A 3D imaging method and system based on structured light phase measurement

Abstract

This application discloses a 3D imaging method and system based on structured light phase measurement, relating to the field of 3D imaging technology. The method includes: calibrating a structured light measurement system to obtain intrinsic and extrinsic parameters, and determining the effective measurement depth range and the corresponding effective absolute phase interval; controlling a projector to project a preset sinusoidal fringe pattern onto the measured scene, acquiring a deformed fringe image modulated by the surface of the measured object, and obtaining a wrapping phase map; calculating the initial fringe order of each pixel based on the effective measurement depth range using a frequency domain order-based fringe order no-neighborhood solution algorithm; performing no-neighborhood verification and error correction on the initial fringe order based on the effective absolute phase interval using an order self-verification and error correction algorithm to obtain the final fringe order; and calculating the absolute phase map of the entire field of view based on the wrapping phase map and the final fringe order, and converting the point cloud data. This application aims to solve the technical problem of existing spatial phase unfolding techniques relying on the continuity of adjacent pixels.

Description

Technical Field

[0001] This invention relates to the field of three-dimensional imaging technology, and in particular to a 3D imaging method and system based on structured light phase measurement. Background Technology

[0002] Structured light phase measurement technology is currently the mainstream non-contact 3D imaging technology. Its core principle is to project a sinusoidal fringe pattern onto the scene being measured using a projector, and a camera collects the deformed fringes modulated by the depth of the surface of the object being measured, thus calculating the wrapped phase (constrained by...). or The absolute phase is then obtained by using phase expansion technology to restore the interval, and finally the 3D coordinates are obtained by combining the triangulation model.

[0003] Phase unwrapping is the core component of this technology. Spatial phase unwrapping technology has become the preferred solution for dynamic scene 3D imaging due to its requirement of only a single frame or a few frames and its fast imaging speed. However, existing spatial phase unwrapping technologies have inherent limitations: all algorithms rely heavily on the phase continuity of adjacent pixels, with the core assumption being that "the depth change of adjacent pixels is extremely small, the corresponding absolute phase difference is less than π, and the fringe level difference is 0 or ±1".

[0004] This inherent limitation results in insurmountable defects in existing technologies: When the scene under test contains phase discontinuities such as depth jumps, shadows, high reflectivity, and noise pollution, the phase difference between adjacent pixels exceeds [a certain threshold]. This will directly lead to misjudgment of stripe order, and the error will spread along the entire neighborhood, causing large-scale 3D reconstruction errors. Unable to handle isolated, disconnected regions in the scene; isolated regions cannot obtain the correct stripe order through neighborhood growth, resulting in reconstruction failure. Existing improvement solutions either introduce multi-frequency, multi-amplitude temporal phase unfolding, sacrificing imaging speed and failing to adapt to dynamic scenarios; or they adopt deep learning solutions, which have poor generalization and lack clear physical model support, resulting in accuracy that cannot meet the needs of industrial-grade detection. Summary of the Invention

[0005] This invention provides a 3D imaging method based on structured light phase measurement, comprising: The structured light measurement system is calibrated to obtain intrinsic and extrinsic parameters, and the effective measurement depth range and the corresponding effective absolute phase interval are determined. The projector projects a preset sinusoidal fringe pattern onto the scene under test, and the camera simultaneously acquires the deformed fringe image modulated by the surface of the object under test. The wrapped phase map is then obtained after calculation. Based on the effective measurement depth range, the initial fringe order of each pixel is calculated using a frequency domain fringe order no-neighbor solution algorithm. Based on the effective range of absolute phase, the initial fringe level is checked and corrected without neighborhood using a level self-checking and error correction algorithm to obtain the final fringe level. Based on the wrapped phase map and the final fringe order, the absolute phase map of the entire field of view is calculated and converted into 3D point cloud data of the scene under test.

[0006] The aforementioned 3D imaging method based on structured light phase measurement calibrates the structured light measurement system, obtains intrinsic and extrinsic parameters, and determines the effective measurement depth range and the corresponding effective absolute phase interval, including: For a structured light measurement system consisting of a monocular camera and a monocular projector, the internal parameters of the camera and projector are calibrated to the system's external parameters. The effective measurement depth range and the corresponding effective absolute phase interval are calculated based on the system parameters obtained from the calibration.

[0007] The aforementioned 3D imaging method based on structured light phase measurement calculates the initial fringe order of each pixel according to the effective measurement depth range using a frequency domain fringe order no-neighborhood solution algorithm, including: Based on the effective measurement depth range of the system, a candidate set of stripe orders is determined; For each candidate fringe level in the candidate set, a corresponding candidate complex phase map is constructed. Based on the candidate complex phase map, the spectral energy amplitude corresponding to each pixel is calculated, and the initial fringe order is calculated in combination with the frequency domain order consensus.

[0008] The aforementioned 3D imaging method based on structured light phase measurement constructs a corresponding candidate complex phase map for each candidate fringe order in the candidate set, including: Calculate the confidence fusion factor for each pixel based on the package phase map; For each level in the candidate level set, construct the corresponding candidate complex phase graph.

[0009] The aforementioned 3D imaging method based on structured light phase measurement calculates the spectral energy amplitude corresponding to each pixel according to the candidate complex phase map, and calculates the initial fringe order by combining the frequency domain order consensus, including: Calculate the spectral energy amplitude corresponding to each pixel based on the candidate complex phase map, and calculate the fundamental order; The consensus degree of the frequency domain level is calculated based on the basic level, and the initial fringe level is determined based on the consensus degree of the frequency domain level.

[0010] The aforementioned 3D imaging method based on structured light phase measurement, according to the effective range of absolute phase, performs neighborhood-free verification and error correction on the initial fringe order through a self-verification and error correction algorithm to obtain the final fringe order, including: Based on the effective range of the absolute phase, an effective probability model is constructed to calculate the verification probability of the initial stripe level; Based on the verification probability, calculate the multi-factor joint verification index, determine the validity of the initial stripe level, and obtain the set of pixels to be corrected. For pixels in the set of pixels to be corrected, a multi-dimensional joint scoring model is constructed to perform high-precision level correction.

[0011] The aforementioned 3D imaging method based on structured light phase measurement constructs a multi-dimensional joint scoring model for high-precision level error correction of pixels in the set of pixels to be corrected, including: The multi-factor joint score for each candidate level is calculated using a multi-factor joint scoring function; Based on the multi-factor joint scoring and combined with the consensus of the frequency domain level, the pixels to be corrected are divided into ordinary pixels to be corrected and high uncertainty pixels, and differential error correction is performed to obtain the final stripe level.

[0012] A 3D imaging system based on structured light phase measurement includes: The system calibration module is used to calibrate the structured light measurement system, obtain intrinsic and extrinsic parameters, and determine the effective measurement depth range and the corresponding effective absolute phase interval. The wrapping phase calculation module is used to control the projector to project a preset sinusoidal stripe pattern onto the scene under test. The camera synchronously acquires the deformed stripe image after modulation on the surface of the object under test, and the wrapping phase map is obtained after calculation. The frequency domain order calculation module is used to calculate the initial fringe order of each pixel based on the effective measurement depth range using a frequency domain order fringe order no-neighborhood calculation algorithm. The order verification and error correction module is used to perform neighborhood-free verification and error correction on the initial fringe order based on the effective range of the absolute phase using an order self-verification and error correction algorithm, so as to obtain the final fringe order. The 3D reconstruction module is used to calculate the absolute phase map of the entire field of view based on the wrapping phase map and the final fringe order, and convert it into 3D point cloud data of the scene under test.

[0013] The beneficial effects achieved by this invention are as follows: It breaks through the strong dependence of traditional spatial phase unfolding on the phase continuity of adjacent pixels at the root, and solves the problems of stripe level misjudgment and error global spread in phase discontinuous scenes such as depth jump, shadow, and high reflectivity caused by this, without the defects of line reconstruction; it can achieve high-precision phase unfolding of isolated and unconnected areas, and its robustness to complex surface scenes far exceeds that of existing similar technologies. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0015] Figure 1 This is a flowchart of a 3D imaging method based on structured light phase measurement provided in Embodiment 1 of this application.

[0016] Figure 2 This is a schematic diagram of a 3D imaging system based on structured light phase measurement provided in Embodiment 2 of this application. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Example 1 like Figure 1 As shown, Embodiment 1 of this application provides a 3D imaging method based on structured light phase measurement, including: S1: Calibrate the structured light measurement system, obtain intrinsic and extrinsic parameters, and determine the effective measurement depth range and the corresponding effective absolute phase interval; The calibration of the structured light measurement system, obtaining intrinsic and extrinsic parameters, and determining the effective measurement depth range and the corresponding effective absolute phase interval includes the following sub-steps: S11: For a structured light measurement system consisting of a monocular camera and a monocular projector, calibrate the internal parameters of the camera and projector and the external parameters of the system. Multiple sets of chessboard calibration board images in different poses were acquired, corner coordinates were extracted, and camera intrinsic parameters and distortion coefficients were solved based on Zhang's calibration method. Camera calibration was completed after passing reprojection error verification. Keeping the calibration board pose unchanged, phase-shifted sinusoidal fringes were projected onto the calibration board, and the camera simultaneously acquired deformed fringe patterns. The absolute phase was obtained through phase shifting and phase unrolling. A pixel coordinate mapping between the projector and camera was established based on the absolute phase. Using an inverse camera model, combined with the calibrated camera intrinsic parameters and the world coordinates of the calibration board corners, the projector intrinsic parameters, distortion coefficients, and system extrinsic parameters (rotation matrix R and translation vector T) were solved. After further error verification, all calibration parameters were stored.

[0019] S12: Calculate the effective measurement depth range and the corresponding effective absolute phase interval based on the system parameters obtained from calibration.

[0020] Based on the system calibration results and the principle of triangulation, a depth measurement system was established. with absolute phase Global linear mapping model The calibration plate is moved along the optical axis to different depths, images are acquired, and 3D reconstruction is performed. The deviation between the reconstructed size and the actual size is calculated. The continuous depth range in which the reconstruction deviation meets the preset accuracy threshold is taken as the effective measurement depth range. .

[0021] Next, the boundary values ​​of the effective measurement depth range will be determined. By using the absolute phase-depth global linear mapping model, the corresponding absolute phase boundary values ​​are obtained through inverse solving; finally, the effective range of the absolute phase of the system is determined as follows: .

[0022] S2: Control the projector to project a preset sinusoidal stripe pattern onto the scene under test, and simultaneously acquire the deformed stripe image modulated by the surface of the object under test through the camera, and obtain the wrapping phase map after calculation. Based on the calibrated system parameters, a preset number of phase-shifted sinusoidal fringe patterns are pre-generated and stored in local high-speed memory. A synchronous communication link between the projector and camera is established through a hardware trigger interface, configuring the projection frame period, camera exposure duration, and trigger timing offset to ensure that the projection stabilization period is perfectly aligned with the camera exposure window. Synchronization commands and acquisition commands control the projector to project the pre-generated fringe patterns sequentially, while the camera acquires them synchronously, obtaining multi-frame deformed fringe images.

[0023] Multiple frames of deformed stripe images are read sequentially and subjected to standardized preprocessing: distortion correction is performed based on camera intrinsic parameters to eliminate optical distortion; gamma linear correction is performed to compensate for photoelectric nonlinear response; and then Gaussian smoothing filtering is applied to suppress noise, resulting in a preprocessed stripe image sequence.

[0024] Based on the preprocessed phase-shifted fringe image, the initial phase value of each pixel in the full field of view is calculated using a phase-shifting algorithm and normalized to... or Within the interval, a full-field-of-view wrap-around phase map is generated.

[0025] S3: Based on the effective measurement depth range, calculate the initial fringe order of each pixel using a frequency domain fringe order no-neighborhood solution algorithm; The initial fringe order of each pixel is calculated based on the effective measurement depth range using a frequency domain fringe order no-neighborhood solution algorithm, including the following sub-steps: S31: Determine the candidate set of stripe levels based on the effective measurement depth range of the system; Based on the effective measurement depth range Composed of absolute phase and enveloping phase Relationship ( For the fringe order, considering the phase boundary, calculate the minimum and maximum values ​​of the fringe order; thus obtaining the candidate order set. This set defines the order search space for each pixel. Then, a complex exponential mapping is performed on the wrapped phase map obtained by the four-step phase-shift algorithm to obtain the reference complex phase map.

[0026] S32: For each candidate fringe level in the candidate set, construct the corresponding candidate complex phase map; Specifically, for each candidate fringe level in the candidate set, a corresponding candidate complex phase map is constructed, including the following sub-steps: S321: Calculate the confidence fusion factor for each pixel based on the wrapper phase map; Calculation of phase quality map based on package phase map (For example, by measuring modulation or phase derivative variance), and then normalizing to obtain the quality weights. ,in This represents the maximum value across the entire image. Simultaneously, the gradient magnitude of the wrapped phase map is calculated. Construct gradient weights: ,in, This represents the gradient magnitude of the wrapper phase map; Indicates gradient weights; This is a preset gradient threshold parameter used to suppress the influence of phase transition regions.

[0027] Next, the quality weights and gradient weights are fused to obtain the confidence fusion factor, and the specific calculation formula is as follows: ,in, Indicates the reliability fusion factor; Indicates quality weight; Indicates gradient weights; The balancing coefficients ensure that reasonable weights are obtained in both high-quality continuous regions and low-quality regions with small gradients.

[0028] S322: For each level in the candidate level set, construct the corresponding candidate complex phase graph; For candidate level set Each level Construct the corresponding candidate complex phase map. Specifically, set... Bandpass filters of different scales Each corresponds to a different neighborhood bandwidth (e.g., from narrowband to broadband). For each scale and each candidate level The weighted candidate complex phase diagram is constructed using the following formula: ,in, Indicates that for the first The first scale, the first A weighted candidate complex phase map constructed from candidate fringe levels; As a reference complex phase diagram; It is a linear phase compensation term used for carrier compensation of the package phase: if the current candidate level If the phase order is equal to the true fringe order of the pixel, the compensated phase will approximate the true absolute phase, thus preserving the complete fundamental frequency modulation component in the frequency domain; if If the order does not match the true order, the compensated phase will introduce additional nonlinear transitions, which will significantly weaken the fundamental frequency component; where j is the imaginary unit. Candidate stripe order; The fundamental frequency of the projected sinusoidal fringes; The image spatial sampling frequency; is the horizontal coordinate of the pixel; As a reliability fusion factor, it enables subsequent frequency domain energy analysis to adaptively focus on regions with continuous phase and high signal-to-noise ratio, significantly improving robustness in complex scenarios such as shadows, high reflectivity, and depth transition edges.

[0029] S33: Based on the candidate complex phase map, calculate the spectral energy amplitude corresponding to each pixel, and calculate the initial fringe order by combining the frequency domain order consensus. The process involves calculating the spectral energy amplitude for each pixel based on the candidate complex phase map, and then calculating the initial fringe order by combining the frequency domain order consensus. This includes the following sub-steps: S331: Calculate the spectral energy amplitude corresponding to each pixel based on the candidate complex phase map, and calculate the fundamental order; For each scale and each candidate level Weighted candidate complex phase diagram Perform a two-dimensional Fast Fourier Transform (FFT) to obtain the spectrum. Use a bandpass filter of the corresponding scale. The spectrum is filtered to obtain the spectrum after bandpass filtering. : And then Perform a two-dimensional inverse fast Fourier transform (IFFT) to obtain the spatial complex signal. The amplitude is taken to obtain the fundamental frequency energy amplitude of each pixel at this scale and this candidate level: For each pixel At each scale Following the principle of energy maximization, the fundamental order at this scale is obtained as follows: S332: Calculate the frequency domain level consensus based on the basic level, and determine the initial fringe level based on the frequency domain level consensus.

[0030] Calculate the frequency domain level consensus using the formula: ,in, For indicator functions; The total number of scales is multi-scale, with each scale corresponding to a bandpass filter with a different bandwidth, used to capture frequency domain features in different neighborhood ranges; for The order that appears most frequently at each scale; Represents pixels At each scale Lower basic level; The frequency domain level consensus reflects the degree of consistency in the level judgments across different scales.

[0031] Next, based on the frequency domain level consensus... Classify pixels: like ( If a pre-set coordination threshold is used (e.g., 0.8), then the level determination of that pixel is considered reliable. As the initial stripe level of this pixel .

[0032] like This indicates that the pixel exhibits divergence at different scales, potentially existing in noise, shadow, or depth-abrupt regions. In this case, an energy-weighted average is chosen as the initial fringe order: ,in, Represents pixel coordinates The initial fringe order at the location; It is a rounding function; The total number of scales is multi-scale, with each scale corresponding to a bandpass filter with a different bandwidth, used to capture frequency domain features in different neighborhood ranges; Representing scale The maximum fundamental frequency energy amplitude of the pixel reflects the strength of the optimal time-frequency domain response at this scale; the larger the energy amplitude, the better the carrier compensation effect at this level, and the higher the reliability of the level judgment. Indicated in scale The optimal stripe order for this pixel is selected based on the energy maximization criterion. As a normalization factor for the weighted average, it ensures that the weighted average falls within the range of values ​​formed by each scale level.

[0033] S4: Based on the effective range of the absolute phase, the initial fringe level is checked and corrected without neighborhood using a level self-checking and error correction algorithm to obtain the final fringe level. Among them, the order self-verification and error correction algorithm utilizes the global linear mapping relationship between the absolute phase and depth after the structured light measurement system is calibrated to construct a probabilistic verification framework based on a physical model, and performs pixel-by-pixel independent validity evaluation and error correction on the initial stripe order.

[0034] The process involves performing neighborhood-free verification and error correction on the initial fringe level based on the effective range of the absolute phase, using a level self-verification and error correction algorithm to obtain the final fringe level. This includes the following sub-steps: S41: Based on the effective range of the absolute phase, construct an effective probability model to calculate the verification probability of the initial stripe level; Based on the effective interval of the absolute phase, a confidence interval extension parameter is introduced. (For example, take) Construct soft check intervals And calculate each candidate level Corresponding absolute phase The probability of belonging to the valid interval is calculated using the following formula: ,in, Indicates pixel coordinates At this point, select candidate levels. The absolute phase corresponding to the time The probability density value that falls within the effective measurement range of the system; For candidate absolute phase, the wrapping phase of the current pixel is... With candidate stripe levels Calculated using the fundamental relationship of phase expansion, it represents the absolute phase value restored under this order of assumptions; The center value of the valid interval; This is the probability transition width, used to transform the physically effective range into a smooth probability transition region, avoiding potential misjudgments that may occur at the boundaries of hard threshold judgments. ,in The confidence interval extension parameter represents the absolute phase under the combined effects of non-ideal factors such as calibration error, system noise, temperature drift, and projector-camera synchronization error. The phase offset relative to the maximum possible deviation of the theoretical linear model is used to accommodate phase fluctuations caused by non-ideal factors such as calibration errors and system noise. The confidence interval width coefficient is preset based on the system calibration accuracy and the tolerance of the measurement task for the misjudgment rate; is the Gaussian distribution normalization coefficient.

[0035] Then for each pixel Read the initial stripe level Combined with the wrap phase of this pixel Calculate the initial absolute phase The verification probability of the initial stripe level is obtained through the effective probability model. .

[0036] S42: Calculate the multi-factor joint verification index based on the verification probability, determine the validity of the initial stripe level, and obtain the set of pixels to be corrected; To further improve the reliability of the verification, the phase quality map is used as an auxiliary criterion to construct a multi-factor joint verification index, which completes the pixel validity determination and obtains the set of pixels to be corrected, as follows: Calculate the multi-factor joint validation index according to the formula: ,in, pixel coordinates The larger the value of the joint verification index at the point, the higher the overall effectiveness of the initial stripe level of the pixel, and the better it meets the system's physical constraints and signal quality requirements; These are preset weighting coefficients, used to control the weight ratio of effective probability density and phase quality map in joint verification. The specific values ​​can be adjusted according to the system calibration accuracy and the tolerance of misjudgment rate in the measurement scenario. For pixels Initial stripe order The corresponding effective probability density value, the larger the value, the higher the credibility of the initial order at the physical model level; For pixels The phase quality map at that location directly reflects the strength of the deformed fringe signal. A larger value indicates better local signal quality for that pixel and higher reliability of the phase calculation results.

[0037] Then, combined with the preset joint verification threshold (For example, Perform validity determination and labeling; if If the initial fringe level of the pixel is valid, it is directly used as the candidate final fringe level for that pixel; if If the initial stripe level of the pixel is determined to be a misjudgment, it is marked as a pixel to be corrected and added to the set of pixels to be corrected.

[0038] S43: For the pixels in the set of pixels to be corrected, construct a multi-dimensional joint scoring model to perform high-precision level correction; Next, for the pixels in the set of pixels to be corrected, a multi-dimensional joint scoring model is constructed, and a differentiated adaptive error correction strategy is adopted according to the uncertainty level of the pixels. Under the premise of ensuring no neighborhood dependency, high-precision level error correction is achieved in complex scenarios.

[0039] Specifically, for pixels in the set of pixels to be corrected, a multi-dimensional joint scoring model is constructed to perform high-precision level correction, including the following sub-steps: S431: Calculate the multi-factor joint score for each candidate level using a multi-factor joint scoring function; For the set of pixels to be corrected Each pixel in Iterate through all candidate levels in the candidate level set and calculate the multi-factor joint score for each candidate level. The scoring function is as follows: ,in, Represents pixels Candidate level Multifactor joint score; Preset weighting coefficients; For pixels The phase quality diagram reflects the local reliability of the signal; This is the soft boundary kurtosis coefficient, which controls the sharpness of the boundary transition in the physical model; Candidate level Corresponding absolute phase; To calibrate the center value of the effective phase interval of the system; For probability transition width; This is a confidence interval expansion parameter used to encompass calibration errors and system noise; This is the preset confidence interval width coefficient; For pixels The mean of the multi-scale fundamental frequency energy amplitudes corresponding to all candidate orders; Candidate level The fundamental frequency energy amplitude below; The total number of candidate levels; This is the energy ratio scale parameter, which controls the smoothness of the frequency domain energy relative intensity factor; In pixels A local reference set centered on the pixel contains only neighboring pixels that have passed validity checks; For Iverson brackets, when the neighboring pixels The final level equal to candidate level The value is 1 if it is true, and 0 otherwise. Represents neighboring pixels The phase quality map value is taken at the location; Use extremely small positive numbers to prevent the denominator from being zero.

[0040] The scoring function comprehensively evaluates candidate levels from three independent dimensions: physical model consistency, signal quality, and frequency domain feature matching degree, to ensure the reliability of the error correction results.

[0041] S432: Based on the multi-factor joint scoring and combined with the consensus of the frequency domain level, the pixels to be corrected are divided into ordinary pixels to be corrected and high uncertainty pixels, and differential error correction is performed to obtain the final stripe level.

[0042] Based on frequency domain level consensus and coordination threshold The pixels to be corrected are divided into ordinary pixels and high-uncertainty pixels, and a differentiated error correction strategy is adopted, as follows: For ordinary pixels to be corrected, that is, satisfying However, the initial fringe level was checked and found to be invalid. This type of pixel has high consistency in multi-scale frequency domain features. Only the initial fringe level does not meet the physical model constraints. Therefore, the candidate level with the highest joint score of multiple factors is directly selected as the final fringe level after error correction. For pixels with high uncertainty, i.e., satisfying The pixels to be corrected are mostly located in shadows, depth transition edges, or areas of strong noise. The frequency domain energy characteristics of a single pixel are ambiguous. A very weak prior statistical analysis of verified valid pixels is introduced for correction, as follows: First, construct a local reference set. Constructed with the pixel as the center The neighborhood is defined as follows: only pixels within the neighborhood that have been verified as valid in step S42 are included in the reference set; if the number of valid pixels is less than 3, the neighborhood is expanded to include... Continue until the number of valid pixels meets the requirement; Next, we construct the corrected scoring function, and the specific calculation formula is as follows: ,in, This indicates the corrected multi-factor joint score; For local reference set hierarchy equal to The number of effective pixels, This represents the total number of valid pixels in the local reference set. This is the neighborhood influence coefficient, used to control the strength of prior constraints; Represents pixels Candidate level Multifactor joint scoring.

[0043] The candidate level with the highest correction score was then selected as the final stripe level after error correction. Specifically, the local reference set only includes valid pixels that have passed independent physical model verification, and pixels that have not passed verification are not included in the reference range; the correction term is only a very weak statistical prior constraint, does not depend on the phase continuity assumption of adjacent pixels, does not participate in the phase difference calculation of adjacent pixels, and only counts the order distribution of valid pixels that have been independently verified. Therefore, it will not produce the problem of error spreading along the neighborhood in traditional spatial phase unfolding.

[0044] Finally, a median filtering smoothing operation is performed on the generated full-field fringe level map: for each pixel, the level distribution in its 3×3 neighborhood is statistically analyzed. If the level of the pixel is different from the mode level of the neighborhood, and the difference between the multi-factor joint score corresponding to the mode level of the neighborhood and the current level score is less than a preset threshold of 0.1, then the level of the pixel is replaced with the mode level of the neighborhood.

[0045] It should be clarified that a smoothing operation can be performed on the obtained hierarchy map to further optimize its visual consistency. It should be noted that this smoothing operation is not necessary for achieving neighborhood-free hierarchy check error correction; even without performing this smoothing operation, the core accuracy of 3D imaging can still be accurately obtained.

[0046] Finally, the initial fringe order of all valid pixels is integrated with the error correction results of the pixels to be corrected to generate the final fringe order map for the entire field of view. The final fringe order diagram has the following characteristics: All pixel levels across the entire field of view have been constrained and verified by the system's global physical model. The core verification and error correction processes are all executed independently on a pixel-by-pixel basis, without relying on the assumption of phase continuity between adjacent pixels; It can effectively restore the hierarchy of isolated and disconnected areas, shadow edges, highly reflective areas, and areas with abrupt changes in depth within the scene.

[0047] S5: Based on the wrapped phase map and the final fringe order, the absolute phase map of the entire field of view is calculated and converted into 3D point cloud data of the scene under test.

[0048] The wrapped phase map obtained in step S3 is added pixel-by-pixel to the final fringe order map obtained in step S4, and the absolute phase value of each pixel is calculated according to the phase unfolding formula to obtain the full field-of-view absolute phase map. If isolated spikes caused by noise exist in the absolute phase map, select... Median filtering is used for smoothing.

[0049] Next, based on the global linear mapping relationship between absolute phase and depth established during the system calibration phase, the absolute phase map is substituted pixel by pixel into this mapping to calculate the full field-of-view depth map. Finally, using the camera intrinsic parameters obtained from calibration (including focal length, principal point coordinates, etc.), the image coordinates of each pixel are back-projected to the depth value in the full field-of-view depth map to calculate the 3D spatial coordinates of each pixel in the camera coordinate system. All pixels are traversed, and points with invalid depth values ​​(such as those outside the effective measurement range or calibration blind zone) are removed. The remaining valid 3D coordinate points are organized into point cloud data and finally output in a standard point cloud format (such as PLY, PCD, or XYZ format) to complete the 3D imaging of the measured scene.

[0050] Example 2 like Figure 2 As shown, Embodiment 2 of this application provides a 3D imaging system based on structured light phase measurement, comprising: System calibration module 21: calibrates the structured light measurement system, obtains intrinsic and extrinsic parameters, and determines the effective measurement depth range and the corresponding effective absolute phase interval; Wrapping phase calculation module 22: controls the projector to project a preset sinusoidal stripe pattern onto the scene under test, and synchronously acquires the deformed stripe image after modulation on the surface of the object under test through the camera, and obtains the wrapping phase map after calculation; Frequency domain order calculation module 23: Based on the effective measurement depth range, calculate the initial fringe order of each pixel using a frequency domain order fringe order no-neighborhood calculation algorithm; Level verification and error correction module 24: Based on the effective range of absolute phase, the initial fringe level is verified and corrected without neighborhood using a level self-verification and error correction algorithm to obtain the final fringe level; 3D Reconstruction Module 25: Based on the wrapping phase map and the final fringe order, calculate the absolute phase map of the entire field of view and convert it into 3D point cloud data of the scene under test.

[0051] Corresponding to the above embodiments, the present invention provides a computer storage medium, including: at least one memory and at least one processor; The memory is used to store one or more program instructions; A processor for running one or more program instructions to execute a 3D imaging method based on structured light phase measurement.

[0052] Corresponding to the above embodiments, this embodiment of the invention provides a computer-readable storage medium containing one or more program instructions, which are executed by a processor to provide a 3D imaging method based on structured light phase measurement.

[0053] The embodiments disclosed in this invention provide a computer-readable storage medium storing computer program instructions that, when executed on a computer, cause the computer to perform the aforementioned 3D imaging method based on structured light phase measurement.

[0054] In this embodiment of the invention, the processor can be an integrated circuit chip with signal processing capabilities. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0055] The various methods, steps, and logic diagrams disclosed in the embodiments of this invention can be implemented or executed. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this invention can be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The processor reads information from the storage medium and, in conjunction with its hardware, completes the steps of the above methods.

[0056] The storage medium can be memory, such as volatile memory or non-volatile memory, or may include both volatile and non-volatile memory.

[0057] Among them, non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory.

[0058] Volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous linked dynamic random access memory (Synchlink DRAM, SLDRAM), and direct memory bus RAM (DRRAM).

[0059] The storage media described in the embodiments of the present invention are intended to include, but are not limited to, these and any other suitable types of memory.

[0060] Those skilled in the art will recognize that, in one or more of the examples above, the functions described in this invention can be implemented using a combination of hardware and software. When applied as software, the corresponding functions can be stored in a computer-readable medium or transmitted as one or more instructions or code on a computer-readable medium. Computer-readable media include computer storage media and communication media, wherein communication media include any medium that facilitates the transmission of computer programs from one place to another. Storage media can be any available medium that can be accessed by a general-purpose or special-purpose computer.

[0061] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made on the basis of the technical solution of the present invention should be included within the scope of protection of the present invention.

Claims

1. A 3D imaging method based on structured light phase measurement, characterized in that, include: The structured light measurement system is calibrated to obtain intrinsic and extrinsic parameters, and the effective measurement depth range and the corresponding effective absolute phase interval are determined. The projector projects a preset sinusoidal fringe pattern onto the scene under test, and the camera simultaneously acquires the deformed fringe image modulated by the surface of the object under test. The wrapped phase map is then obtained after calculation. Based on the effective measurement depth range, the initial fringe order of each pixel is calculated using a frequency domain fringe order no-neighbor solution algorithm. Based on the effective range of absolute phase, the initial fringe level is checked and corrected without neighborhood using a level self-checking and error correction algorithm to obtain the final fringe level. Based on the wrapped phase map and the final fringe order, the absolute phase map of the entire field of view is calculated and converted into 3D point cloud data of the scene under test.

2. The 3D imaging method based on structured light phase measurement according to claim 1, characterized in that, The structured light measurement system is calibrated to obtain intrinsic and extrinsic parameters, and the effective measurement depth range and corresponding effective absolute phase interval are determined, including: For a structured light measurement system consisting of a monocular camera and a monocular projector, the internal parameters of the camera and projector are calibrated to the system's external parameters. The effective measurement depth range and the corresponding effective absolute phase interval are calculated based on the system parameters obtained from the calibration.

3. The 3D imaging method based on structured light phase measurement according to claim 1, characterized in that, Based on the effective measurement depth range, the initial fringe order of each pixel is calculated using a frequency domain fringe order no-neighborhood solution algorithm, including: Based on the effective measurement depth range of the system, a candidate set of stripe orders is determined; For each candidate fringe level in the candidate set, a corresponding candidate complex phase map is constructed. Based on the candidate complex phase map, the spectral energy amplitude corresponding to each pixel is calculated, and the initial fringe order is calculated in combination with the frequency domain order consensus.

4. The 3D imaging method based on structured light phase measurement according to claim 3, characterized in that, For each candidate fringe level in the candidate set, a corresponding candidate complex phase map is constructed, including: Calculate the confidence fusion factor for each pixel based on the package phase map; For each level in the candidate level set, construct the corresponding candidate complex phase graph.

5. A 3D imaging method based on structured light phase measurement according to claim 3, characterized in that, Based on the candidate complex phase map, the spectral energy amplitude corresponding to each pixel is calculated, and the initial fringe order is calculated by combining the frequency domain order consensus, including: Calculate the spectral energy amplitude corresponding to each pixel based on the candidate complex phase map, and calculate the fundamental order; The consensus degree of the frequency domain level is calculated based on the basic level, and the initial fringe level is determined based on the consensus degree of the frequency domain level.

6. The 3D imaging method based on structured light phase measurement according to claim 1, characterized in that, Based on the effective range of the absolute phase, the initial fringe order is checked and corrected without neighborhood using a self-checking and error correction algorithm, resulting in the final fringe order, including: Based on the effective range of the absolute phase, an effective probability model is constructed to calculate the verification probability of the initial stripe level; Based on the verification probability, calculate the multi-factor joint verification index, determine the validity of the initial stripe level, and obtain the set of pixels to be corrected. For pixels in the set of pixels to be corrected, a multi-dimensional joint scoring model is constructed to perform high-precision level correction.

7. A 3D imaging method based on structured light phase measurement according to claim 6, characterized in that, For pixels in the set of pixels to be corrected, a multi-dimensional joint scoring model is constructed to perform high-precision level correction, including: The multi-factor joint score for each candidate level is calculated using a multi-factor joint scoring function; Based on the multi-factor joint scoring and combined with the consensus of the frequency domain level, the pixels to be corrected are divided into ordinary pixels to be corrected and high uncertainty pixels, and differential error correction is performed to obtain the final stripe level.

8. A 3D imaging system based on structured light phase measurement, characterized in that, include: The system calibration module is used to calibrate the structured light measurement system, obtain intrinsic and extrinsic parameters, and determine the effective measurement depth range and the corresponding effective absolute phase interval. The wrapping phase calculation module is used to control the projector to project a preset sinusoidal stripe pattern onto the scene under test. The camera synchronously acquires the deformed stripe image after modulation on the surface of the object under test, and the wrapping phase map is obtained after calculation. The frequency domain order calculation module is used to calculate the initial fringe order of each pixel based on the effective measurement depth range using a frequency domain order fringe order no-neighborhood calculation algorithm. The order verification and error correction module is used to perform neighborhood-free verification and error correction on the initial fringe order based on the effective range of the absolute phase using an order self-verification and error correction algorithm, so as to obtain the final fringe order. The 3D reconstruction module is used to calculate the absolute phase map of the entire field of view based on the wrapping phase map and the final fringe order, and convert it into 3D point cloud data of the scene under test.

9. A computer-readable storage medium, characterized in that, It includes one or more program instructions, which are executed by a processor as described in any one of claims 1-7, to provide a 3D imaging method based on structured light phase measurement.

Images