A method and system for nonlinear carrier phase removal of fringe projection images

Abstract

The application discloses a kind of nonlinear carrier phase removal methods and systems of fringe projection image.The method obtains the fringe projection image of first and second view angle with nonlinear carrier;Including the following steps: (1) the fringe projection image of first and second view angle is fused using image splicing algorithm;(2) using singular value decomposition, the first principal component obtained is regarded as background component, and other components are regarded as foreground component;(3) detect foreground region and background region;(4) polynomial curved surface fitting is carried out to the background region of first and / or second view angle, and the first and / or second nonlinear phase is obtained;(5) subtract respective carrier phase and the respective nonlinear phase obtained in step (4).The application obtains global fusion fringe projection image without obstruction by double camera fringe imaging system, and detects background region based on singular value decomposition, so that nonlinear carrier phase is accurately removed.

Description

Technical Field

[0001] This invention belongs to the field of image processing, and more specifically, relates to a nonlinear carrier phase removal method and system for striped projection images. Background Technology

[0002] With the development of technology, people's demand for information is increasing, and three-dimensional information contains more information than one-dimensional and two-dimensional information. Acquiring three-dimensional information is usually achieved through three-dimensional vision imaging technology, and fringe projection profilometry is currently the most widely used three-dimensional vision imaging technology. The core problem in fringe projection profilometry is phase extraction from a single frame of fringe projection. With the advancement of machine learning, various algorithms, such as PCA-based algorithms and phase fitting algorithms, have been developed and can measure the phase of real objects.

[0003] Although many algorithms can now extract the phase of objects in a single frame of fringe projection image with high accuracy, they still have many challenges and shortcomings:

[0004] 1. The assumptions are too stringent, limiting applicability. Existing methods require strong assumptions; only by meeting these assumptions can high detection accuracy be achieved, making them unsuitable for large-scale application in real-world environments.

[0005] 2. Requires a large amount of data and has high data requirements. Neural network-based methods have high accuracy but require a large amount of training data and a high degree of correlation between the data, which limits their practicality.

[0006] 3. Lack of convenience. Currently used methods usually require additional capture of stripe projection images of the background plane, which is not convenient and cannot be applied to dynamic scenes. Summary of the Invention

[0007] To address the aforementioned deficiencies or improvement needs of existing technologies, this invention provides a nonlinear carrier phase removal method and system for striped projection images. The aim is to obtain an unobstructed, globally fused striped projection image using a dual-camera striped imaging system, detect background regions based on singular value decomposition, and obtain the nonlinear carrier phase through polynomial fitting, thereby accurately removing it. This solves the technical problem of unsatisfactory nonlinear carrier phase removal in single-camera striped projection images due to inaccurate background region detection caused by occlusion.

[0008] To achieve the above objectives, according to one aspect of the present invention, a nonlinear carrier phase removal method for fringe projection images is provided, which is applied to a dual-camera fringe imaging system with a planar background to obtain fringe projection images with a first and a second viewpoint having a nonlinear carrier; the method includes the following steps:

[0009] (1) The stripe projection images of the first view and the second view are fused using an image stitching algorithm to obtain a global fused stripe projection image with a nonlinear carrier.

[0010] (2) The global fused stripe projection image with nonlinear carrier obtained in step (1) is subjected to singular value decomposition, and the first principal component is used as the background component and the other components are used as the foreground components.

[0011] (3) Detect the foreground region and background region based on the foreground component obtained in step (2), and separate the foreground region and background region from the first and second viewpoint stripe projection images by globally fused stripe projection images.

[0012] (4) Background areas obtained from the first and / or second perspectives in step (3) Perform polynomial surface fitting to obtain the fitted phase as the first and / or second nonlinear phase;

[0013] (5) After phase unfolding the stripe projection images of the first view and / or the second view obtained in step (1), subtract their respective carrier phases and the respective nonlinear phases obtained in step (4) to obtain the phase of the object under test after removing the nonlinear carrier.

[0014] Preferably, the nonlinear carrier phase removal method for the striped projection image, wherein the dual-camera striped imaging system includes a grating projector, a first and a second camera, and a planar display panel; the principal optical axis of the grating projector is perpendicular to the planar display panel; the first and second cameras are symmetrical about the principal optical axis and have the same imaging range; the imaging range is located within the planar display panel;

[0015] The grating projector projects a nonlinear carrier wave onto a flat plate on which the object to be tested is placed. The first and second cameras image the flat plate on which the object to be tested is placed, respectively obtaining striped projection images with a first and a second viewpoint of nonlinear carrier wave.

[0016] Preferably, the nonlinear carrier phase removal method for the striped projection image includes step (1) as follows:

[0017] (1-1) Based on the imaging of the first and second cameras on the same mark, calibrate the first and second cameras, obtain the intrinsic and extrinsic parameters of the first and second cameras, and calculate the coordinate mapping function between the first and second cameras; the intrinsic parameters include the lens focal length and the coordinates of the camera image center; the extrinsic parameters include the rotation matrix R and the translation vector T of the camera in the world coordinate system.

[0018] After calibrating the intrinsic and extrinsic parameters of the first and second cameras in the world coordinate system, the affine transformation relationship between the first and second cameras and the world coordinate system was established, and the mapping relationship between the coordinates of the first and second cameras, i.e., the mapping function, was obtained.

[0019] (1-2) The stripe projection images of the first viewpoint and the second viewpoint are mapped to the same coordinate system through the coordinate mapping function between the first camera and the second camera obtained in step (1-1). The stripe projection images of the first viewpoint and the second viewpoint are registered based on the similarity of local image features in units of superpixels and then fused into a globally fused stripe projection image with a nonlinear carrier. The superpixel is a grid centered on a pixel. The local image features are the geometric or non-set features of the superpixel.

[0020] Preferably, the nonlinear carrier phase removal method for the striped projection image specifically includes step (2):

[0021] (2-1) Globally fused stripe projection image Normalization is performed to obtain the normalized image matrix I. N :

[0022]

[0023] Where, min(I) is the maximum gray level in the globally fused stripe projection image I, and max(I) is...

[0024] (2-2) The normalized image matrix I obtained in step (2-1) N Perform singular value decomposition:

[0025] I N =US T

[0026] in, The left matrix, The right matrix, It is a singular value matrix;

[0027] (2-3) Based on the singular value decomposition matrix obtained in step (2-2), calculate the first principal component I of the globally fused fringe projection image. f As the background component, the other components are treated as the foreground component I. e :

[0028] I f =U(:,1)·V t (1,:)·S(1,1)

[0029] Among them, I f: indicates the maximum principal component, and : indicates taking all rows or columns of the matrix;

[0030] Calculate the other components I of the globally fused fringe projection image. e ;

[0031]

[0032] Where k is min(m,n), that is, k takes the minimum value of m and n.

[0033] Preferably, the nonlinear carrier phase removal method for the striped projection image specifically includes step (3):

[0034] (3-1) The component I obtained in step (2) e Perform filtering and noise reduction, and then normalize to obtain the normalized background component. A 3×3 Gaussian filter is preferred.

[0035] (3-2) Normalize the background components obtained in step (3-1) Binarization is performed using a preset threshold θ to obtain the binarized image I. bin :

[0036]

[0037] Among them, I bin ∈{0,1} represents the binarized image, and θ is the threshold;

[0038] (3-3) The binarized image I obtained in step (3-2) bin Filtering and denoising are performed to obtain the denoised binarized image I. binmed The preferred method is to perform a 5×5 median filter, i.e.:

[0039] I binmed =medfilter(I bin ,5)

[0040] Where medfilter(·,a) represents median filtering with a window size of a×a.

[0041] (3-4) Obtain the binarized image I from step (3-3). binmed Perform object region determination to obtain the binarized image region (r) of the object to be tested. t ,r b ,r l ,r r ), where r t Let r be the upper boundary of the object to be measured. b Let r be the lower boundary of the object to be measured. l Let r be the left boundary of the object to be measured.r The right boundary of the object to be measured;

[0042] (3-5) For the background area obtained in step (3-4) Based on the pixel mapping relationship between the fringe projection images obtained from the first and second viewpoints and the globally fused fringe projection image, the background regions of the first and second viewpoints are obtained, and are denoted as follows: and

[0043] Preferably, in the nonlinear carrier phase removal method for the striped projection image, the threshold θ in step (3-2) is adaptively determined according to the following steps: S1, calculate the normalization Histogram H I The histogram divides [0,1] into 100 regions; S2, for H I After reversing the sequence, find the index corresponding to the first minimum point after the first maximum point, and map the index to a real number θ in the range [0,1]. m To enhance robustness, the threshold is set to θ = θ m +0.05.

[0044] Preferably, the nonlinear carrier phase removal method for the striped projection image specifically includes step (3-4):

[0045] For binarized image I binmed Morphological etching was performed to obtain the etching image I binop A 5×5 matrix structuring element is preferably used for morphological erosion, and the eroded image is subjected to I... binop Summing the rows and columns yields I sumx and I sumy ; Calculate I sumx The average value M sumx And traverse I sumx , will I sumx (i) <M sumx sumx (i+1) is used as the condition for determining the upper boundary of the object region, and I sumx (i+1) <M sumx sumx (i) As a condition for determining the lower boundary of the object region, the upper boundary r of the object region is obtained. t and lower boundary r b ; Calculate I sumy The average value M sumy And traverse I sumy , will I sumy (i) <M sumy sumy (i+1) is used as the condition for determining the left boundary of the object region, and I sumy (i+1) <M​​​sumy sumy (i) As a condition for determining the lower boundary of the object region, the left boundary r of the object region is obtained. l and right boundary r r The object region (r) is obtained by combining the results. t ,r b ,r l ,r r ), and background area

[0046] Preferably, the nonlinear carrier phase removal method for the striped projection image, specifically step (4) is as follows:

[0047] (4-1) For the background area in the first or second perspective Among them, point (x) p ,y p The phase Φ is extracted and expanded using a phase extraction method. M (x p ,y p )=2πf0x+Φ0(x p ,y p ), removing the carrier phase yields the nonlinear phase Φ0(x) p ,y p ):

[0048] Φ0(x p ,y p )=Φ M (x p ,y p )-2πf0x

[0049] Where f0 is the fringe frequency of the fringe projection image, which is determined by the imaging system;

[0050] (4-2) Nonlinear phase for the first and second perspectives Based on the background areas obtained in step (4-1) from the second perspective nonlinear phase The least squares method is used to fit the polynomial surface p(x,y) as the nonlinear phase of the first and second perspectives.

[0051] Preferably, in the nonlinear carrier phase removal method for the striped projection image, step (4-2) employs a third-order bivariate polynomial for surface fitting:

[0052] p(x,y)=c 30 x 3 +c 03 y 3 +c​21 x 2 y+c 12 y 2 x+c 20 x 2 +c 02 y 2 +c 30 x 3

[0053] +c 11 xy+c 10 x+c 01 y+c 00

[0054] Substitute data The coefficients c of the polynomial p are fitted using the least squares method. ij The fitting result is used as the nonlinear phase Φ0(x) p ,y p ).

[0055] According to another aspect of the present invention, a nonlinear carrier phase removal system for striped projection images is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the program to implement the steps of the nonlinear carrier phase removal method for striped projection images provided by the present invention.

[0056] In summary, compared with the prior art, the above-described technical solutions conceived by this invention can achieve the following beneficial effects:

[0057] The present invention provides a nonlinear carrier phase removal method and system for striped projection images. By using a dual-camera striped imaging system, striped projection images are acquired from symmetrical first and second perspectives. A globally fused striped projection image is obtained through image fusion. Striped projection information of occluded parts is removed, and background areas are detected more accurately through singular value decomposition, thereby obtaining a more accurate nonlinear carrier phase. After removal, a more accurate phase of the object under test is obtained.

[0058] Furthermore, the new assumptions proposed in this invention are weaker, making them more adaptable to real-world environments and thus more applicable. Secondly, a single singular value decomposition can separate the background region from the actual object region, enabling effective data utilization and addressing the shortcomings of existing technologies in data utilization and unnecessary repetitive processes, thereby improving efficiency and reducing costs. The first principal component obtained by decomposing the globally fused fringe projection image primarily includes background region information, while other components contain object phase information. Based on this novel assumption, the object region can be determined with a single decomposition of the fringe projection image, significantly reducing interference with the object phase during subsequent fitting of the nonlinear carrier phase. This greatly improves detection accuracy and efficiency, reduces the interference of the object phase on the actual detected phase, and represents a more robust background region detection algorithm at a lower cost. Attached Figure Description

[0059] Figure 1 This is a schematic diagram of the dual-camera stripe imaging system used in this invention;

[0060] Figure 2 This is a schematic diagram of the nonlinear carrier phase removal method for striped projection images provided by the present invention;

[0061] Figure 3 This is a schematic diagram of the region decomposition of a striped projection image according to the present invention. Detailed Implementation

[0062] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0063] The nonlinear carrier phase removal method for stripe projection images provided by this invention is applied to a dual-camera stripe imaging system with a planar background.

[0064] The dual-camera stripe imaging system has the following structure: Figure 1 As shown, it includes a grating projector, a first and a second camera, and a flat panel; the main optical axis of the grating projector is perpendicular to the flat panel; the first and second cameras are symmetrical about the main optical axis and have the same imaging range; the imaging range is located within the flat panel.

[0065] The grating projector projects a nonlinear carrier wave onto a flat plate on which the object to be tested is placed. The first and second cameras image the flat plate on which the object to be tested is placed, respectively obtaining striped projection images with a first and a second viewpoint having a nonlinear carrier wave.

[0066] Includes the following steps:

[0067] (1) The fringe projection images from the first and second perspectives are fused using an image stitching algorithm to obtain a globally fused fringe projection image with a nonlinear carrier wave; the following method is preferred:

[0068] (1-1) Based on the imaging calibration of the first and second cameras on the same mark, the first and second cameras are calibrated to obtain the intrinsic and extrinsic parameters of the first and second cameras, and the coordinate mapping function between the first and second cameras is calculated; the intrinsic parameters include the lens focal length and the coordinates of the camera image center; the extrinsic parameters include the rotation matrix R and the translation vector T of the camera in the world coordinate system.

[0069] After calibrating the intrinsic and extrinsic parameters of the first and second cameras in the world coordinate system, the affine transformation relationship between the first and second cameras and the world coordinate system was established. Thus, the mapping relationship between the coordinates of the first and second cameras, i.e., the mapping function, was obtained.

[0070] (1-2) The stripe projection images of the first viewpoint and the second viewpoint are mapped to the same coordinate system through the coordinate mapping function between the first camera and the second camera obtained in step (1-1). The stripe projection images of the first viewpoint and the second viewpoint are registered based on the similarity of local image features in units of superpixels and then fused into a globally fused stripe projection image with a nonlinear carrier. The superpixel is a grid centered on a pixel. The local image features are the geometric or non-set features of the superpixel.

[0071] After local image feature registration, the stripe projection images from the first and second perspectives are stitched and fused. In the fused global stripe projection image, each point includes stripe brightness information obtained by the dual cameras, which complement each other and avoid stripe information occlusion.

[0072] The accuracy of fringe projection images is crucial for singular value decomposition and for detecting foreground and background. However, in single-camera fringe imaging systems, the spatial steric hindrance between the grating projector and the camera inevitably creates a certain angle, making it easy for the object under test to be obscured. A single camera cannot guarantee the extraction of all fringe projection information, resulting in inaccurate or lost fringe phase information during foreground-background separation.

[0073] This invention employs a dual-camera stripe imaging system, which corrects phase difference by fusing stripe projection images from a first viewpoint and a second viewpoint to obtain a globally fused stripe projection image equivalent to a front-facing camera, making foreground-background separation more accurate.

[0074] (2) The globally fused fringe projection image with a nonlinear carrier obtained in step (1) is subjected to singular value decomposition. The first principal component is used as the background component, and the other components are used as the foreground components. The specific steps are as follows:

[0075] (2-1) Globally fused stripe projection image Normalization is performed to obtain the normalized image matrix I. N :

[0076]

[0077] Where, min(I) is the maximum gray level in the globally fused stripe projection image I, and max(I) is...

[0078] (2-2) The normalized image matrix I obtained in step (2-1) N Perform singular value decomposition:

[0079] I N =USV T

[0080] in, The left matrix, The right matrix, It is a singular value matrix;

[0081] (2-3) Based on the singular value decomposition matrix obtained in step (2-2), calculate the first principal component I of the globally fused fringe projection image. f As the background component, the other components are treated as the foreground component I. e :

[0082] I f =U(:,1)·V T (1,:)·S(1,1)

[0083] Among them, I f `:` indicates the maximum principal component, and `:` indicates taking all rows or columns of the matrix. f It is the most prominent linear component in the striped projection image. Its shape is basically the same as that of the background striped projection image, but the rows and columns of deformed stripes produce blurring.

[0084] The first principal component image obtained after SVD decomposition of the fringe projection image has high similarity to the fringe image on the reference plane without the measured object.

[0085] Calculate the other components I of the globally fused fringe projection image. e ;

[0086]

[0087] Where k is min(m,n), that is, k takes the minimum value of m and n.

[0088] (3) Detect the foreground and background regions based on the foreground components obtained in step (2), and separate the foreground and background regions in the fringe projection images from the first and second viewpoints using the globally fused fringe projection images; specifically, this includes the following steps:

[0089] (3-1) The component I obtained in step (2) e Perform filtering and noise reduction, and then normalize to obtain the normalized background component. A 3×3 Gaussian filter is preferred.

[0090] (3-2) Normalize the background components obtained in step (3-1) Binarization is performed using a preset threshold θ to obtain the binarized image I. bin :

[0091]

[0092] Among them, I bin ∈{0,1} represents the binarized image, and θ is the threshold. The threshold θ is adaptively determined according to the following steps: S1, calculate the normalization. Histogram H I The histogram divides [0,1] into 100 regions; S2, for H I After reversing the sequence, find the index corresponding to the first minimum point after the first maximum point, and map the index to a real number θ in the range [0,1]. m To enhance robustness, the threshold is set to θ = θ m +0.05.

[0093] (3-3) The binarized image I obtained in step (3-2) bin Filtering and denoising are performed to obtain the denoised binarized image I. binmed The preferred method is to perform a 5×5 median filter, i.e.:

[0094] I binmed =medfilter(I bin ,5)

[0095] Where medfilter(·,a) represents median filtering with a window size of a×a.

[0096] For the binarized image I obtained in step (3-3) binmed perform object region judgment to obtain the binarized image region (r t , r b , r l , r r ) of the object to be measured, where r t is the upper boundary of the object to be measured, r b is the lower boundary of the object to be measured, r l is the left boundary of the object to be measured, r r is the right boundary of the object to be measured; specifically:

[0097] Perform morphological erosion on the binarized image I binmed to obtain the eroded image I binop . Preferably, use a 5×5 matrix structuring element for morphological erosion. Sum the rows and columns of the eroded image I binop to obtain I sumx and I sumy ; Calculate the average value M sumx of I sumx and traverse I sumx . Take I sumx (i) < M sumx < I sumx (i + 1) as the condition for judging the upper boundary of the object region, and take I sumx (i + 1) < M sumx < I sumx (i) as the condition for judging the lower boundary of the object region to obtain the upper boundary r t and the lower boundary r b of the object region; Calculate the average value M sumy of I sumy and traverse I sumy . Take I sumy (i) < M sumy < I sumy (i + 1) as the condition for judging the left boundary of the object region, and take I sumy (i + 1) < M sumy < I sumy (i) as the condition for judging the lower boundary of the object region to obtain the left boundary r l and the right boundary r r of the object region. Combine to obtain the object region (r t , r b , r l , r r ), and the background region = {(x, y)|0 ≤ x < r t or r b < x < m or 0 ≤ y < r l or r r < y < n}.

[0098] (3-5) For the background area obtained in step (3-4) Based on the pixel mapping relationship between the fringe projection images obtained from the first and second viewpoints and the globally fused fringe projection image, the background regions of the first and second viewpoints are obtained, and are denoted as follows: and

[0099] (4) Background areas obtained from the first and / or second perspectives in step (3) Perform polynomial surface fitting to obtain the fitted phase as the first and / or second nonlinear phase; specifically, this includes the following steps:

[0100] (4-1) For the background area in the first or second perspective Among them, point (x) p ,y p The phase Φ is extracted and expanded using a phase extraction method. M (x p ,y p )=2πf0x+Φ0(x p ,y p ), removing the carrier phase yields the nonlinear phase Φ0(x) p ,y p ):

[0101] Φ0(x p ,y p )=Φ M (x p ,y p )-2πf0x

[0102] Where f0 is the fringe frequency of the fringe projection image, which is determined by the imaging system.

[0103] (4-2) Nonlinear phase for the first and second perspectives Based on the background areas obtained in step (4-1) from the second perspective nonlinear phase The least squares method is used to fit the polynomial surface p(x,y) as the nonlinear phase of the first and second perspectives. For example, for a region Surface fitting is performed using a third-order bivariate polynomial:

[0104] p(x,y)=c 30 x 3 +c 03 y 3 +c 21 x 2 y+c 12y 2 x+c 20 x 2 +c 02 y 2 +c 30 x 3

[0105] +c 11 xy+c 10 x+c 01 y+c 00

[0106] Substitute the data Φ0(x) p ,y p ), The coefficients c of the polynomial p are fitted using the least squares method. ij The fitting result is used as the nonlinear phase Φ0(x) p ,y p ).

[0107] (5) After phase unwrapping the stripe projection images of the first and / or second viewpoints obtained in step (1), subtract their respective carrier phases and the respective nonlinear phases obtained in step (4) to obtain the phase of the object under test after removing the nonlinear carrier; that is:

[0108] For the first-person perspective, the unfolded phase is obtained by performing phase unfolding on the fringe projection image. Subtract nonlinear carrier phase Obtain the phase Φ1 of the object under test from a first-person perspective after removing the nonlinear carrier.

[0109]

[0110] For the second perspective, the unfolded phase is obtained by performing phase unfolding on the fringe projection image. Subtract nonlinear carrier phase Determined by the imaging system, the phase Φ2 of the object under test, after removing the nonlinear carrier, is obtained from the second viewpoint;

[0111]

[0112] Where f0 is the fringe frequency of the fringe projection image, which is determined by the imaging system.

[0113] The phases of the object under test from the first and second perspectives, after removing the nonlinear carrier, can be independently recovered into three-dimensional images, and can be mutually checked and fused to obtain a more accurate overall three-dimensional image.

[0114] The implementation process of this invention is as follows: Figure 2As shown. First, this invention proposes a new theoretical assumption: the foreground-background separability assumption of fringe projection images. Based on this assumption, a fringe projection image can be separated into a background fringe component image and a foreground object component image. Then, the separated background region is processed to obtain the actual phase of the object. The new method based on this assumption has higher detection accuracy, stronger stability compared to existing methods, and is less affected by the phase of the object itself, thus obtaining a more accurate object phase.

[0115] A schematic diagram of region decomposition of a fringe projection image is shown below. Figure 3 As shown, a fringe projection image typically consists of two parts: an object region and a background region. In the object region, the projected fringes are highly modulated by the object's surface, resulting in deformation and high recognizability. In the background region, the projected fringes generally contain only carrier and nonlinear components. The fringes exhibit non-uniform variations laterally. The present invention aims to separate the object region from the background region in the fringe projection image to facilitate subsequent phase detection of the actual object and reduce interference from the actual object on the phase detection results.

[0116] This invention provides a new theoretical assumption for nonlinear carrier phase extraction, and based on this assumption, proposes a novel method for better fitting the phase of a background stripe image using matrix decomposition. This method separates the background region from the object region through singular value decomposition and calculates the phase of the actual object, achieving high accuracy and stability.

[0117] This invention first proposes a novel theoretical assumption: the foreground-background separability assumption of fringe projection images. Based on this assumption, singular value decomposition can be used to separate the fringe projection image into a background fringe component image and a foreground object component image. The separated background region is then processed to obtain the actual phase of the object. This new method based on this assumption exhibits high detection accuracy, stronger stability compared to existing methods, and is less affected by the object's own phase, thus yielding a more accurate object phase. This detection method can remove nonlinear carrier phase without human intervention and with limited training data, demonstrating high applicability and flexibility.

[0118] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A nonlinear carrier phase removal method for striped projection images, characterized in that, A dual-camera fringe imaging system with a planar background is applied to obtain fringe projection images with a first and a second viewpoint having a nonlinear carrier wave; the process includes the following steps: (1) The stripe projection images of the first view and the second view are fused using an image stitching algorithm to obtain a global fused stripe projection image with a nonlinear carrier. (1-1) Based on the imaging of the first and second cameras onto the same mark, calibrate the first and second cameras to obtain their intrinsic and extrinsic parameters, and calculate the coordinate mapping function between the first and second cameras; the intrinsic parameters include the lens focal length and the coordinates of the camera image center; the extrinsic parameters include the rotation matrix of the camera in the world coordinate system. and translation vector ; After calibrating the intrinsic and extrinsic parameters of the first and second cameras in the world coordinate system, the affine transformation relationship between the first and second cameras and the world coordinate system was established, and the mapping relationship between the coordinates of the first and second cameras, i.e., the mapping function, was obtained. (1-2) The stripe projection images from the first and second perspectives are mapped to the same coordinate system using the coordinate mapping function between the first and second cameras obtained in step (1-1). The stripe projection images from the first and second perspectives are then registered based on the similarity of local image features at the superpixel level and fused into a globally fused stripe projection image with a nonlinear carrier. The superpixel is a grid centered on a pixel. The local image features are the geometric or non-set features of the superpixel. (2) The global fused stripe projection image with nonlinear carrier obtained in step (1) is subjected to singular value decomposition, and the first principal component is used as the background component and the other components are used as the foreground components. (3) Detect the foreground and background regions of the globally fused fringe projection image based on the foreground component obtained in step (2), and obtain the background regions of the first and second views respectively based on the pixel mapping relationship between the fringe projection images obtained from the first and second views and the globally fused fringe projection image. These are denoted as... and ; (4) The first-view background area obtained in step (3) Polynomial surface fitting is performed to obtain the fitted phase as the nonlinear phase of the first perspective; The background area obtained from the second perspective in step (3) Polynomial surface fitting is performed to obtain the fitted phase as the nonlinear phase of the second perspective; (5) After phase unfolding the stripe projection images of the first and second perspectives obtained in step (1), subtract their respective carrier phases and the respective nonlinear phases obtained in step (4) to obtain the phase of the object under test after removing the nonlinear carrier.

2. The nonlinear carrier phase removal method for striped projection images as described in claim 1, characterized in that, The dual-camera stripe imaging system includes a grating projector, a first and a second camera, and a planar display panel; the main optical axis of the grating projector is perpendicular to the planar display panel; the first and second cameras are symmetrical about the main optical axis and have the same imaging range; the imaging range is located within the planar display panel. The grating projector projects a nonlinear carrier wave onto a flat plate on which the object to be tested is placed. The first and second cameras image the flat plate on which the object to be tested is placed, respectively obtaining striped projection images with a first and a second viewpoint of nonlinear carrier wave.

3. The nonlinear carrier phase removal method for striped projection images as described in claim 1, characterized in that, Step (2) specifically includes: (2-1) For the globally fused fringe projection image I ∈ ℝ m×n Normalization is performed to obtain a normalized image matrix. : ； in, The minimum gray level in the globally fused fringe projection image I. The maximum grayscale value in the globally fused striped projection image I; (2-2) The normalized image matrix obtained in step (2-1) Perform singular value decomposition: ； Where, U ∈ ℝ m×m Let V be a left matrix, ∈ ℝ n×n Let S be a right matrix, ∈ ℝ m×n It is a singular value matrix; (2-3) Based on the singular value decomposition matrix obtained in step (2-2), calculate the first principal component of the globally fused fringe projection image. As the background component, the other components are treated as foreground components. : ； in, : indicates the maximum principal component, and : indicates taking all rows or columns of the matrix; Calculate the other components of the globally fused stripe projection image. ; ； in, It is min(m, n), that is, k takes the minimum value of m and n.

4. The nonlinear carrier phase removal method for striped projection images as described in claim 1, characterized in that, Step (3) specifically includes: (3-1) Take the components obtained in step (2) Filtering and noise reduction are performed, followed by normalization to obtain the normalized background component. 3×3 Gaussian filtering is preferred; (3-2) Normalized background components obtained in step (3-1) Use a preset threshold Binarization is performed to obtain the binarized image. : ； in, The image after binarization. For threshold; (3-3) The binarized image obtained in step (3-2) Filtering and denoising are performed to obtain the denoised binarized image. The preferred method is to perform a 5×5 median filter, i.e.: ； in This represents median filtering with a window size of a×a; (3-4) Obtain the binarized image from step (3-3). Perform object region identification to obtain the binarized image region of the object to be tested. , , , ),in The upper boundary of the object to be measured. The lower boundary of the object to be measured. The left boundary of the object to be measured. The right boundary of the object to be measured; (3-5) For the background area obtained in step (3-4) Based on the pixel mapping relationship between the fringe projection images obtained from the first and second viewpoints and the globally fused fringe projection image, the background regions of the first and second viewpoints are obtained, and are denoted as follows: and .

5. The nonlinear carrier phase removal method for striped projection images as described in claim 4, characterized in that, The threshold mentioned in step (3-2) The following steps are used to adaptively determine: S1, calculate normalization. histogram The histogram divides [0, 1] into 100 regions; S2, for After reversing the sequence, find the index corresponding to the first minimum point after the first maximum point, and map the index to a real number in the range [0,1]. m To enhance robustness, the threshold is set to = m +0.

05.

6. The nonlinear carrier phase removal method for striped projection images as described in claim 4, characterized in that, Steps (3-4) are as follows: For binarized images Morphological etching was performed to obtain etched images. A 5×5 matrix structuring element is preferably used for morphological erosion, and the eroded image is then processed. Summing the rows and columns yields and ;calculate average and traverse ,Will As a condition for determining the upper boundary of an object's region, As a condition for determining the lower boundary of an object region, the upper boundary of the object region is obtained. and lower boundary ; calculate average and traverse ,Will As a condition for determining the left boundary of an object's region, As a condition for determining the lower boundary of an object region, the left boundary of the object region is obtained. and right boundary The object region is obtained by combining the results. , , , ), and background area .

7. The nonlinear carrier phase removal method for striped projection images as described in claim 1, characterized in that, Step (4) is as follows: (4-1) For the background area in the first or second perspective , , midpoint The phase is extracted and unfolded using the phase extraction method. Removing the carrier phase yields the nonlinear phase. : ； in, The fringe frequency of the fringe projection image is determined by the imaging system; (4-2) Nonlinear phase for the first and second perspectives , Based on the background area obtained in step (4-1) from the second perspective , nonlinear phase , Polynomial surfaces using the least squares method Fitting, as nonlinear phase of the first and second perspectives , .

8. The nonlinear carrier phase removal method for striped projection images as described in claim 7, characterized in that, Step (4-2) uses a third-order bivariate polynomial for surface fitting: ； Substitute data , Fitting the polynomial using the least squares method coefficient The fitting result is used as the nonlinear phase. .

9. A nonlinear carrier phase removal system for striped projection images, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the nonlinear carrier phase removal method for striped projection images as described in any one of claims 1 to 8.

Images