Absolute Phase Recovery Method Based on Non-Uniform Phase Shift Optimization of Dual-Frequency Composite Fringe Projection

By using a dual-frequency fringe projection method optimized by non-uniform phase shift, the problem of insufficient measurement speed and accuracy in traditional methods is solved. This method enables high-precision three-dimensional measurement with only 5 fringe patterns, effectively suppressing harmonic crosstalk between high and low frequency channels.

CN123111136APending Publication Date: 2026-11-17TIANJIN UNIVERSITY OF TECHNOLOGY
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202610990999.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-07-04
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing dual-frequency composite fringe projection technology struggles to balance measurement speed and accuracy due to nonlinear response and spectral aliasing effects. Traditional methods require multiple projection fringe patterns and lack adaptive harmonic energy suppression, making it impossible to effectively suppress harmonic crosstalk between high and low frequency channels under conditions of less than 8 phase shifts.

Method used

A dual-frequency fringe projection method based on non-uniform phase shift optimization is adopted. By constructing a cost function with zero-point constraint of frequency transfer function and weighted penalty of harmonic energy, the optimal non-uniform phase shift sequence is adaptively solved. Only 5 fringe patterns are needed to achieve high and low frequency demodulation. Combined with alternating iterative optimization algorithm and rotation invariance principle, a physical projection sequence containing 5 independent phase shifts is generated.

Benefits of technology

While maintaining high precision, the number of projected patterns was reduced, the measurement speed was improved, harmonic noise was effectively suppressed, and the signal-to-noise ratio of the wrapped phase was increased.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123111136A_ABST
    Figure CN123111136A_ABST
Patent Text Reader

Abstract

This invention relates to a dual-frequency composite fringe projection absolute phase recovery method based on non-uniform phase shift optimization. The steps are as follows: S1: Construct a dual-frequency anti-crosstalk joint optimization model, define a target frequency response set covering high-frequency and low-frequency signals, set high-frequency demodulation targets and low-frequency demodulation targets respectively, construct a weighted loss function, and strengthen the suppression of crosstalk frequencies through weight coefficients; S2: Use an alternating iterative optimization algorithm to solve for the optimal non-uniform phase shift sets for high and low frequencies, perform sequence alignment based on rotation invariance, and perform phase shift calibration on the obtained high-frequency and low-frequency sequences; S3: Obtain a physical projection set containing 5 independent phase shifts, generate 5 non-uniform phase shift fringe images and project them, and use a camera to acquire the corresponding 5 deformed fringe images; S4: Demodulate the acquired images, obtain the high-frequency wrapped phase and the low-frequency wrapped phase respectively, and unfold them to obtain the absolute phase.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of optical three-dimensional measurement technology, specifically relating to an absolute phase recovery method based on dual-frequency composite fringes. Background Technology

[0002] Fringe projection profilometry, due to its non-contact, full-field measurement, high resolution, and high precision, has been widely used in reverse engineering, quality inspection, biomedical imaging, and micro / nano manufacturing. Its core principle is to project a sinusoidal fringe pattern onto the surface of an object, use a camera to capture the deformed fringes modulated by the object's height, and then recover the object's three-dimensional shape using phase demodulation technology.

[0003] To overcome the ambiguity caused by phase truncation, the acquired wrapped phase needs to be unfolded to obtain the absolute phase of the object. Currently, phase unwrapping methods are mainly divided into two categories: spatial phase unwrapping and temporal phase unwrapping. Temporal phase unwrapping algorithms introduce auxiliary frequencies into the time series, independently calculate the phase of each pixel, and complete stereo matching, effectively avoiding error accumulation and possessing high practicality. The two most widely used temporal phase unfolding methods are Gray code and multi-frequency heterodyne.

[0004] However, both of the aforementioned algorithms suffer from an unavoidable problem: they require a large number of projected fringe patterns, sacrificing measurement efficiency. Dual-frequency composite fringe projection technology, by reusing temporal or spatial resources within the same physical frame, can simultaneously acquire high and low frequency phase information with only a few phase shift steps, theoretically significantly reducing the number of projected patterns. However, in practical applications, this method is limited by the inherent nonlinear response of digital micromirror devices (such as the gamma effect), resulting in severe non-sinusoidal distortion and the inclusion of high-order harmonics in the projected fringes, leading to reduced measurement accuracy.

[0005] In recent years, many methods have been developed to eliminate nonlinear harmonic errors in fringe projection profilometry. Traditional methods, such as increasing the number of phase shift steps, use uniform sampling to bring the frequency transfer function to zero at specific higher harmonics, thereby filtering out harmonic interference. However, this method is highly dependent on increasing the number of phase shift steps, severely sacrificing measurement speed. Furthermore, this method cannot eliminate inter-channel crosstalk in composite frequency fringes, resulting in limited reconstruction accuracy in high-precision microscopy. A dual-frequency time-division multiplexing method based on a guard band mechanism addresses this by increasing the ratio of high- and low-frequency time carrier frequencies... M Set as M With a frequency ≥3, the fundamental frequency and second harmonic are physically isolated in the frequency spectrum, thus achieving crosstalk-free demodulation using eight fringe patterns. This method effectively solves the harmonic crosstalk problem in dual-frequency composite fringes and ensures the robustness of phase demodulation.

[0006] However, this method requires 8 phase shifts, resulting in a large number of projected patterns. Furthermore, it lacks adaptive weighted suppression of harmonic energy; when the system exhibits strong nonlinearity or lighting fluctuations, residual harmonic responses still reduce measurement accuracy, and it cannot achieve effective anti-crosstalk demodulation with fewer than 8 steps.

[0007] In summary, current dual-frequency composite fringe projection technology is limited by nonlinear response and spectral aliasing effects. Traditional uniform phase shift methods cannot suppress harmonic crosstalk between high and low frequency channels under phase shift conditions of less than 8 steps, thus making it difficult to balance speed and accuracy.

[0008] The following publicly disclosed existing patent technologies were found through a search: A composite fringe phase-shifting demodulation method, apparatus, and related medium (application number 202311060480.6) are disclosed. The method includes: superimposing two sets of different frequency fringes required for dual-frequency heterodyne into a single fringe pattern using a cosine function; generating and projecting only four composite fringe patterns according to a four-step phase-shifting formula; directly solving two sets of enclosed phases using pixel-level independent equations; and finally completing phase unfolding based on the multi-frequency heterodyne principle. This method replaces additional Gray codes or multiple projections of a second set of frequencies with mathematical decoupling, significantly reducing the number of projected patterns and significantly improving measurement efficiency while maintaining the advantage of error-free time-phase unfolding.

[0009] A method for obtaining the phase required for 3D measurement using frequency reuse (application number 202411720530.3) includes: creating multi-frequency fringe patterns at various phase shift angles according to the four-step phase shift three-frequency heterodyne requirement; merging different frequency fringe patterns at specific phase shift angles to create a composite fringe pattern; and extracting the phase shift of each frequency by directly subtracting the composite patterns with a phase shift difference of π, thereby skipping the inverse trigonometric function calculation and information loss steps in the traditional composite fringe separation process. Finally, three-frequency truncated phase acquisition and continuous phase unfolding are completed using eight composite patterns. This method effectively avoids high-frequency information loss while further compressing the number of projection frames, thus improving the accuracy of high-speed 3D measurement.

[0010] A three-dimensional imaging method and system for phase unwrapping assisted by color Gray code encoding and its application (202510768560X) involves constructing two sets of sinusoidal fringe patterns with different frequencies, calculating the wrapped phase using the phase shift formula, and then unwrapping the wrapped phase using the multi-frequency heterodyne method.

[0011] Upon comparison, the methods used in this patent application and the prior art documents are significantly different. Summary of the Invention

[0012] The purpose of this invention is to overcome the low measurement efficiency and lack of adaptive weighted suppression of harmonic energy that require multiple projected fringe patterns. It provides a dual-frequency fringe projection absolute phase recovery method based on non-uniform phase shift optimization. By constructing a cost function of zero-point constraint of frequency transfer function and weighted penalty of harmonic energy, the optimal non-uniform phase shift sequence is adaptively solved. Demodulation of different frequency fringes can be achieved using only 5 fringe patterns.

[0013] The present invention adopts the following technical solution to achieve the above objectives: An absolute phase recovery method for dual-frequency composite fringe projection based on non-uniform phase shift optimization is characterized by the following steps: S1: Construct a dual-frequency anti-crosstalk joint optimization model, define the target frequency response set, covering high-frequency signal frequency and low-frequency signal frequency, set high-frequency demodulation target and low-frequency demodulation target respectively, that is, the gain at high-frequency signal frequency and low-frequency signal frequency, construct a weighted loss function, and strengthen the suppression of crosstalk frequency through weight coefficients; S2: Using an alternating iterative optimization algorithm, solve for the optimal set of non-uniform phase shifts at high and low frequencies. Each set contains 3 phase shifts; S3: Based on rotation invariance, sequence alignment is performed. The high-frequency sequence and low-frequency sequence obtained by S2 iterative optimization are phase-shifted and calibrated to eliminate redundant phase shifts. S4: Construct a five-step physical projection sequence, and for the phase shifts that were eliminated in S3, obtain a physical projection set containing five independent phase shifts. ; S5: Based on Five non-uniform phase-shifted fringe images are generated and projected, and five corresponding deformed fringe images are acquired using a camera. ; S6: Using the weight vector calculated in S2, demodulate the acquired image to obtain the high-frequency encapsulation phase. and low-frequency wrap phase ; S7: Uses a dual-frequency hierarchical algorithm, leveraging... and Perform phase unfolding to obtain the absolute phase. .

[0014] Furthermore, in S1, the high-frequency demodulation target is set as follows: the gain is 1 at the high-frequency signal frequency and the gain is 0 at the low-frequency signal frequency; the low-frequency demodulation target is set as follows: the gain is 1 at the low-frequency signal frequency and the gain is 0 at the high-frequency signal frequency, specifically manifested as follows: For high-frequency demodulators, targeting the high-frequency fundamental frequency component Set the target gain to 1; for interference frequencies including , , , and Set the target gain to 0; For low-frequency demodulators, targeting the low-frequency fundamental component Set the target gain to 1; for interference frequencies including , , , and Set the target gain to 0; For normalized frequency, This represents the ratio of the time carrier frequency to the high-frequency frequency.

[0015] Furthermore, the specific steps for constructing the weighted loss function are as follows: Set the target frequency response of the high-frequency demodulator respectively and the target frequency response of the low-frequency demodulator To construct high-frequency weighted loss functions respectively. and low-frequency weighted loss function :

[0016]

[0017] in, For a set containing the frequencies, and These are the target gain vectors for the high-frequency and low-frequency components, respectively. It is a frequency weight, and it is a high-frequency crosstalk component. Assign a higher weight to the frequencies than to other frequencies. and The calculation formula is:

[0018]

[0019] in and The first n Phase shift Phase shift The corresponding high-frequency and low-frequency complex demodulation weighting coefficients, It is a complex exponential phase factor. N The total number of steps for phase measurement, the and The calculation formula is:

[0020]

[0021] in , , For matrix The algebraic cofactor, , , For matrix The algebraic cofactor, It is a high-frequency demodulation matrix. For low-frequency demodulation matrix, the and The calculation formula is:

[0022]

[0023] Furthermore, the alternating iterative optimization calculation process of S2 is as follows: The solution to the optimal non-uniform phase shift set for high and low frequencies Each set contains 3 phase shifts, i.e. , Randomly generated N One initial phase shift; Based on the current phase shift High-frequency demodulation matrices were constructed respectively. and low-frequency demodulation matrix ; Using the constructed high-frequency demodulation matrix and low-frequency demodulation matrix, calculate the high-frequency complex demodulation weighting coefficients. and low-frequency complex demodulation weighting coefficients ; Calculate the target frequency of the high-frequency demodulator and the target frequency of the low-frequency demodulator The phase shift was finely adjusted one by one using an unconstrained optimization algorithm. until the weighted loss function and They converge simultaneously.

[0024] Furthermore, the S3 sequence alignment process based on rotation invariance is as follows: The set of high-frequency phase shifts obtained through alternating iterative optimization of S2 and low-frequency phase shift set ; Select As a reference sequence, the phase offset calibration amount is calculated. ; For the set of high-frequency phase shifts Perform an overall phase shift to generate a calibrated set of high-frequency phase shifts. This ensures that the calibrated high-frequency initial phase shift is equal to the low-frequency initial phase shift. Eliminate overlapping redundant phase shifts.

[0025] Furthermore, in S5, the physical projection set is based on 5 independent variables. Generate composite stripe pattern For the nth step The light intensity of the fringe pattern is represented as:

[0026] in, For the projector pixel coordinates, For high-frequency component modulation, For low-frequency component modulation, and These are the stripe frequencies for high and low frequencies, respectively.

[0027] Furthermore, the process of performing wrapper phase demodulation in S6 is as follows: Let the number of collected data be... n Amplitude phase-shifted image at pixel The gray value at that location is Multiply the gray value by the corresponding high-frequency complex demodulation weighting coefficient. and low-frequency complex demodulation weighting coefficients The complex summation was then performed to obtain the complex light intensity values ​​for both high and low frequencies. in, ; ; Using complex light intensity values real part and the virtual part The high-frequency enveloping phase was calculated using the arctangent function. .

[0028] Furthermore, the dual-frequency hierarchical algorithm process used in S7 is as follows: Through formula Calculate the order of high-frequency stripes The formula calculates This refers to the compensation required during high-frequency phase demodulation. Integer period, of which This represents the rounding function. and These are the stripe frequencies for high and low frequencies, respectively. Then through the formula Recovering the absolute phase yields the final phase distribution that continuously corresponds to the object's shape, where... The order phase of high-frequency stripes.

[0029] On the other hand, the present invention also provides a computer-readable storage medium for storing a computer program, which, when executed by a processor, implements the aforementioned dual-frequency composite fringe projection absolute phase recovery method based on non-uniform phase shift optimization.

[0030] On the other hand, the present invention also provides a dual-frequency fringe projection absolute phase recovery system based on non-uniform phase shift optimization, comprising: Digital projectors are used to project images with non-uniform phase-shifted fringes. Industrial cameras are used to capture images of deformed stripes; A computer, connected to a digital projector and an industrial camera respectively, includes: a memory for storing computer programs; and a processor for executing the computer programs to implement the aforementioned dual-frequency composite fringe projection absolute phase recovery method based on non-uniform phase shift optimization.

[0031] Compared with the prior art, the present invention has the following beneficial effects: To address the issues of low efficiency and motion artifacts caused by the traditional dual-frequency phase-shift method for composite stripes requiring the projection of eight or more images, this invention utilizes the principle of overall rotational invariance of phase shifts to fuse two independently optimized phase shift sequences into a physical projection sequence containing five independent phase shifts. While maintaining the high accuracy of dual-frequency demodulation, only five images need to be projected to complete the measurement, thus improving the measurement speed.

[0032] To address the shortcomings of existing uniform phase shift methods that cannot effectively suppress harmonic crosstalk between high and low frequency channels within 5 steps, this invention constructs a dual-frequency anti-crosstalk joint optimization model. Target frequency responses are set for the high and low frequency demodulators respectively, and a weighted loss function is used to strengthen the penalty for crosstalk frequencies. Combined with an alternating iterative optimization algorithm, the optimal non-uniform phase shift sequence is adaptively solved, effectively filtering out harmonic noise and improving the signal-to-noise ratio of the wrapped phase. Attached Figure Description

[0033] Figure 1 This is an overall flowchart of the dual-frequency composite fringe projection absolute phase recovery method in this embodiment of the invention; Figure 2 This is a schematic diagram comparing the frequency response of non-uniform phase shift and traditional uniform phase shift in an embodiment of the present invention; Figure 3 This is an absolute phase result image obtained by using the method of the present invention in an embodiment of the present invention; Figure 4 This is the absolute phase error map recovered using the method of the present invention in an embodiment of the present invention. Detailed Implementation

[0034] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0035] An absolute phase recovery method for dual-frequency composite fringe projection based on non-uniform phase shift optimization is described in the following flowchart: Figure 1 As shown. The overall process of this method is as follows: Step 1: Set up a dual-frequency composite stripe projection measurement system.

[0036] A fringe projection measurement system is constructed, consisting of a digital projector (DLP), an industrial camera, a computer, and the object under test. The projector and camera are fixed in place, and their relative positions and lens focal lengths are adjusted to ensure the camera can clearly capture the fringe image projected onto the surface of the object. The computer is connected to both the projector and camera to control the fringe projection and image acquisition. In this embodiment, a frequency ratio between high and low frequencies is set. M =3, meaning the high-frequency fringe frequency is 3 times the low-frequency fringe frequency.

[0037] Step 2: Construct a dual-frequency anti-crosstalk joint optimization model and solve for the optimal non-uniform phase shift sequence.

[0038] The core of this step is to optimize the algorithm to solve for a set of three non-uniform phase shifts used for demodulating the high-frequency enveloping phase. and a set of 3-step non-uniform phase shifts for demodulating low-frequency wrapping phase. .

[0039] Specifically, the target frequency response of the high-frequency demodulator is first set. At the normalized frequency k=3 (i.e., the high-frequency fundamental frequency component), the target gain is set to 1; at the main interference frequencies, including the DC component (k=0) and the low-frequency fundamental frequency component... Low-frequency second harmonic negative frequency and high-frequency second harmonic Set the target gain to 0. Similarly, set the target frequency response of the low-frequency demodulator and... The gain is 1 at k=1 (low-frequency fundamental frequency). The gain is 0 at the (high-frequency base frequency) and other interference frequencies.

[0040] Then construct the weighted loss function. and :

[0041]

[0042] in, Include Key frequencies, and This represents the target gain vector. Specifically, to suppress interference from high-frequency components, [the vector is defined as follows]... Weight at point Set to 10, which is significantly higher than the weight of other frequencies (set to 1).

[0043] and The calculation formula is:

[0044]

[0045] in and The first n Phase shift Phase shift The corresponding high-frequency and low-frequency complex demodulation weighting coefficients, It is a complex exponential phase factor. N The total number of steps for phase measurement, the and The calculation formula is:

[0046]

[0047] in , , For matrix The algebraic cofactor, , , For matrix The algebraic cofactor, It is a high-frequency demodulation matrix. For low-frequency demodulation matrix, the and The calculation formula is:

[0048]

[0049] An alternating iterative optimization algorithm is used to solve the problem. Taking high-frequency sequence optimization as an example: 1. Randomly generate the initial 3-step phase shift, for example... radian.

[0050] 2. Based on the current phase shift Construct a high-frequency demodulation matrix .

[0051] 3. Based on the constructed high-frequency demodulation matrix Calculate the complex demodulation weights .

[0052] 4. Calculate the current frequency response using the formula. This leads to the loss function. .

[0053] 5. Fine-tune each step using an unconstrained optimization algorithm. Repeat steps 2-4 until... convergence.

[0054] 6. Similarly, using a low-frequency demodulation matrix Repeat the above process to optimize and obtain the low-frequency phase shift. .

[0055] After optimization, the typical non-uniform phase shift values ​​obtained in this embodiment are as follows: radian. Figure 2 The comparison of frequency response between the traditional uniform phase shift and the non-uniform phase shift optimized by this method is shown, demonstrating that the proposed method has advantages in frequency response. A deeper "zero trap" is formed, which significantly enhances the ability to suppress high-frequency component interference.

[0056] Step 3: Sequence alignment and fusion based on rotation invariance.

[0057] Because the phase shifts are rotationally invariant—meaning that all phase shifts increase by the same constant simultaneously—the demodulation result remains unchanged. Therefore, independently optimized high-frequency sequences can be demodulated. and low frequency sequences Align on the phase reference to fuse into fewer independent projection steps.

[0058] Select As a reference sequence, the calibration quantity is calculated. The high-frequency sequence was then shifted as a whole: .

[0059] After removing duplicate phase shifts, the merged result is a physical projection set containing 5 independent phase shifts. radian.

[0060] Step 4: Generate and project a non-uniform phase-shifted fringe pattern.

[0061] Based on the 5-step non-uniform phase shift after fusion Generate composite stripe pattern For the nth step The intensity distribution of the fringe pattern can be represented as:

[0062] in, For the projector pixel coordinates, For high-frequency component modulation, For low-frequency component modulation, and These are the stripe frequencies for high and low frequencies, respectively.

[0063] Step 5: Acquire images of deformed stripes.

[0064] The projector sequentially projects the aforementioned 10 fringe images onto the surface of the object under test, while simultaneously triggering the camera to acquire the corresponding deformed fringe images. The camera acquires the first... n The deformed stripe pattern is denoted as ,in These are the camera pixel coordinates.

[0065] Step 6: Wrap phase demodulation.

[0066] Take demodulating low-frequency wrapper phase as an example. For each pixel in the camera image... The grayscale values ​​of the first three deformed stripe images to The low-frequency complex demodulation weights obtained in step 2 Perform a weighted summation of complex numbers:

[0067] in, The weight coefficients obtained in the nth step of optimization (n=0~2) are then used through the arctangent function. Calculate the low-frequency wrap phase of this pixel. : Similarly, high-frequency complex demodulation weights are used. High-frequency deformed fringe pattern (n=1,4,5) undergo the same processing, and the high-frequency wrapping phase is obtained through the arctangent function. Due to the adoption of anti-crosstalk optimization, The contamination from low-frequency components has been largely eliminated. The contamination of high-frequency components has also been largely eliminated.

[0068] Step 7: Dual-frequency layered phase expansion.

[0069] After obtaining the high- and low-frequency wrapped phases, a dual-frequency layering algorithm is used for phase unwrapping. The order of the high-frequency fringes... It can be calculated using the following formula:

[0070] in, This represents the rounding function. After obtaining the order, the high-frequency absolute phase can be recovered. :

[0071] This absolute phase This is the phase distribution that is ultimately required and corresponds continuously to the shape of the object.

[0072] This embodiment uses the above method to conduct a simulation experiment. Figure 3 The recovered absolute phase result is shown. Experimental results demonstrate that this invention can achieve phase demodulation using only five composite frequency fringe patterns. Furthermore, based on an eight-step uniform phase shift capable of resisting second harmonics, Figure 4 The phase error of the non-uniform phase shift algorithm is shown. As can be seen from the figure, the average absolute phase error after processing is suppressed to within 0.15 rad, which proves the effectiveness of the present invention.

[0073] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any changes, modifications, substitutions, combinations, or simplifications made without departing from the spirit and principle of the present invention shall be considered equivalent substitutions and shall be included within the protection scope of the present invention.

Claims

1. A dual-frequency composite fringe projection absolute phase recovery method based on non-uniform phase shift optimization, characterized in that: Includes the following steps: S1: Construct a dual-frequency anti-crosstalk joint optimization model, define the target frequency response set, covering high-frequency signal frequency and low-frequency signal frequency, set high-frequency demodulation target and low-frequency demodulation target respectively, that is, the gain at high-frequency signal frequency and low-frequency signal frequency, construct a weighted loss function, and strengthen the suppression of crosstalk frequency through weight coefficients; S2: Using an alternating iterative optimization algorithm, solve for the optimal set of non-uniform phase shifts at high and low frequencies. Each set contains 3 phase shifts; S3: Based on rotation invariance, sequence alignment is performed. The high-frequency sequence and low-frequency sequence obtained by S2 iterative optimization are phase-shifted and calibrated to eliminate redundant phase shifts. S4: Construct a five-step physical projection sequence, and for the phase shifts that were eliminated in S3, obtain a physical projection set containing five independent phase shifts. ; S5: Based on Five non-uniform phase-shifted fringe images are generated and projected, and five corresponding deformed fringe images are acquired using a camera. ; S6: Using the weight vector calculated in S2, demodulate the acquired image to obtain the high-frequency encapsulation phase. and low-frequency wrap phase ; S7: Uses a dual-frequency hierarchical algorithm, leveraging... and Perform phase unfolding to obtain the absolute phase. .

2. The absolute phase recovery method for dual-frequency composite fringe projection based on non-uniform phase shift optimization according to claim 1, characterized in that: In step S1, the high-frequency demodulation target is set to a gain of 1 at the high-frequency signal frequency and a gain of 0 at the low-frequency signal frequency; the low-frequency demodulation target is set to a gain of 1 at the low-frequency signal frequency and a gain of 0 at the high-frequency signal frequency, specifically as follows: For high-frequency demodulators, targeting the high-frequency fundamental frequency component Set the target gain to 1; for interference frequencies including , , , and Set the target gain to 0; For low-frequency demodulators, targeting the low-frequency fundamental component Set the target gain to 1; for interference frequencies including , , , and Set the target gain to 0; For normalized frequency, This represents the ratio of the time carrier frequency to the high-frequency frequency.

3. The absolute phase recovery method for dual-frequency composite fringe projection based on non-uniform phase shift optimization according to claim 2, characterized in that: The specific steps for constructing the weighted loss function are as follows: Set the target frequency response of the high-frequency demodulator respectively and the target frequency response of the low-frequency demodulator To construct high-frequency weighted loss functions respectively. and low-frequency weighted loss function : , , in, For a set containing the frequencies, and These are the target gain vectors for the high-frequency and low-frequency components, respectively. It is a frequency weight, and it is a high-frequency crosstalk component. Assign a higher weight to the frequencies than to other frequencies. and The calculation formula is: , , in and The first n Phase shift Phase shift The corresponding high-frequency and low-frequency complex demodulation weighting coefficients, It is a complex exponential phase factor. N The total number of steps for phase measurement, the and The calculation formula is: , ,in , , For matrix The algebraic cofactor, , , For matrix The algebraic cofactor, It is a high-frequency demodulation matrix. For low-frequency demodulation matrix, the and The calculation formula is: , 。 4. The absolute phase recovery method for dual-frequency composite fringe projection based on non-uniform phase shift optimization according to claim 3, characterized in that: The alternating iterative optimization algorithm process of S2 is as follows: The solution to the optimal non-uniform phase shift set for high and low frequencies Each set contains 3 phase shifts, i.e. , Randomly generated N One initial phase shift; Based on the current phase shift High-frequency demodulation matrices were constructed respectively. and low-frequency demodulation matrix ; Using the constructed high-frequency demodulation matrix and low-frequency demodulation matrix, calculate the high-frequency complex demodulation weighting coefficients. and low-frequency complex demodulation weighting coefficients ; Calculate the target frequency of the high-frequency demodulator and the target frequency of the low-frequency demodulator The phase shift was finely adjusted one by one using an unconstrained optimization algorithm. Until the weighted loss function and They converge simultaneously.

5. The absolute phase recovery method for dual-frequency composite fringe projection based on non-uniform phase shift optimization according to claim 4, characterized in that: The S3 sequence alignment process based on rotation invariance is as follows: The set of high-frequency phase shifts obtained through alternating iterative optimization of S2 and low-frequency phase shift set ; Select As a reference sequence, the phase offset calibration amount is calculated. ; For the set of high-frequency phase shifts Perform an overall phase shift to generate a calibrated set of high-frequency phase shifts. This ensures that the calibrated high-frequency initial phase shift is equal to the low-frequency initial phase shift. Eliminate overlapping redundant phase shifts.

6. The absolute phase recovery method for dual-frequency composite fringe projection based on non-uniform phase shift optimization according to claim 5, characterized in that: In S5, the physical projection set is based on 5 independent variables. Generate composite stripe pattern For the nth step The intensity of the fringe pattern is represented as: ,in, For the projector pixel coordinates, For high-frequency component modulation, For low-frequency component modulation, and These are the stripe frequencies for high and low frequencies, respectively.

7. The absolute phase recovery method for dual-frequency composite fringe projection based on non-uniform phase shift optimization according to claim 6, characterized in that: The process of S6 performing wrapper phase demodulation is as follows: Let the number of collected data be... n Amplitude phase-shifted image at pixel The gray value at that location is Multiply the gray value by the corresponding high-frequency complex demodulation weighting coefficient. and low-frequency complex demodulation weighting coefficients The complex summation was then performed to obtain the complex light intensity values ​​for both high and low frequencies. in, ; ; Using complex light intensity values real part and the virtual part The high-frequency enveloping phase was calculated using the arctangent function. .

8. The absolute phase recovery method for dual-frequency composite fringe projection based on non-uniform phase shift optimization according to claim 7, characterized in that: The dual-frequency hierarchical algorithm used in S7 is as follows: Through formula Calculate the order of high-frequency stripes The formula calculates This refers to the compensation required during high-frequency phase demodulation. Integer period, of which This represents the rounding function. and These are the stripe frequencies for high and low frequencies, respectively; Then through the formula Recovering the absolute phase yields the final phase distribution that continuously corresponds to the object's shape, where... The order phase of high-frequency stripes.

9. A computer-readable storage medium, characterized in that: Used to store a computer program, which, when executed by a processor, implements the dual-frequency composite fringe projection absolute phase recovery method based on non-uniform phase shift optimization as described in any one of claims 1 to 8.

10. A dual-frequency fringe projection absolute phase recovery system based on non-uniform phase shift optimization, characterized in that, include: Digital projector used to project images with non-uniform phase-shifted fringes; Industrial cameras are used to capture images of deformed stripes; A computer, connected to a digital projector and an industrial camera respectively, the computer comprising: a memory for storing a computer program; and a processor for executing the computer program to implement the absolute phase recovery method for dual-frequency composite fringe projection based on non-uniform phase shift optimization as described in any one of claims 1 to 8.

Citation Information

Patent Citations

  • A composite fringe phase shift phase demodulation method, device and related medium

    CN117419661B

  • Method for obtaining crosstalk-free phase required by three-dimensional measurement by using frequency multiplexing

    CN119687826A