in Figure 3 if N=4 and k=k0, k1, 8 images of 4x2 of 201 are generated by formula (3).

[0090] In some possible embodiments of the present application, the step of preprocessing the double-frequency phase shift reflection fringe image specifically comprises:

[0091] Performing fast Fourier transform on the double-frequency phase shift reflection fringe image to generate a Fourier transform image;

[0092] Filtering the Fourier transform image to obtain fringe information of the front and back surfaces of the object to be measured;

[0093] Based on the fringe information of the front and back surfaces of the object to be measured, the truncated phases of the front and back surfaces of the object to be measured are calculated.

[0094] Based on the truncated phases of the front and back surfaces of the object to be measured, the phase order and the unwrapped phase of the front and back surfaces of the object to be measured are calculated.

[0095] The unwrapped phase of the front surface of the object to be measured is taken as the first phase initial value, and the unwrapped phase of the back surface of the object to be measured is taken as the second phase initial value.

[0096] Specifically, the embodiment provides an implementation of pre-processing of a dual-frequency phase shift reflection fringe image, as shown in the figure. Figure 3 For a dual-layer transparent object such as a lens, a picture captured by a camera is as shown in 201; a fast Fourier transform (FFT) is respectively performed on the image to generate a Fourier transform image, as shown in 202; the fringe density and direction of the front and back surfaces of the object to be measured can be determined according to the shape of the front and back surfaces; two groups of center-symmetric peaks are respectively taken as filters, so that the fringe information of the front and back surfaces can be separated, as shown in 203; after filtering, the truncated phase of the front and back surfaces can be respectively calculated by using a four-step phase shift method; then, a time phase unwrapping technology based on dual-frequency fringes is used to calculate the phase order and unwrapped phase of the front and back surfaces, as shown in 204.

[0097] In some possible embodiments of the present application, the step of filtering the Fourier transform image specifically comprises:

[0098] The log spectrum of the Fourier transform image is calculated, and the calculation formula is:

[0099] (4),

[0100] wherein, x and y respectively represent the coordinates of a pixel point in x and y directions; represents a Fourier transform image, represents the result of taking a logarithm of the Fourier transform image;

[0101] The position coordinates of a center peak on a frequency spectrum plane formed by the log spectrum are obtained;

[0102] Based on the position coordinates of the center peak, the fringe information of the front and back surfaces of the object to be measured is separated.

[0103] Specifically, the embodiment provides an implementation of filtering a Fourier transform image obtained by performing a Fourier transform. For a dual-layer transparent object, a camera will capture superimposed fringes after reflection of the front and back surfaces 201. In order to decouple the phases of the front and back surfaces, a phase initial value pre-processing method is adopted, still referring to Figure 3 a fast Fourier transform (FFT) is respectively performed on the image to generate a Fourier transform image, and then a log spectrum of the Fourier transform image is calculated , to improve the dynamic range and visualization effect, as shown in 202; due to the different surface shapes of the front and back surfaces, the spatial modulation of the sinusoidal fringes is also different, so that a central peak representing the ambient light part and two sets of central symmetric peaks representing the fringes modulated by the front and back surfaces can be seen on the spectral plane. According to the shape of the side surface, the fringe density and direction of the front and back surfaces can be determined, and the fringe information of the front and back surfaces can be separated by taking the two sets of central symmetric peaks as filters, as shown in 203.

[0104] In some possible embodiments of the present application, the truncated phases of the front and back surfaces of the object to be measured are calculated based on the fringe information of the front and back surfaces of the object to be measured by using a four-step phase shift method.

[0105] Specifically, the embodiment provides an implementation for calculating the truncated phases of the front and back surfaces of the object to be measured, and the truncated phases of the front and back surfaces can be calculated by using a four-step phase shift method after filtering.

[0106] In some possible embodiments of the present application, the phase orders and unwrapped phases of the front and back surfaces of the object to be measured are calculated based on the truncated phases of the front and back surfaces of the object to be measured.

[0107] The phase orders of the front and back surfaces of the object to be measured are calculated based on the truncated phases of the front and back surfaces of the object to be measured.

[0108] The unwrapped phases of the front and back surfaces of the object to be measured are calculated based on the truncated phases and the phase orders of the front and back surfaces of the object to be measured by using a time phase unwrapping technology based on double-frequency fringes.

[0109] Specifically, the embodiment provides an implementation for calculating the phase orders and unwrapped phases of the front and back surfaces of the object to be measured, and the phase orders and unwrapped phases of the front and back surfaces are calculated by using a time phase unwrapping technology based on double-frequency fringes after the truncated phases of the front and back surfaces of the object to be measured are obtained, as shown in 204.

[0110] Specifically, initial phase measurement is performed by using two signals with different frequencies to obtain two corresponding phase maps, and a high-frequency signal and a low-frequency signal are usually selected; the phase difference between the two frequency signals is calculated, and this setting is used to eliminate the 2π ambiguity problem in phase measurement; the phase of the high-frequency signal is unwrapped according to the calculated phase difference information to obtain a high-precision phase value; the phase unwrapping process of the high-frequency signal is guided through the phase information of the low-frequency signal to improve the measurement accuracy; and the unwrapped high-frequency phase map is synthesized into a final high-resolution phase map, so that a high-precision three-dimensional reconstruction result or phase distribution is obtained.

[0111] In possible embodiments, since the filter is set in the form of a Gaussian function, the standard deviation of the Gaussian filter satisfies a minimum value under certain conditions, i.e., a minimum value of the average value of the log spectrum intensity within the range defined by the standard deviation (e.g., less than 0.9 times the peak log spectrum value). This method of dynamic adjustment ensures that the filter adapts to the data characteristics and sets an appropriate standard deviation for each peak position, thereby achieving effective filtering. However, in practice, the peaks corresponding to the front and back surfaces are not Gaussian, but are affected by the direction-specificity of the surface profile and the off-axis nature of the deflection system. This can cause distortion of the fringe information after Gaussian filtering, which in turn affects the phase order jump and the error of the unwrapped phase. As can be seen from 204, there are phase jump errors in some parts.

[0112] To further eliminate the errors caused by filtering and restore the phase information, a global optimization method based on pattern reconstruction is used to accurately reconstruct the front and back surface phases through continuous iteration.

[0113] In some possible embodiments of the present application, the step of globally optimizing the modulation and phase of the front surface and the modulation and phase of the back surface of the double-layer transparent object to be measured specifically includes:

[0114] setting the modulation and phase of the front surface of the double-layer transparent object to be measured as initial iteration parameters; using a pattern reconstruction method to decompose the initial iteration parameters ,

[0115] , and to obtain decomposition parameters J; setting to represent the residual error, and

[0116] taking 1, 2, 3, and 4 respectively, representing i , , and respectively, and setting a global optimization objective function :

[0117] (5)

[0118] wherein is a parameter representation of the modulation of the front surface of the object to be measured, is a parameter representation of the modulation of the back surface of the object to be measured, is a parameter representation of the phase of the front surface of the object to be measured, and ​​a parameter representation of the phase of the rear surface of the object to be measured;

[0119] Based on the first phase initial value, the second phase initial value, and the first modulation degree initial value of the front surface of the double-layer transparent object to be measured and the second modulation degree initial value of the rear surface of the double-layer transparent object to be measured, the global optimization objective function is iteratively solved until a preset iteration number is reached or a preset step threshold is reached.

[0120] output the calculated front surface modulation degree , rear surface modulation degree , front surface phase , and rear surface phase .

[0121] Specifically, the embodiment provides an implementation of global optimization of the modulation degree and phase of the front and rear surfaces of a double-layer transparent object to be measured. In order to further eliminate the error caused by filtering and restore the phase information, a global optimization method based on pattern reconstruction is used, and the unwrapped phase with error is taken as the initial value of the global optimization algorithm based on pattern reconstruction J The model minimizes the objective function through iterative calculation, ensuring accurate reconstruction of the front and rear surface phases through continuous iteration.

[0122] In possible embodiments, as shown in Figure 4 , the specific process of the global optimization method based on pattern reconstruction is as follows:

[0123] First, the initial iteration parameters are selected as the modulation degree and phase of the front and rear surfaces , , and ; the pattern reconstruction method is used to decompose the four parameters to obtain , which converts the surface information into a parameter set, including modulation and phase parameters related to the front and rear surfaces of the object; the iteration parameters are subjected to nonlinear global optimization according to the global optimization objective function ; whether to exit is determined by judging the iteration stopping condition: k > k max or If any of the conditions is met, the iteration process is terminated and the next step is entered; otherwise, the next iteration is returned; finally, the iteration convergence is restored as the optimized modulation degree and phase , , and . The final iteration parameters will be used for surface reconstruction of the transparent object, ensuring high accuracy, high speed, and high dynamic range of measurement.

[0124] Further, according to the phase information, the light emitted by the same camera pixel point can be made to pass through the front and back surface reflection and match with two pixel points on the display screen. In this way, the shape and position of the front surface can be determined by the pre-known surface iteration method, and the shape and position of the back surface need to be corrected for the light refracted into the medium inside. The back surface surface iteration is converged by optimizing the light tracing path.

[0125] In possible embodiments, the wavefront reconstruction is performed by mode reconstruction technology, which can use a set of orthogonal modes (such as Zernike polynomials or other suitable basis functions) to represent the wavefront surface, and determine the coefficients of each mode one by one to approximate or reconstruct the complex wavefront. Generally, least squares method or other optimization algorithms are used to fit these modes based on known measurement data, so as to obtain the coefficients of each mode. The combination of these coefficients can reconstruct the shape of the original wavefront.

[0126] It can be understood that the four unknowns of the modulation degree and phase of the front surface and the back surface of the object to be measured have continuity in the x and y directions, and the modal wavefront reconstruction is feasible. Therefore, the embodiment proposes a global optimization method based on modal wavefront reconstruction. This technology uses a series of mutually orthogonal polynomials to simulate the modulation intensity and phase of the front and back surfaces. Generally, the selection of polynomials (such as Legendre polynomials, Chebyshev polynomials or Zernike polynomials) depends on the specific surface shape of the analysis. Zernike polynomials are usually used for circular regions, and Legendre polynomials are used for rectangular regions. Modal wavefront reconstruction uses fewer parameters to represent the entire surface. This allows the phase and modulation of the entire surface to be optimized in the nonlinear programming (NLP) process, significantly reducing the need to calculate the modulation intensity and phase pixel by pixel. The phase decoupling of the entire front and back surface can be achieved through a single iteration, ensuring phase continuity and improving the robustness of phase reconstruction, while avoiding phase ambiguity and phase order jumps.

[0127] For example, when using Zernike polynomials, the four unknowns 、 、 and are represented as a linear combination of Zernike polynomial terms , where n is from 0 to l, and m varies from -n in steps of 2. The coefficients represent the fitting parameters of these polynomials. The appropriate polynomial order l can be selected according to the complexity of the front and back surface profiles. If l is set to 5, the first 21 terms of are covered, and the operations of other parameters are similar, as shown in equations (6) or (7). Similarly, , 、 and It can also be expressed as a linear combination of the product of Legendre polynomials D(x) and D(y) in the x and y directions, where k represents the order of the Legendre polynomials, as shown in formula (6) or (7).

[0128] (6),

[0129] (7),

[0130] In the subsequent calculation process, this embodiment uses Zernike fitting as an example to illustrate the modeling process. The process of selecting other polynomials for modeling can be analogous to this method.

[0131] On this basis, based on the global optimization method of modal wavefront reconstruction, the objective function can be set as shown in formula (8):

[0132] (8),

[0133] Taking the dual-frequency phase shift method as an example, its objective function is defined as the residual 、 、 and The square sum of the optimization parameters J represents the Zernike fitting parameters , by applying nonlinear least squares optimization, a set of optimal parameters can be determined .

[0134] The above configuration of this embodiment effectively minimizes the discrepancy between the predicted model and actual measurements by treating the entire surface as a whole. This holistic perspective not only improves model accuracy but also addresses fundamental issues such as local phase ambiguity and order jumps. Furthermore, this method significantly improves measurement efficiency, providing a more streamlined and reliable means of phase acquisition.

[0135] In a specific embodiment, Figure 5 As shown, Figure 5 This is the reconstruction result of a plano-convex lens using the double-layer transparent object 3D reconstruction method provided by the present invention, wherein (a) and (b) are the reconstruction results of the front and back surfaces, respectively; (c) and (d) are the residuals after comparison with the standard model, respectively. In order to verify the effectiveness of the above method, a K9 plano-convex lens commonly used in optical experiments was reconstructed. Its focal length is 1000mm, its front surface is a sphere with a curvature radius of 516.82mm, the refractive index of the material is 1.5163, and the reconstruction range of the front surface is 20mm×20mm; while the reconstruction range of the back surface is slightly smaller than the front surface due to ray tracing. Figure 5In the figures, (a), (b) are the front, back surface reconstruction results, the measurement results are compared with the surface profile provided by the manufacturer, as Figure 5 shown in (c), (d).

[0136] In another specific embodiment, as Figure 6 shown, Figure 6 is the reconstruction result of the astigmatic lens provided by the method for three-dimensional reconstruction of double-layer transparent objects. Among them, (a), (b) are the real object diagram and the measurement scatter diagram of the progressive spectacle lens using a three-dimensional profiler, respectively; (c), (d) are the reconstruction results of the front and back surfaces, respectively; (e), (f) are the residual errors after comparison with the standard model. Another measurement result of a spectacle lens with astigmatism proves the universal applicability of the method. Such a lens has different radii of curvature and thickness in different directions, as Figure 6 shown in (a). The shape of the lens surface can be described by an elliptical parabolic equation , where x and y are along the two principal axes, respectively. Its back surface is a spherical surface with a radius of R. Due to the large variation of the curvature of the lens, the dynamic range of the white light interferometer cannot meet the measurement requirements. The surface is sprayed with powder with diffuse reflection characteristics, and then the object is scanned by ATOS capsule 12M to obtain the three-dimensional topography of the surface, as Figure 6 shown in (b), and the measurement data are used as standard values. The deviation of the reconstruction result obtained by the method of the present application from the standard value is shown in (c)-(f) of Figure 6 .

[0137] From the above verification results, it can be proved that the global optimization iterative algorithm based on the pattern reconstruction method can realize the decoupling of the aliasing fringe of the double-layer transparent object. This method fully utilizes the physical space continuity of the front and back surface modulation and phase, avoiding the problems of phase ambiguity, order jump and low solving efficiency caused by pixel-by-pixel phase solving. On this basis, a phase deflectometry system based on this method is built, which can realize the reconstruction of the double-layer surface of any free-form lens. The measurement accuracy can reach the micron level, which is comparable to the three-dimensional measurement profiler. In addition, its measurement dynamic range can be from microns to millimeters, far exceeding the white light interferometer, and it is better at processing large curvature surfaces. There is no need for any pretreatment (powder spraying) and post-processing (cleaning) during measurement, which can realize in-situ detection during industrial production and processing. At the same time, the measurement speed is greatly improved, and the measurement time of a single element can be completed within a few seconds. This method has very important significance for the field of free-form surface precision machining and manufacturing.

[0138] In some possible embodiments of the present application, based on the modulation degree solution and the phase solution, the front surface of the object to be measured is reconstructed by using a pre-known surface shape iteration method, and the rear surface of the object to be measured is reconstructed by using a light beam tracing iteration reflection point method.

[0139] Specifically, the present embodiment provides an implementation of reconstructing the front and rear surfaces of an object to be measured. When selecting a reconstruction method, the shape and characteristics of the front and rear surfaces of the object to be measured are considered, and different reconstruction methods are used for reconstruction.

[0140] The three-dimensional reconstruction method of the double-layer transparent object provided by the present application is essentially a global iteration method based on free-form surface phase modeling. The phase and modulation degree of the front and rear surfaces are constructed by using a pattern reconstruction method, and the fitting parameters are used as optimization parameters for global nonlinear optimization iteration of the entire surface shape. The problems of phase confusion, order jump and low demodulation efficiency can be solved at the same time. By designing an optimization method based on global free-form surface phase modeling, the phase decoupling of the front and rear surfaces can be realized, and the method has the characteristics of high robustness, accuracy and fast decoupling. When the present application is applied to phase deflectometry, three-dimensional reconstruction of a double-layer transparent object can be realized. A double-layer lens deflectometry system with high precision, high speed and high dynamic range has a profound significance for the field of precise free-form surface processing and detection.

[0141] In some specific embodiments of the present application, as shown in Figure 7 The present application provides a three-dimensional reconstruction device of a double-layer transparent object, which is applied to a transparent object deflectometry system. The system comprises a display screen, an object to be measured and a camera which are sequentially installed. The object to be measured is a double-layer transparent object. The display screen is used to display a pre-generated double-frequency phase shift initial fringe image and project the double-frequency phase shift initial fringe image on the object to be measured. The camera is used to capture a double-frequency phase shift reflection fringe image formed by the double-frequency phase shift initial fringe image reflected by the front surface and the rear surface of the object to be measured. The front surface and the rear surface respectively represent the surface of the object to be measured which first contacts the projected light and the surface of the object to be measured which later contacts the projected light.

[0142] The device comprises:

[0143] The reflection fringe image acquisition module 71 is configured to acquire a double-frequency phase shift reflection fringe image formed by the double-frequency phase shift initial fringe image reflected by the object to be measured.

[0144] The preprocessing module 72 is configured to pre-process the double-frequency phase shift reflection fringe image to obtain a first phase initial value of the front surface of the object to be measured and a second phase initial value of the rear surface of the object to be measured.

[0145] The global optimization module 73 is configured to perform global optimization on the modulation and phase of the front surface and the modulation and phase of the rear surface of the object to be measured based on a mode reconstruction method based on the first phase initial value, the second phase initial value, and the first modulation initial value of the front surface of the object to be measured and the second modulation initial value of the rear surface of the object to be measured, and obtain modulation calculation results and phase calculation results of the front surface and the rear surface, respectively.

[0146] The reconstruction module 74 is configured to perform three-dimensional reconstruction on the object to be measured based on the modulation calculation results and the phase calculation results.

[0147] The three-dimensional reconstruction device for a double-layer transparent object provided in the embodiment of the present application has similar implementation principles and beneficial effects to those of the three-dimensional reconstruction method for a double-layer transparent object, and reference can be made to the implementation principles and beneficial effects of the three-dimensional reconstruction method for a double-layer transparent object, which will not be repeated here.

[0148] Figure 8 An example of a schematic diagram of a physical structure of an electronic device is shown in FIG. 8. Figure 8 As shown in FIG. 8, the electronic device can include a processor 810, a communications interface 820, a memory 830, and a communications bus 840, wherein the processor 810, the communications interface 820, and the memory 830 can communicate with each other through the communications bus 840. The processor 810 can invoke a logical instruction in the memory 830 to execute a three-dimensional reconstruction method for a double-layer transparent object, which includes: obtaining a double-frequency phase shift reflection fringe image formed by reflection of an initial double-frequency phase shift fringe image on an object to be measured; performing preprocessing on the double-frequency phase shift reflection fringe image to obtain a first phase initial value of a front surface of the object to be measured and a second phase initial value of a rear surface of the object to be measured; performing global optimization on the modulation and phase of the front surface and the modulation and phase of the rear surface of the object to be measured based on a mode reconstruction method based on the first phase initial value, the second phase initial value, and the first modulation initial value of the front surface of the object to be measured and the second modulation initial value of the rear surface of the object to be measured, and obtaining modulation calculation results and phase calculation results of the front surface and the rear surface, respectively; and performing three-dimensional reconstruction on the object to be measured based on the modulation calculation results and the phase calculation results.

[0149] In addition, the logic instructions in the memory 830 described above can be implemented in the form of software functional units and sold or used as independent products, and can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or part of the technical solutions can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the methods described in various embodiments of the present application. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0150] In another aspect, the present application also provides a computer program product, the computer program product comprising a computer program, the computer program being stored on a non-transitory computer readable storage medium, and the computer program being executable by a processor to cause a computer to perform the method for three-dimensional reconstruction of a double-layer transparent object, the method comprising: obtaining a double-frequency phase shift reflection fringe image formed by reflection of an initial double-frequency phase shift fringe image on an object to be measured; preprocessing the double-frequency phase shift reflection fringe image to obtain a first phase initial value of a front surface of the object to be measured and a second phase initial value of a back surface of the object to be measured; based on the first phase initial value, the second phase initial value, a first modulation initial value of the front surface of the object to be measured, and a second modulation initial value of the back surface of the object to be measured, performing global optimization of modulation and phase of the front surface and modulation and phase of the back surface of the object to be measured based on a pattern reconstruction method to obtain modulation and phase calculation results of the front surface and the back surface, respectively; and performing three-dimensional reconstruction of the object to be measured based on the modulation and phase calculation results.

[0151] In yet another aspect, the present application also provides a non-transitory computer readable storage medium having stored thereon a computer program, which, when executed by a processor, implements the method for three-dimensional reconstruction of a double-layer transparent object provided by any of the above methods, and the method comprises: acquiring a double-frequency phase shift reflection fringe image formed by reflection of an initial double-frequency phase shift fringe image on an object to be measured; pre-processing the double-frequency phase shift reflection fringe image to obtain a first phase initial value of a front surface of the object to be measured and a second phase initial value of a rear surface of the object to be measured; based on the first phase initial value, the second phase initial value, a first modulation initial value of the front surface of the object to be measured, and a second modulation initial value of the rear surface of the object to be measured, globally optimizing the modulation and phase of the front surface and the modulation and phase of the rear surface of the object to be measured based on a pattern reconstruction method to obtain modulation and phase calculation results of the front surface and the rear surface, respectively; and based on the modulation and phase calculation results, performing three-dimensional reconstruction on the object to be measured.

[0152] The device embodiments described above are merely illustrative, wherein the units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, i.e., can be located in one place or distributed on multiple network units. Part or all of the modules can be selected to achieve the purpose of the embodiment according to actual needs. Those skilled in the art can understand and implement without creative labor.

[0153] From the above description of the embodiments, those skilled in the art can clearly understand that the embodiments can be implemented by means of software plus necessary general hardware platforms, and of course can also be implemented by hardware. Based on such understanding, the above technical solutions, essentially or in terms of contribution to the prior art, can be embodied in the form of a software product. The computer software product can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0154] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for some technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for 3D reconstruction of a double-layer transparent object, applied to a transparent object deflectometer system, characterized in that: The system includes a display screen, an object to be measured, and a camera, which are installed in sequence. The object to be measured is a double-layer transparent object. The display screen is used to display a pre-generated dual-frequency phase-shifted initial fringe image and project the dual-frequency phase-shifted initial fringe image onto the object to be measured. The camera is used to capture a dual-frequency phase-shifted reflected fringe image formed by reflecting the dual-frequency phase-shifted initial fringe image from the front surface and the rear surface of the object to be measured. The front surface and the rear surface respectively represent the surface of the object to be measured that first contacts the projected light and the surface of the object to be measured that contacts the projected light later. The method for three-dimensional reconstruction of a double-layer transparent object comprises: Acquire a dual-frequency phase-shifted reflected fringe image formed by reflecting the dual-frequency phase-shifted initial fringe image from the object to be measured; Preprocessing the dual-frequency phase-shifted reflected fringe image to obtain a first initial phase value of the front surface of the object to be measured and a second initial phase value of the rear surface of the object to be measured; Based on the first phase initial value, the second phase initial value, and the preset first modulation index initial value of the front surface of the object to be measured and the preset second modulation index initial value of the rear surface of the object to be measured, the modulation index and phase of the front surface of the object to be measured and the modulation index and phase of the rear surface of the object to be measured are globally optimized based on the pattern reconstruction method to obtain modulation index solution results and phase solution results of the front surface and the rear surface of the object to be measured, respectively; Based on the modulation depth calculation result and the phase calculation result, the object to be measured is reconstructed in three dimensions.

2. The method for 3D reconstruction of a double-layer transparent object according to claim 1, wherein: Continuously capturing, by a camera, the reflected fringes of the dual-frequency phase-shifted initial fringe image reflected by the front surface and the rear surface of the object to be measured, to generate the dual-frequency phase-shifted reflected fringe image; The light intensity distribution of the reflected stripes captured by the camera for the nth time I n (x,y) is represented as: ， in, is the ambient background light, and are the modulation intensities of the front and back surfaces, and represent the phase of the front and back surfaces respectively, , n=0,1,2…N-1, n represents the phase shift of the display screen, N represents the total phase shift, and k represents the frequency ratio in the case of multi-frequency stripes.

3. The method for 3D reconstruction of a double-layer transparent object according to claim 2, wherein: The step of globally optimizing the modulation degree and phase of the front surface and the modulation degree and phase of the rear surface of the object to be measured specifically includes: The modulation degree of the front surface of the object to be measured B f ( x,y ) and phase φ f ( x,y ), modulation of the rear surface B r ( x,y ) and phase φ r ( x,y ) is set as the initial iteration parameter; The initial iteration parameters are reconstructed using the pattern reconstruction method. 、 、 and Decompose to obtain decomposition parameter J; set up Residual error, i When 1, 2, 3, and 4 are taken respectively, Use 、 、 and Indicates that the global optimization objective function is set : ， in, is the parameter representation of the modulation degree of the front surface of the object to be measured, is the parameter representation of the back surface modulation of the object to be measured, is the parameter representation of the phase of the front surface of the object to be measured, is the parameter representation of the phase of the rear surface of the object to be measured; Iteratively solving the global optimization objective function based on the first initial phase value, the second initial phase value, and a preset first modulation index initial value of the front surface of the object to be measured and a preset second modulation index initial value of the rear surface of the object to be measured until a preset number of iterations is reached or a preset step size threshold is reached; Output the calculated front surface modulation , rear surface modulation , front surface phase and rear surface phase .

4. The method for 3D reconstruction of a double-layer transparent object according to any one of claims 1 to 3, wherein: The step of preprocessing the dual-frequency phase-shifted reflection fringe image specifically includes: performing a fast Fourier transform on the dual-frequency phase-shifted reflected fringe image to generate a Fourier transform image; Filtering the Fourier transform image to obtain fringe information on the front and rear surfaces of the object to be measured; Calculating the truncation phases of the front and rear surfaces of the object to be measured based on the fringe information of the front and rear surfaces of the object to be measured; Calculating the phase order and the unwrapped phase of the front and rear surfaces of the object to be measured based on the truncated phases of the front and rear surfaces of the object to be measured; The unwrapped phase of the front surface of the object to be measured is used as a first phase initial value, and the unwrapped phase of the rear surface of the object to be measured is used as a second phase initial value.

5. The method for 3D reconstruction of a double-layer transparent object according to claim 4, characterized in that: The step of filtering the Fourier transform image specifically includes: The logarithmic spectrum of the Fourier transform image is calculated using the following formula: ； in, Represents the pixels at Coordinates of direction; represents the Fourier transform image, Represents the result of taking the logarithm of the Fourier transform image; Obtaining the position coordinates of the central peak on the frequency spectrum plane formed by the logarithmic spectrum; Based on the position coordinates of the central peak, the stripe information on the front and rear surfaces of the object to be measured is separated.

6. The method for 3D reconstruction of a double-layer transparent object according to claim 4, wherein: Based on the fringe information of the front and rear surfaces of the object to be measured, the truncated phases of the front and rear surfaces of the object to be measured are calculated using a four-step phase shift method.

7. The method for 3D reconstruction of a double-layer transparent object according to claim 4, wherein: The step of calculating the phase order and the unwrapped phase of the front and rear surfaces of the object to be measured based on the truncated phases of the front and rear surfaces of the object to be measured specifically includes: Calculating the phase orders of the front and rear surfaces of the object to be measured based on the truncated phases of the front and rear surfaces of the object to be measured; Based on the truncated phases and phase orders of the front and rear surfaces of the object to be measured, the unwrapped phases of the front and rear surfaces of the object to be measured are calculated by using a temporal phase unwrapping technique of dual-frequency fringes.

8. The method for 3D reconstruction of a double-layer transparent object according to claim 1, wherein: Based on the modulation depth solution results and the phase solution results, the front surface of the object to be measured is reconstructed using a method based on predicted surface iteration, and the back surface of the object to be measured is reconstructed using a method based on beam tracing iterative reflection points.

9. A double-layer transparent object 3D reconstruction device, applied to a transparent object deflectometer system, characterized in that: The system includes a display screen, an object to be measured, and a camera, which are installed in sequence. The object to be measured is a double-layer transparent object. The display screen is used to display a pre-generated dual-frequency phase-shifted initial fringe image and project the dual-frequency phase-shifted initial fringe image onto the object to be measured. The camera is used to capture a dual-frequency phase-shifted reflected fringe image formed by reflecting the dual-frequency phase-shifted initial fringe image from the front surface and the rear surface of the object to be measured. The front surface and the rear surface respectively represent the surface of the object to be measured that first contacts the projected light and the surface of the object to be measured that contacts the projected light later. The double-layer transparent object three-dimensional reconstruction device comprises: A reflection fringe image acquisition module, used for acquiring a dual-frequency phase-shifted reflected fringe image formed by the dual-frequency phase-shifted initial fringe image being reflected by the object to be measured; a preprocessing module, configured to preprocess the dual-frequency phase-shifted reflected fringe image to obtain a first initial phase value of the front surface of the object to be measured and a second initial phase value of the rear surface of the object to be measured; a global optimization module, configured to perform global optimization of the modulation and phase of the front surface of the object to be measured and the modulation and phase of the rear surface of the object to be measured based on the first phase initial value, the second phase initial value, and a preset first modulation initial value of the front surface of the object to be measured and a preset second modulation initial value of the rear surface of the object to be measured, based on a pattern reconstruction method, to obtain modulation solution results and phase solution results of the front surface and the rear surface, respectively; A reconstruction module is used to perform three-dimensional reconstruction of the object to be measured based on the modulation depth solution result and the phase solution result.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the method for three-dimensional reconstruction of a double-layer transparent object as claimed in any one of claims 1 to 8 is implemented.

Images