A method for constructing a true value of a micro-morphology of an inner wall of a pipeline based on a spatial domain trigger and a dynamic light gate control

Abstract

This invention discloses a method and system for constructing the true value of the microscopic morphology of the inner wall of a pipeline based on spatial domain triggering and dynamic shutter control, belonging to the field of pipeline inspection technology. It employs a dynamic time-division stroboscopic acquisition strategy based on spatial sampling consistency to ensure that the four image sequences required for photometric stereo vision overlap in physical space. It introduces dynamic shutter control based on instantaneous velocity to reduce the non-uniformity of imaging quality caused by the variable speed movement of multiple sensor modules. A robust photometric stereo algorithm is used to decouple and separate the pure pipe wall texture and micro-geometry, and reconstruct a high-precision 3D depth map. Finally, standardized true value data integrating position, texture, and geometry is constructed. This invention solves the problem in existing photometric stereo vision methods where multiple frames acquired under different lighting conditions are difficult to keep spatially consistent when the carrier is in continuous motion or experiencing mechanical micro-vibration, leading to significant errors in normal vector calculation and 3D reconstruction results.

Description

Technical Field

[0001] This invention belongs to the field of pipeline inspection technology, specifically relating to a method for constructing the true value of the microstructure of the inner wall of a pipeline based on spatial domain triggering and dynamic shutter control. Background Technology

[0002] Long-distance oil and gas pipelines serve as strategic "main arteries" for national energy transportation, and their operational safety is directly related to national energy security and social stability. However, these pipelines operate under harsh conditions, including underground laying, high-pressure transmission, and complex geological environments, making them highly susceptible to structural damage such as electrochemical corrosion, stress corrosion cracking, and mechanical damage. Since external pipeline inspections cover many areas inaccessible to humans and machines, using pipeline inspection robots equipped with high-precision sensors to enter the pipeline and conduct non-destructive inspections covering the entire mileage has become a primary technical means for assessing pipeline health and preventing catastrophic leaks.

[0003] Among existing internal inspection technologies, electromagnetic-acoustic (EMS) non-destructive testing methods such as magnetic flux leakage and ultrasonic testing offer high-precision detection capabilities for defects such as corrosion and thinning in ferromagnetic pipes. However, internal inspection equipment requires a full-coverage arrangement of related sensors, resulting in high costs and impacting robot throughput. In contrast, machine vision inspection technology, with its non-contact, lightweight, and low-cost characteristics, demonstrates unique advantages in the macroscopic detection of defects on the inner surface of pipes, making it an important means of early inspection and macroscopic damage screening for pipelines.

[0004] However, existing pipeline visual inspection robots mostly rely on two-dimensional industrial cameras as their primary sensing method. The two-dimensional images they acquire can only record surface texture information and cannot directly reflect the microscopic undulations and depth changes of the inner wall surface. In practical engineering, relying solely on two-dimensional images often makes it difficult to effectively distinguish between "surface stains caused by color changes" and "corrosion pits or microcracks with actual structural significance," thereby increasing the risk of misjudgment and limiting the reliability of visual inspection results in engineering decision-making.

[0005] To compensate for the limitations of 2D vision in acquiring depth information, existing technologies have attempted to introduce 3D measurement methods such as line laser scanning or structured light. However, these technologies (line laser / structured light) are limited by sensor size and scanning modes, making it difficult to mount them on the body of a pipeline robot within confined pipe spaces; the pipeline robot lacks a fixed and stable track during its movement inside the pipe, failing to provide a stable scanning environment; and the circumferential scanning mode requires the pipeline robot to maintain a low speed, resulting in large data storage requirements, making it difficult to adapt to long-distance, long-term continuous scanning.

[0006] Against this backdrop, photometric stereo vision technology, due to its ability to derive surface normal vectors from shadow variations under multi-directional lighting and its high sensitivity to minute surface undulations, is considered a promising method for 3D shape perception. Compared to line laser or structured light methods, photometric stereo vision requires relatively simple hardware structures, theoretically making it more suitable for integration into compact pipeline inspection robots. However, existing photometric stereo vision methods are generally based on the assumptions that "the camera and the object being measured are relatively stationary" or "multiple frames of images can be considered as samples from the same spatial location." When the carrier is in continuous motion or experiences mechanical micro-vibrations, it is difficult for multiple frames acquired under different lighting conditions to maintain spatial consistency, leading to significant errors in normal vector calculation and 3D reconstruction results, making this technology difficult to directly apply in actual pipeline inspection conditions. Summary of the Invention

[0007] The purpose of this invention is to solve the above problems and provide a method for constructing the true value of the microstructure of the inner wall of a pipe based on spatial domain triggering and dynamic shutter control.

[0008] A method for constructing the true microstructure of the inner wall of a pipe based on spatial domain triggering and dynamic shutter control includes the following steps: (1) Dynamic time-division stroboscopic acquisition based on spatial sampling consistency A multi-sensor module is positioned on the central axis inside the pipeline. The multi-sensor module includes an active binocular camera at the front end for forward sensing and macroscopic modeling, and eight array cameras distributed circumferentially on the sides for photometric stereo imaging. Each array camera is equipped with four independently controlled zone light sources L1, L2, L3, and L4. The four zone light sources are centrally symmetrically distributed on a projection plane perpendicular to the optical axis of the array camera. The multi-sensor module is fed along the pipeline axis by feeding hardware powered by a servo motor. The acquisition process is initiated by a spatial state perception gating unit of the underlying FPGA, which confirms that the multi-sensor module has reached its position and its attitude is stable, i.e., the spatial sampling consistency constraint is met. By introducing dynamic shutter control, the FPGA calculates the time difference between adjacent pulses of the servo motor encoder in real time to obtain the current instantaneous feed speed V of the multi-sensor module. To achieve consistent imaging quality across all operating conditions, from low-speed crawling to high-speed cruising, a pulse width is established with the allowable image plane motion blur threshold as the core constraint. The inverse mapping function between the instantaneous feed rate V and the light source driver hardware, and the minimum and maximum pulse width physical boundaries, are expressed as follows: in: The real-time strobe exposure pulse width output by the FPGA to the LED light source, in seconds; V represents the current instantaneous feed speed of the multi-sensor module fed back by the servo motor encoder, in mm / s; The preset threshold for allowable motion blur is set to [value]. To ensure subpixel-level clarity; The physical size of the image camera's pixels, in mm; The optical magnification of the imaging system is dimensionless. The minimum narrowest pulse width allowed by the light source driver hardware is set to 10μs. Under this state, the system automatically triggers the overdrive mode and synchronously increases the instantaneous drive current of the LED to compensate for the luminous flux. To prevent motion blur overflow, the maximum allowable pulse width is set to 2000μs. Under this condition, the system maintains the rated current to achieve a high signal-to-noise ratio. After receiving the trigger signal Within the window, the system calculates according to the above... The pulse width drives the four zone light sources to blink sequentially: At time T0: Light up the first zone light source L1, then the camera exposes and acquires image I1; Time T1: Light up the second zone light source L2, then the camera exposes and acquires image I2; Time T2: Light up the third zone light source L3, then the camera exposes and acquires image I3; At time T3: Light up the fourth zone light source L4, then the camera exposes and acquires image I4; As a result, the alignment accuracy of images I1, I2, I3, and I4 in physical space reaches the sub-pixel level; (2) Surface gradient field calculation based on robust photometric stereo Mathematical model construction: Based on the Lambertian diffuse reflection model, the reflection model was rewritten in a vectorized form, wherein, Indicates the first The unit incident direction vector of each light source. The unit surface normal vector representing the spatial position of the inner wall of the pipe. For the reflectivity coefficient at this location, a composite normal vector parameter is introduced. The brightness expression is: That is, the composite normal vector parameter; Iterative weighted solution: Constructing a weighted objective function based on sparse induced norm: in The confidence weight for the k-th image is calculated iteratively as follows: Initial estimation: Let the initial weight matrix be... Given the identity matrix, calculate the initial pseudo-normal vector using the least squares method. : Residual calculation and weight update: In the t-th iteration, calculate the residual between the observed brightness and the theoretical brightness. The weights are dynamically updated based on the residuals. ; Weighted solution: using the updated weight matrix Resolve the algorithm vector: Repeat the above steps until convergence, finally obtaining the pure surface normal vector. and albedo ; (3) Variational fusion reconstruction based on macroscopic constraints Gradient field calculation: Based on the unit normal vector n obtained in step (2), calculate the gradient field (p, q) on the inner wall surface of the pipe: Construction of variational energy equation: Introducing macroscopic sparse depth obtained from active binocular cameras As a low-frequency constraint, a global variational energy functional E(Z) containing gradient fidelity terms and depth approximation terms is constructed: Where Z is the high-precision depth map to be reconstructed. The gradient vector field derived from the photometric stereo. The regularization coefficient for balancing the weights; Frequency domain solution: According to the variational principle, minimizing the energy functional is equivalent to solving the following Euler-Lagrange equations: Expanded into partial differential form: By directly solving in the frequency domain using the Fast Fourier Transform, the final depth map Z(x,y) that integrates micro-texture details and macro-geometric structure is obtained. (4) Data Mapping The physical mileage index at the trigger time is read, and the generated 3D topography data is mapped and bound to the physical mileage of the encoder, ultimately producing a 3D image of the pipe inner wall with mileage coordinates.

[0009] Furthermore, in step (1), the four zone light sources L1, L2, L3, and L4 are located in the upper left, upper right, lower left, and lower right positions of the four quadrants of the camera's field of view, respectively, forming an X-shaped layout.

[0010] Furthermore, in step (1), the four zone light sources L1, L2, L3, and L4 are located at the top, bottom, left, and right orthogonal positions of the camera's field of view, respectively, forming a cross-shaped layout.

[0011] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: 1. Elimination of motion artifacts and vibration interference: This invention deeply integrates the hardware platform's "micro-latency elastic capture" and "active deceleration" mechanisms. On the one hand, "micro-latency" actively avoids the peak moments of human disturbance or obstacle crossing vibration; on the other hand, "active deceleration" significantly reduces the amount of physical displacement during the four exposures. This ensures that the four images involved in the calculation are strictly overlapped in physical space, enabling the algorithm to output high-fidelity microscopic three-dimensional morphology even under dynamic and even strongly disturbed conditions.

[0012] 2. Pure physical calculation: The reconstruction process is entirely based on the principles of optical geometry (integration), without relying on the conjectures of neural networks, and the generated three-dimensional data has objective physical meaning.

[0013] 3. Consistent Imaging Quality: This invention uses "dynamic shutter control" to forcibly lock the motion blur of all acquired images to near the pixel level (e.g., <0.5 pixels). Regardless of the motion state of the multi-sensor module (cruising or decelerating to overcome obstacles), the reconstructed 3D ground truth has uniform frequency domain response characteristics.

[0014] 4. High-precision restoration of microscopic morphology: This invention can restore depth information that ordinary two-dimensional images cannot present, and intuitively present the tiny undulations of the inner wall of the pipe. Attached Figure Description

[0015] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort, wherein: Figure 1 This is a schematic diagram of the logic flow of the present invention.

[0016] Figure 2 This is a schematic diagram of the multi-sensor module of the present invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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 embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0018] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0019] It should be noted that the labels and letters in the following figures represent similar items, therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0020] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of this invention is in use. They are only used for the purpose of simplifying the description of this invention and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention. In addition, the terms "first," "second," and "third," etc., are only used to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0021] Furthermore, terms such as "horizontal" and "vertical" do not imply that components must be absolutely horizontal or suspended, but rather that they can be slightly tilted. For example, "horizontal" simply means that its direction is more horizontal than "vertical," and does not mean that the structure must be completely horizontal, but can be slightly tilted.

[0022] In the description of this invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal communication between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0023] A method for constructing the true microstructure of the inner wall of a pipe based on spatial domain triggering and dynamic shutter control includes the following steps: (1) Dynamic time-division stroboscopic acquisition based on spatial sampling consistency Mechanical structure design: A multi-sensor module is positioned on the central axis inside the pipeline. The multi-sensor module includes an active binocular camera at the front end for forward sensing and macroscopic modeling, and eight array cameras distributed circumferentially on the sides for photometric stereo imaging. Each array camera is equipped with four independently controlled zone light sources L1, L2, L3, and L4. The four zone light sources are centrally symmetrically distributed on a projection plane perpendicular to the optical axis of the array camera. The multi-sensor module is fed along the pipeline axis by feeding hardware powered by a servo motor. The acquisition process is initiated by a spatial state perception gating unit of the underlying FPGA, which confirms that the multi-sensor module has reached its position and its attitude is stable, i.e., the spatial sampling consistency constraint is met. A multi-sensor module is positioned on the central axis inside the pipeline. The multi-sensor module includes an active binocular camera at the front end for forward sensing and macroscopic modeling, and eight array cameras distributed circumferentially on the sides for photometric stereo imaging. Each array camera is equipped with four independently controlled zone light sources L1, L2, L3, and L4. The four zone light sources are centrally symmetrically distributed on a projection plane perpendicular to the optical axis of the array camera. The multi-sensor module is fed along the pipeline axis by feeding hardware powered by a servo motor. The acquisition process is initiated by a spatial state perception gating unit of the underlying FPGA, which confirms that the multi-sensor module has reached its position and its attitude is stable, i.e., the spatial sampling consistency constraint is met. By introducing dynamic shutter control, the FPGA calculates the time difference between adjacent pulses of the servo motor encoder in real time to obtain the current instantaneous feed speed V of the multi-sensor module. To achieve consistent imaging quality across all operating conditions, from low-speed crawling to high-speed cruising, a pulse width is established with the allowable image plane motion blur threshold as the core constraint. The inverse mapping function between the instantaneous feed rate V and the light source driver hardware, and the minimum and maximum pulse width physical boundaries, are expressed as follows: in: The real-time strobe exposure pulse width output by the FPGA to the LED light source, in seconds; V represents the current instantaneous feed speed of the multi-sensor module fed back by the servo motor encoder, in mm / s; The preset threshold for allowable motion blur is set to [value]. To ensure subpixel-level clarity; The physical size of the image camera's pixels, in mm; The optical magnification of the imaging system is dimensionless. The minimum narrowest pulse width allowed by the light source driver hardware is set to 10μs. Under this state, the system automatically triggers the overdrive mode and synchronously increases the instantaneous drive current of the LED to compensate for the luminous flux. To prevent motion blur overflow, the maximum allowable pulse width is set to 2000μs. Under this condition, the system maintains the rated current to achieve a high signal-to-noise ratio. After receiving the trigger signal Within the window, the system calculates according to the above... The pulse width drives the four zone light sources to blink sequentially: Time T0: Light up the first zone light source L1 (such as the upper left zone) → camera exposure → image I1 acquired; Time T1: Illuminate the second zone light source L2 (such as the lower left zone) → Camera exposure → Image acquisition I2; Time T2: Illuminate the third zone light source L3 (such as the lower right zone) → Camera exposure → Image acquisition I3; Time T3: Illuminate the fourth zone light source L4 (such as the upper right zone) → Camera exposure → Image acquisition I4; Results: Thanks to the dual protection of "state gating (stable)" and "active deceleration (slow)," the alignment accuracy of these four images in physical space is significantly improved (close to pixel level), effectively suppressing registration errors caused by motion blur or vibration, and meeting the solution requirements of photometric stereo algorithm.

[0024] (2) Step 2: Surface gradient field calculation based on robust photometric solid Mathematical model construction: Based on the Lambertian diffuse reflection model, the reflection model was rewritten in a vectorized form, wherein, Indicates the first The unit incident direction vector of each light source. The unit surface normal vector representing the spatial position of the inner wall of the pipe. For the reflectivity coefficient at this location, a composite normal vector parameter is introduced. The brightness expression is: That is, the composite normal vector parameter; Iterative Weighted Solution (IRLS) Process: To eliminate the damage to the Lambertian model caused by the highlights and shadows of oil stains on the inner wall of the pipe, this invention constructs a weighted objective function based on the sparse induced norm: in Let be the confidence weight for the k-th image. The calculation process is performed iteratively according to the following steps: Initial estimation: Let the initial weight matrix be... Given the identity matrix, calculate the initial pseudo-normal vector using the least squares method. : Residual calculation and weight update: In the t-th iteration, calculate the residual between the observed brightness and the theoretical brightness. The weights are dynamically updated based on the residuals. (For example, using Tukey's biweight function).

[0025] Weighted solution: using the updated weight matrix Resolve the algorithm vector: Repeat the above steps until convergence, finally obtaining the pure surface normal vector. and albedo .

[0026] (3) Step 3: Variational fusion reconstruction based on macroscopic constraints Gradient field calculation: Based on the unit normal vector n obtained in step two, calculate the gradient field (p, q) on the inner wall surface of the pipe: Construction of variational energy equation: To address the low-frequency geometric drift problem caused by direct integration, macroscopic sparse depth acquired through binocular vision from a related patent is introduced. As a low-frequency constraint, a global variational energy functional E(Z) is constructed, which includes a "gradient fidelity term" and a "depth approximation term": Where Z is the high-precision depth map to be reconstructed. The gradient vector field derived from the photometric stereo. This is the regularization coefficient for the balancing weights.

[0027] Frequency domain solution (screening the Poisson equation): According to the variational principle, minimizing the above energy functional is equivalent to solving the following Euler-Lagrange equation (i.e., the Poisson equation with a screening term): Expanded into partial differential form: The above equations are solved directly in the frequency domain using Fast Fourier Transform (FFT) to obtain the final depth map Z(x,y) that integrates micro-texture details and macro-geometric structure.

[0028] (4) Step Four: Data Mapping As shown in the sub-process in Figure 1, the physical mileage index at the trigger time is read. The generated 3D topography data is then mapped and bound to the physical mileage of the encoder.

[0029] Final output: A 3D image of the pipe's inner wall with mileage coordinates.

[0030] Specifically, the hardware platform's micro-delay elastic capture and active deceleration mechanism is as follows: The hardware platform includes an FPGA, feeding hardware for feeding the multi-sensor module, a servo motor as the power source for the feeding hardware, and an IMU located within the multi-sensor module. The execution process of the FPGA's spatial state awareness gating unit is as follows: S1, Interference Monitoring: The FPGA reads the triaxial acceleration and triaxial angle feedback from the IMU in real time, and determines whether the current acceleration or angle deviation exceeds the safety threshold based on the preset safety threshold. If so, it determines that the multi-sensor module will not sample. S2, Step-size modulation based on parallax perception: An inverse coupling relationship between sampling density and feed rate is established using an active binocular camera, based on the vertical field of view height of a single array camera distributed circumferentially on the side. and preset image overlap rate Calculate the maximum allowable basic step size. ,in, This ensures seamless coverage of the circumferential panoramic image under standard acquisition mode. The FPGA calculates the disparity change rate of the active binocular camera's ROI region in real time. When the disparity d of the active binocular camera stabilizes, the system executes the basic step size. When a parallax step is detected, the FPGA immediately switches the trigger step size to a fine step size and controls the servo motor to decelerate. S3, Physical Gated Arbitration: When the motor encoder pulse count reaches the sampling step size, a sampling request is generated. The FPGA checks the current IMU status: if the IMU value is lower than the safety threshold, it immediately triggers the image capture; if the IMU value is higher than the safety threshold, the FPGA refuses to trigger immediately and enters the S4 micro-delay elastic capture mode; the trigger signal is resent the instant the vibration wave passes and the IMU value falls back below the threshold. S4, micro-delay elastic capture: The system does not skip the current sampling point and actively introduces a dynamic Δt micro-delay. The FPGA tracks the IMU value in real time. When the IMU value falls below the threshold, a physical trigger signal is sent again. S5, a source-level hard implantation of spacetime: In the same clock cycle as the release trigger signal, the FPGA extracts the current absolute physical mileage and instantaneous attitude quaternion in parallel, and combines the absolute physical mileage and instantaneous attitude quaternion data into a 64-bit spatiotemporal status stamp, which is then hard-embedded into the frame header of the original data of the multi-sensor sampling unit.

[0031] The above description constitutes an embodiment of the present invention. The foregoing descriptions are preferred embodiments of the present invention. Unless there is a clear contradiction or a prerequisite for a particular preferred embodiment, the preferred embodiments can be arbitrarily combined and used. The embodiments and specific parameters described are merely for clearly illustrating the verification process of the invention and are not intended to limit the scope of patent protection of the present invention. The scope of patent protection of the present invention is still determined by its claims. Similarly, any equivalent structural changes made based on the description and drawings of the present invention should also be included within the scope of protection of the present invention.

Claims

1. A method for constructing the true microstructure of the inner wall of a pipe based on spatial domain triggering and dynamic shutter control, characterized in that, Includes the following steps: (1) Dynamic time-division stroboscopic acquisition based on spatial sampling consistency A multi-sensor module is positioned on the central axis inside the pipeline. The multi-sensor module includes an active binocular camera at the front end for forward sensing and macroscopic modeling, and eight array cameras distributed circumferentially on the sides for photometric stereo imaging. Each array camera is equipped with four independently controlled zone light sources L1, L2, L3, and L4. The four zone light sources are centrally symmetrically distributed on a projection plane perpendicular to the optical axis of the array camera. The multi-sensor module is fed along the pipeline axis by feeding hardware powered by a servo motor. The acquisition process is initiated by a spatial state perception gating unit of the underlying FPGA, which confirms that the multi-sensor module has reached its position and its attitude is stable, i.e., the spatial sampling consistency constraint is met. By introducing dynamic shutter control, the FPGA calculates the time difference between adjacent pulses of the servo motor encoder in real time to obtain the current instantaneous feed speed V of the multi-sensor module. To achieve consistent imaging quality across all operating conditions, from low-speed crawling to high-speed cruising, a pulse width is established with the allowable image plane motion blur threshold as the core constraint. The inverse mapping function between the instantaneous feed rate V and the light source driver hardware, and the minimum and maximum pulse width physical boundaries, are expressed as follows: ; in: The real-time strobe exposure pulse width output by the FPGA to the LED light source, in seconds; V represents the current instantaneous feed speed of the multi-sensor module fed back by the servo motor encoder, in mm / s; The preset threshold for allowable motion blur is set to [value]. To ensure subpixel-level clarity; The physical size of the image camera's pixels, in mm; The optical magnification of the imaging system is dimensionless. The minimum narrowest pulse width allowed by the light source driver hardware is set to 10μs. Under this state, the system automatically triggers the overdrive mode and synchronously increases the instantaneous drive current of the LED to compensate for the luminous flux. To prevent motion blur overflow, the maximum allowable pulse width is set to 2000μs. Under this condition, the system maintains the rated current to achieve a high signal-to-noise ratio. After receiving the trigger signal Within the window, the system calculates according to the above... The pulse width drives the four zone light sources to blink sequentially: At time T0: Light up the first zone light source L1, then the camera exposes and acquires image I1; Time T1: Light up the second zone light source L2, then the camera exposes and acquires image I2; Time T2: Light up the third zone light source L3, then the camera exposes and acquires image I3; At time T3: Light up the fourth zone light source L4, then the camera exposes and acquires image I4; As a result, the alignment accuracy of images I1, I2, I3, and I4 in physical space reaches the sub-pixel level; (2) Surface gradient field calculation based on robust photometric stereo Mathematical model construction: Based on the Lambertian diffuse reflection model, the reflection model was rewritten in a vectorized form, wherein, Indicates the first The unit incident direction vector of each light source. The unit surface normal vector representing the spatial position of the inner wall of the pipe. For the reflectivity coefficient at this location, a composite normal vector parameter is introduced. The brightness expression is: ; That is, the composite normal vector parameter; Iterative weighted solution: Constructing a weighted objective function based on sparse induced norm: ; in The confidence weight for the k-th image is calculated iteratively as follows: Initial estimation: Let the initial weight matrix be... Given the identity matrix, calculate the initial pseudo-normal vector using the least squares method. : ; Residual calculation and weight update: In the t-th iteration, calculate the residual between the observed brightness and the theoretical brightness. The weights are dynamically updated based on the residuals. ; Weighted solution: using the updated weight matrix Resolve the algorithm vector: ; Repeat the above steps until convergence, finally obtaining the pure surface normal vector. and albedo ; (3) Variational fusion reconstruction based on macroscopic constraints Gradient field calculation: Based on the unit normal vector n obtained in step (2), calculate the gradient field (p,q) on the inner wall surface of the pipe: ; Construction of variational energy equation: Introducing macroscopic sparse depth obtained from active binocular cameras As a low-frequency constraint, a global variational energy functional E(Z) containing gradient fidelity terms and depth approximation terms is constructed: ; Where Z is the high-precision depth map to be reconstructed. The gradient vector field derived from the photometric stereo. The regularization coefficient for balancing the weights; Frequency domain solution: According to the variational principle, minimizing the energy functional is equivalent to solving the following Euler-Lagrange equations: ; Expanded into partial differential form: ; By directly solving in the frequency domain using the Fast Fourier Transform, the final depth map Z(x,y) that integrates micro-texture details and macro-geometric structure is obtained. (4) Data Mapping The physical mileage index at the trigger time is read, and the generated 3D topography data is mapped and bound to the physical mileage of the encoder, ultimately producing a 3D image of the pipe inner wall with mileage coordinates.

2. The method for constructing the true value of the microstructure of the inner wall of a pipe based on spatial domain triggering and dynamic shutter control according to claim 1, characterized in that, In step (1), the four zone light sources L1, L2, L3, and L4 are located in the upper left, upper right, lower left, and lower right positions of the four quadrants of the camera's field of view, respectively, forming an X-shaped layout.

3. The method for constructing the true value of the microstructure of the inner wall of a pipe based on spatial domain triggering and dynamic shutter control as described in claim 1, characterized in that, In step (1), the four zone light sources L1, L2, L3, and L4 are located at the top, bottom, left, and right orthogonal positions of the camera's field of view, forming a cross-shaped layout.

Images