Three-dimensional reconstruction motion error compensation method based on phase shift fringe profilometry

Abstract

The invention discloses a phase shift fringe profilometry-based three-dimensional reconstruction motion error compensation method, which relates to structured light three-dimensional reconstruction, and comprises the following steps of: acquiring a large number of deformed sine fringe images projected to the surfaces of various objects, calculating a first image characteristic value and a second image characteristic value, and constructing a data set; then training the image feature extraction model, and obtaining a first image feature value and a second image feature value of a multi-frame deformed sine stripe image of a to-be-detected object in a dynamic scene through the model; the method comprises the following steps: acquiring a coding phase and background light intensity of each frame of deformed sine stripe image of a to-be-measured object, then performing registration by using an optical flow method to obtain motion displacement of multiple frames of deformed sine stripe images of the to-be-measured object, and registering the multiple frames of deformed sine stripe images of the to-be-measured object to the same position to obtain the motion displacement of the to-be-measured object. And the coded phase of the sine fringe image after motion error compensation is obtained by using the registered sine fringe image of the object to be measured, so that the accuracy of three-dimensional reconstruction of the object is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of structured light three-dimensional reconstruction, and in particular to a three-dimensional reconstruction motion error compensation method based on phase-shift fringe profilometry. Background Art

[0002] Three-dimensional reconstruction refers to the process of using computer technology to reconstruct a three-dimensional geometric model based on the object images in a single view or multiple views in a two-dimensional space. Among them, phase-shift fringe profilometry (PSP) is widely used in many industries due to its advantages such as imaging accuracy, speed, and cost.

[0003] In the traditional multi-frequency phase-shift fringe profilometry, since multiple structured light patterns need to be projected, it is required that the object to be measured must be in a static state. However, realizing high-precision three-dimensional reconstruction of objects in a dynamic scene is an important requirement in industrial inspection. In order to meet the three-dimensional reconstruction requirements of objects in a dynamic scene, Fourier transform profilometry (FTP) can be used for single-frame three-dimensional reconstruction. However, Fourier transform profilometry requires careful design and calibration of the system pose, and the operation is complex. In addition, the object to be reconstructed often has a complex surface. When performing Fourier transform to extract the wrapped phase, due to the spectral leakage of the zero frequency, the extracted phase has errors, which affects the reconstruction accuracy of the object to be three-dimensionally reconstructed. Summary of the Invention

[0004] In order to overcome the defects in the prior art that there are motion errors in the object to be three-dimensionally reconstructed in a dynamic scene, resulting in low reconstruction accuracy and complex operation, the present invention proposes a three-dimensional reconstruction motion error compensation method based on phase-shift fringe profilometry.

[0005] To achieve the above object, the present invention adopts the following technical solutions, including:

[0006] S1: Calibrate the camera-projector system to obtain the parameters of the camera-projector system.

[0007] S2: Collect a large number of deformed sine fringe images, calculate the first image eigenvalue and the second image eigenvalue, and construct an image feature data set.

[0008] S3: Use a single-frame deformed sine fringe image as the input of the model, and the first image eigenvalue and the second image eigenvalue as the output of the model to construct an image feature extraction network model, train the model, and obtain a trained image feature extraction network model.

[0009] S4: Input the multi-frame deformed sine fringe images of the object to be measured obtained in the dynamic scene into the trained image feature extraction network model, and obtain the first image eigenvalue M and the second image eigenvalue D of each frame of deformed sine fringe image of the object to be measured.

[0010] S5: Based on the first image eigenvalue M and the second image eigenvalue D of each frame of the deformed sine stripe image of the object to be measured, register multiple frames of the deformed sine stripe images of the object to be measured to the same position, and obtain the registered sine stripe image of the object to be measured.

[0011] Preferably, in step S1, calibrate the camera-projector system to obtain the parameters of the camera-projector system, including:

[0012] S11: Calibrate the camera to obtain the internal parameters of the camera and the pose relationship between the camera and the calibration board;

[0013] S12: Set the sine stripe image of the projector;

[0014] S13: The projector projects the sine stripe image onto the calibration board, and use the camera to collect the projected image on the calibration board to obtain the pixel coordinates (x p , y p ) of the calibration board projected image;

[0015] S14: Based on the pixel coordinates (x p , y p ) of the calibration board projected image, calibrate the internal parameters of the projector and obtain the pose relationship between the projector and the calibration board;

[0016] S15: Based on the pose between the camera and the calibration board and the pose between the projector and the calibration board, obtain the pose relationship between the projector and the camera; wherein, the pose relationship between the projector and the camera includes the angle and the distance between the projector and the camera.

[0017] Preferably, in step S12, when setting the sine stripe image of the projector, based on the chip size of the projector, use the following formula to obtain the gray value of the pixel points in the sine stripe image:

[0018]

[0019] where, I W-n (x, y) is the gray value at the pixel coordinates (x, y) of the sine horizontal stripe image, and I H-n (x, y) is the gray value at the pixel coordinates (x, y) of the sine vertical stripe image; (x, y) are the pixel coordinates of the sine stripe image; a(x, y) is the background light intensity at the pixel coordinates (x, y); b(x, y) is the modulation degree at the pixel coordinates (x, y); f is the set sine stripe frequency; W and H are the total numbers of the horizontal and vertical coordinates of the projector chip pixel points respectively; n and N are the frame numbers and the total number of frames respectively, that is, the phase shift number and the number of phase shift steps.

[0020] Preferably, in step S13, the pixel coordinates (xp , y p ) is obtained based on the following formula:

[0021]

[0022] where (x c , y c ) are the pixel coordinates of the corner points of the projected image on the calibration board, and Φ v (x c , y c ) and Φ u (x c , y c ) are the horizontal and vertical sine phases of the corner points of the projected image on the calibration board, respectively.

[0023] Preferably, in step S2, the collecting a large number of deformed sine stripe images, calculating the first image eigenvalue and the second image eigenvalue, and constructing an image feature dataset includes:

[0024] S21: Set the projector to generate high-frequency sine stripe images with several phase shifting steps;

[0025] S22: Use the projector to project the high-frequency sine stripe images onto the surfaces of various objects, and obtain deformed sine stripe images with several phase shifting steps on the surfaces of each type of object;

[0026] S23: Based on the deformed sine stripe images with several phase shifting steps, obtain the first image eigenvalue M and the second image eigenvalue D.

[0027] Preferably, in step S23, the calculation formulas for the first image eigenvalue M and the second image eigenvalue D are respectively:

[0028]

[0029] where I n is the nth frame of deformed sine stripe image, and n and N are the frame number and the total number of frames, respectively.

[0030] Preferably, the image feature extraction network model includes an encoder-decoder network; the encoding layer contains c1 downsampling modules, and each downsampling module consists of a convolutional layer, a relu activation layer, a batch normalization layer, and a residual layer; the convolutional layer in the encoding layer has a convolution kernel size of ks1 and a stride of s1; the decoding layer contains c1 upsampling modules, and each upsampling module consists of a bilinear interpolation module, a convolutional layer, a relu activation layer, a batch normalization layer, and a residual layer; the convolutional layer in the decoding layer has a convolution kernel size of ks2 and a stride of s2.

[0031] Preferably, in step S5, based on the first image eigenvalue M and the second image eigenvalue D of each frame of the deformed sinusoidal fringe image of the object to be measured, multiple frames of the deformed sinusoidal fringe images of the object to be measured are registered to the same position, and the registered sinusoidal fringe image of the object to be measured is obtained, including:

[0032] S51: Based on the first image eigenvalue M and the second image eigenvalue D of each frame of the deformed sinusoidal fringe image, the encoded phase of each frame of the deformed sinusoidal fringe image is obtained and the background light intensity A;

[0033] S52: Based on the background light intensity A of each frame of the deformed sinusoidal fringe image, optical flow method is used for registration to obtain the motion displacement of multiple frames of the deformed sinusoidal fringe images;

[0034] S53: Based on the motion displacement of multiple frames of the deformed sinusoidal fringe images, multiple frames of the deformed sinusoidal fringe images are registered to the same position, and the registered sinusoidal fringe image of the object to be measured is obtained.

[0035] Preferably, after step S5, there is also S6: Based on the registered sinusoidal fringe image of the object to be measured, the encoded phase of the sinusoidal fringe image after motion error compensation is obtained The calculation formula is:

[0036]

[0037] where, δ n is the phase difference between the encoded phase of the nth frame of the deformed sinusoidal fringe image and the encoded phase of the first frame of the deformed sinusoidal fringe image ; I n is the nth frame of the deformed sinusoidal fringe image, n and N are the frame numbers and the total number of frames respectively, A' is the background light intensity of the registered sinusoidal fringe image of the object to be measured, and M' and D' are the first image eigenvalue and the second image eigenvalue of the registered sinusoidal fringe image of the object to be measured respectively.

[0038] Preferably, in step S51, the obtaining of the encoded phase of each frame of the deformed sinusoidal fringe image

[0039]

[0040] and the background light intensity A, the calculation formulas are respectively:

[0041] The advantages of the present invention are:

[0042] (1) The present invention collects a large number of deformed sine stripe images projected onto the surfaces of various objects, calculates the first image eigenvalue and the second image eigenvalue, constructs a data set, then trains an image feature extraction model to obtain a trained image feature extraction model. Next, the first image eigenvalue and the second image eigenvalue are obtained by passing multiple frames of deformed sine stripe images of the object to be measured in a dynamic scene through the model. Then, the multiple frames of deformed sine stripe images of the object to be measured are registered to the same position to obtain the registered sine stripe image of the object to be measured, and three-dimensional reconstruction of the subsequent object is performed, thereby improving the accuracy of the subsequent three-dimensional reconstruction of the object.

[0043] (2) The present invention designs a deep learning technology to train an image feature extraction network model, uses a U-shaped encoding-decoding structure, effectively combines the object image features of the deep layer and the shallow layer, can extract accurate encoded phases from a single frame of deformed sine stripe image, makes the phase difference between two adjacent frames of deformed sine stripe images of the object to be measured more real, and improves the accuracy of phase calculation.

[0044] (3) Based on the first image eigenvalue M and the second image eigenvalue D of each frame of deformed sine stripe image of the object to be measured, the present invention obtains the encoded phase and the background light intensity of each frame of deformed sine stripe image of the object to be measured, then uses the optical flow method for registration to obtain the motion displacement of multiple frames of deformed sine stripe images of the object to be measured. Based on the motion displacement of multiple frames of deformed sine stripe images of the object to be measured, the multiple frames of deformed sine stripe images of the object to be measured are registered to the same position. Then, using the registered sine stripe image of the object to be measured, the encoded phase of the sine stripe image after motion error compensation is obtained, which simplifies the operation steps for the subsequent three-dimensional reconstruction of the object and reduces the motion error of the object in the dynamic scene. Description of the Drawings

[0045] Figure 1 is the flow chart of the steps of the method of the present invention;

[0046] Figure 2 is the projection image collected by projecting the sine stripe image onto the object in the method of the present invention;

[0047] Figure 3 is the three-dimensional reconstruction effect diagram without using the method of the present invention;

[0048] Figure 4 is the three-dimensional reconstruction effect diagram after being processed by the method of the present invention. Detailed Embodiments

[0049] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0050] As Figures 1-4 shown, the present invention proposes a three-dimensional reconstruction motion error compensation method based on phase-shifting fringe profilometry for three-dimensional reconstruction of moving objects in a dynamic scene, including:

[0051] S1: In the camera-projector system, calibrate the camera and the projector to obtain the parameters of the camera and the projector, including:

[0052] S11: Calibrate the camera to obtain the internal parameter A of the camera c , and the pose relationship between the camera and the calibration board;

[0053] S12: Based on the chip size of the projector, set the sine fringe image of the projector, and obtain the gray value of the pixel point in the sine fringe image. The calculation formula is as follows:

[0054]

[0055] where I W-n (x, y) is the gray value at the pixel coordinates (x, y) of the sine horizontal fringe image, I H-n (x, y) is the gray value at the pixel coordinates (x, y) of the sine vertical fringe image, (x, y) is the pixel coordinates of the sine fringe image, a(x, y) is the background light intensity at the pixel coordinates (x, y), b(x, y) is the modulation degree at the pixel coordinates (x, y), f is the set sine fringe frequency, W and H are the total numbers of the horizontal and vertical coordinates of the projector chip pixels respectively, n and N are the frame numbers (phase-shifting numbers) and the total number of frames (phase-shifting steps) respectively;

[0056] S13: The projector projects the sine fringe image onto the calibration board, and the camera collects the projected image on the calibration board to obtain the pixel coordinates (x p , y p ) of the projected image on the calibration board. The calculation formula is:

[0057]

[0058] where (x c , y c ) are the pixel coordinates of the corner points of the projected image on the calibration board, Φ v (x c , y c ) and Φu (x c , y c ) are the horizontal and vertical sine phases of the corner points of the projected image on the calibration board, respectively.

[0059] S14: Based on the pixel coordinates (x p , y p ) of the projected image on the calibration board, calibrate the internal parameter A p of the projector, and obtain the pose relationship between the projector and the calibration board;

[0060] S15: Based on the pose relationship between the camera and the calibration board, and the pose relationship between the projector and the calibration board, obtain the pose relationship between the projector and the camera; among them, the pose relationship between the projector and the camera includes the angle R and the distance T between the projector and the camera.

[0061] Before step S1, it is first necessary to build a camera-projector system.

[0062] S2: Collect a large number of deformed sine fringe images, calculate the first image eigenvalue and the second image eigenvalue, and construct an image feature dataset, including:

[0063] S21: Set the projector to generate high-frequency sine fringe images with several phase-shifting steps;

[0064] S22: Use the projector to project the high-frequency sine fringe images onto the surfaces of various objects, and obtain deformed sine fringe images with several phase-shifting steps on the surfaces of each type of object;

[0065] S23: Based on the deformed sine fringe images with several phase-shifting steps, obtain the first image eigenvalue M and the second image eigenvalue D, and the calculation formulas are respectively:

[0066]

[0067] where I n is the nth frame of deformed sine fringe image, and n and N are the frame numbers (phase-shifting numbers) and the total number of frames (phase-shifting steps), respectively;

[0068] In this embodiment, the high-frequency sine fringe images with several phase-shifting steps are 12-step high-frequency sine fringe images.

[0069] S3: Build an image feature extraction network model, use a single-frame deformed sine fringe image as the input of the model, and the first image eigenvalue and the second image eigenvalue as the output of the model, train the model, and obtain a trained image feature extraction network model;

[0070] The image feature extraction network model includes an encoding-decoding network with a U-shaped structure. Among them, the encoding layer contains c1 downsampling modules, and each downsampling module consists of a convolutional layer, a ReLU activation layer, a batch normalization layer, and a residual layer. The convolutional layer in the encoding layer has a convolutional kernel size of ks1 and a stride of s1. The decoding layer contains c1 upsampling modules, and each upsampling module consists of a bilinear interpolation module, a convolutional layer, a ReLU activation layer, a batch normalization layer, and a residual layer. The convolutional layer in the decoding layer has a convolutional kernel size of ks2 and a stride of s2.

[0071] Using the convolutional layer can better focus on the local information on the deformed sine deformation image and extract the required phase information from the local. The U-shaped structure can well fuse the shallow and deep features, can well perceive the shallow texture information such as edges, and make the training more stable.

[0072] The image feature dataset is allocated into a training set, a validation set, and a test set in a ratio of 8:1:1, and the image feature extraction network model is trained until the model converges.

[0073] S4: Input the multi-frame deformed sine fringe images of the object to be measured obtained in the dynamic scene into the trained image feature extraction network model, and obtain the first image feature value M and the second image feature value D of each frame of the deformed sine fringe image of the object to be measured.

[0074] S5: Based on the first image feature value M and the second image feature value D of each frame of the deformed sine fringe image of the object to be measured, register the multi-frame deformed sine fringe images of the object to be measured to the same position, and obtain the registered sine fringe image of the object to be measured.

[0075] S51: Based on the first image feature value M and the second image feature value D of each frame of the deformed sine fringe image, obtain the encoded phase of each frame of the deformed sine fringe image and the background light intensity A, and the calculation formulas are respectively:

[0076]

[0077] where I is the deformed sine fringe image.

[0078] S52: Based on the background light intensity of each frame of the deformed sine fringe image, use the optical flow method for registration to obtain the motion displacement of the multi-frame deformed sine fringe images.

[0079] S53: Based on the motion displacement of the multi-frame deformed sine fringe images, register the multi-frame deformed sine fringe images to the same position, and obtain the registered sine fringe image of the object to be measured.

[0080] S6: Obtain the encoded phase of the sine fringe image after motion error compensation based on the registered sine fringe image of the object to be measured. The calculation formula is:

[0081]

[0082] where δ n is the encoded phase of the nth deformed sine fringe image and the encoded phase of the first deformed sine fringe image The phase difference between them; I n is the nth deformed sine fringe image, n and N are the frame numbers (phase shift numbers) and the total number of frames (phase shift steps) respectively, A' is the background light intensity of the registered sine fringe image of the object to be measured, and M' and D' are the first and second image eigenvalues of the registered sine fringe image of the object to be measured respectively.

[0083] The encoded phase of the sine fringe image after motion error compensation is used for the three-dimensional reconstruction of the moving object to be measured.

[0084] Of course, for those skilled in the art, the details of the above exemplary embodiments are not limited to the present invention, but also include the same or similar structures that can be implemented in other specific forms without departing from the spirit or basic characteristics of the present invention. Therefore, in any aspect, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be included in the present invention. Any reference signs in the claims should not be construed as limiting the claims involved.

[0085] In addition, it should be understood that although this specification is described in terms of embodiments, not every embodiment only contains an independent technical solution. This narrative way of the specification is only for clarity. Those skilled in the art should regard the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

[0086] The technologies, shapes, and structures not detailed in the present invention are all well-known technologies.

Claims

1. A three-dimensional reconstruction motion error compensation method based on phase-shifting fringe profilometry, characterized in that Including: S1: Calibrate the camera-projector system to obtain the parameters of the camera-projector system; S2: Collect a large number of deformed sinusoidal fringe images, calculate the first image eigenvalue and the second image eigenvalue, and construct an image feature dataset; S3: Use a single-frame deformed sinusoidal fringe image as the input of the model, and the first image eigenvalue and the second image eigenvalue as the output of the model to construct an image feature extraction network model, train the model, and obtain a trained image feature extraction network model; S4: Input the multi-frame deformed sinusoidal fringe images of the object to be measured obtained in the dynamic scene into the trained image feature extraction network model to obtain the first image eigenvalue M and the second image eigenvalue D of each frame of the deformed sinusoidal fringe image of the object to be measured; S5: Based on the first image eigenvalue M and the second image eigenvalue D of each frame of the deformed sinusoidal fringe image of the object to be measured, register the multi-frame deformed sinusoidal fringe images of the object to be measured to the same position to obtain the registered deformed sinusoidal fringe image of the object to be measured.

2. The three-dimensional reconstruction motion error compensation method based on phase-shifting fringe profilometry according to claim 1, characterized in that, In step S1, when calibrating the camera-projector system to obtain the parameters of the camera-projector system, it includes: S11: Calibrate the camera to obtain the internal parameters of the camera and the pose relationship between the camera and the calibration board; S12: Set the sinusoidal fringe image of the projector; S13: The projector projects a sinusoidal fringe image onto the calibration board, and the camera is used to collect the projected image on the calibration board to obtain the pixel coordinates (x p , y p ) of the projected image of the calibration board; S14: Based on the pixel coordinates (x p , y p ) of the calibration board projection image, calibrate the internal parameters of the projector and obtain the pose relationship between the projector and the calibration board; S15: Based on the pose between the camera and the calibration board and the pose between the projector and the calibration board, obtain the pose relationship between the projector and the camera; wherein, the pose relationship between the projector and the camera includes the angle and distance between the projector and the camera.

3. The three-dimensional reconstruction motion error compensation method based on phase-shifting fringe profilometry according to claim 2, wherein In the step S12, when setting the sinusoidal fringe image of the projector, based on the chip size of the projector, use the following formula to obtain the gray value of the pixel points in the sinusoidal fringe image: Among them, I W-n (x, y) is the gray value at the pixel coordinates (x, y) of the sine horizontal stripe image, and I H-n (x, y) is the gray value at the pixel coordinates (x, y) of the sine vertical stripe image; (x, y) are the pixel coordinates of the sine stripe image; a(x, y) is the background light intensity at the pixel coordinates (x, y); b(x, y) is the modulation degree at the pixel coordinates (x, y); f is the set sine stripe frequency; W and H are the total numbers of the horizontal and vertical coordinates of the projector chip pixels respectively; n and N are the frame number and the total number of frames respectively, that is, the phase shift number and the phase shift steps.

4. The three-dimensional reconstruction motion error compensation method based on phase-shifting fringe profilometry according to claim 3, wherein, In step S13, the pixel coordinates (x p , y p ) of the calibration plate projection image are obtained based on the following formula: Among them, (x c , y c ) is the pixel coordinate of the corner point of the projected image on the calibration board, and Φ v (x c , y c ) and Φ u (x c , y c ) are the horizontal and vertical sine phases of the corner point of the projected image on the calibration board, respectively.

5. The three-dimensional reconstruction motion error compensation method based on phase-shifting fringe profilometry according to claim 1, wherein In step S2, the process of collecting a large number of deformed sinusoidal fringe images, calculating the first image eigenvalue and the second image eigenvalue, and constructing an image feature dataset includes: S21: Set the projector to generate high-frequency sinusoidal fringe images with several phase-shifting steps; S22: Use the projector to project the high-frequency sinusoidal fringe images onto the surfaces of various objects to obtain the deformed sinusoidal fringe images with several phase-shifting steps on the surfaces of each object; S23: Based on the deformed sinusoidal fringe images with several phase-shifting steps, obtain the first image eigenvalue M and the second image eigenvalue D.

6. The three-dimensional reconstruction motion error compensation method based on phase-shifting fringe profilometry according to claim 5, characterized in that In step S23, the calculation formulas for the first image eigenvalue M and the second image eigenvalue D are respectively: Among them, I n is the deformed sine fringe image of the nth frame, where n and N are the frame number and the total number of frames, respectively.

7. The three-dimensional reconstruction motion error compensation method based on phase-shifting fringe profilometry according to claim 1, characterized in that The image feature extraction network model includes an encoder-decoder network; the encoding layer contains c1 downsampling modules, and each downsampling module consists of a convolutional layer, a relu activation layer, a batch normalization layer, and a residual layer; the convolutional layer in the encoding layer has a convolutional kernel size of ks1 and a stride of s1; the decoding layer contains c1 upsampling modules, and each upsampling module consists of a bilinear interpolation module, a convolutional layer, a relu activation layer, a batch normalization layer, and a residual layer; the convolutional layer in the decoding layer has a convolutional kernel size of ks2 and a stride of s2.

8. The three-dimensional reconstruction motion error compensation method based on phase-shifting fringe profilometry according to claim 1, wherein In the step S5, based on the first image eigenvalue M and the second image eigenvalue D of each frame of the deformed sinusoidal fringe image of the object to be measured, multiple frames of the deformed sinusoidal fringe images of the object to be measured are registered to the same position, and the registered sinusoidal fringe image of the object to be measured is obtained, including: S51: Obtain the encoded phase of each deformed sinusoidal fringe image based on the first image eigenvalue M and the second image eigenvalue D of each deformed sinusoidal fringe image and the background light intensity A; S52: Based on the background light intensity A of each frame of the deformed sinusoidal fringe image, optical flow method is used for registration to obtain the motion displacements of multiple frames of the deformed sinusoidal fringe images; S53: Based on the motion displacements of multiple frames of the deformed sinusoidal fringe images, multiple frames of the deformed sinusoidal fringe images are registered to the same position, and the registered sinusoidal fringe image of the object to be measured is obtained.

9. The three-dimensional reconstruction motion error compensation method based on phase-shifting fringe profilometry according to claim 1, wherein After step S5, it further includes S6: based on the registered sinusoidal fringe image of the object to be measured, obtain the encoded phase of the sinusoidal fringe image after motion error compensation The calculation formula is as follows: Among them, δ n is the encoded phase of the deformed sinusoidal fringe image of the nth frame and the encoded phase of the deformed sinusoidal fringe image of the first frame The phase difference between them; I n is the deformed sinusoidal fringe image of the nth frame, n and N are the frame numbers and the total number of frames respectively, A′ is the background light intensity of the sinusoidal fringe image of the object to be measured after registration, and M′ and D′ are the first image eigenvalue and the second image eigenvalue of the sinusoidal fringe image of the object to be measured after registration.

10. The three-dimensional reconstruction motion error compensation method based on phase-shifting fringe profilometry according to claim 8, characterized in that, In step S51, the encoded phase of each frame of deformed sinusoidal fringe image is obtained and the background light intensity A, and the calculation formulas are respectively as follows: Wherein, I is the deformed sinusoidal fringe image.

Images