A three-dimensional defect detection method and system based on complex color stripe projection

Abstract

The application relates to the technical field of photoelectric detection, and particularly discloses a three-dimensional defect detection method and system based on complex color stripe projection, which comprises an upper computer control and processing module, a structured light projection module and an imaging module. The upper computer control and processing module is used for generating a complex color stripe pattern coded with phase difference information and coordinating and controlling the synchronous operation of the whole system. The structured light projection module is used for receiving the complex color stripe pattern and projecting the complex color stripe pattern onto the surface of a measured object at one time. The imaging module is used for receiving the reflected complex color light modulated by the surface of the measured object, separating the complex color light into monochromatic light signals of different colors and synchronously collecting the monochromatic light signals. The imaging module further comprises an active high light suppression unit which is used for dynamically adjusting the polarization direction according to the instruction of the upper computer control and processing module so as to suppress the high light caused by the mirror reflection. By projecting a plurality of groups of sinusoidal stripes with fixed phase difference onto the surface of the measured object at one time, a plurality of deformed stripe patterns with phase shift can be synchronously collected under single exposure.

Description

Technical Field

[0001] This invention relates to the field of photoelectric detection technology, specifically a three-dimensional defect detection method and system based on polychromatic fringe projection. Background Technology

[0002] Traditional structured light 3D measurement technology faces significant challenges in the online inspection of fast-moving targets with highly reflective properties, such as automotive sheet metal parts. On the one hand, traditional phase-shifting methods require multiple time-division projection acquisitions, making them unsuitable for high-speed moving workpieces on production lines, and resulting in low defect detection efficiency. On the other hand, the clear coat on the sheet metal surface causes strong specular reflection, leading to overexposure of highlights at specific viewing angles, resulting in the loss of local stripe information and making it impossible to detect minute defects.

[0003] Existing highlight suppression methods mostly employ orthogonal polarizer arrays, but their fixed polarization direction cannot adapt to the dynamic changes in the polarization state of highlights caused by the continuous change of the surface normal of a moving curved workpiece, resulting in limited suppression effects. Although high dynamic range (HDR) imaging technology can be used, it requires multiple exposures and still cannot meet the real-time requirements of dynamic detection.

[0004] Therefore, there is an urgent need for a three-dimensional detection system that can simultaneously acquire multiphase shifting fringes in a single exposure and dynamically suppress highlights that change with angle and color. Summary of the Invention

[0005] The purpose of this invention is to provide a three-dimensional defect detection method and system based on polychromatic stripe projection to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a three-dimensional defect detection system based on polychromatic fringe projection, comprising:

[0007] The system comprises a host computer control and processing module, a structured light projection module, and an imaging module.

[0008] The host computer control and processing module is used to generate composite color stripe patterns encoded with phase difference information and to coordinate and control the synchronous operation of the entire system.

[0009] The structured light projection module is used to receive the composite color stripe pattern and project it onto the surface of the object being measured in one go;

[0010] The imaging module is used to receive composite color light reflected after being modulated by the surface of the object under test, and to separate it into monochromatic light signals of different colors for synchronous acquisition.

[0011] The imaging module also includes an active specular highlight suppression unit, which dynamically adjusts the polarization direction according to the instructions of the host computer control and processing module to suppress specular highlights caused by specular reflection.

[0012] As a further embodiment of the present invention, the composite color stripe pattern is obtained by encoding three sinusoidal stripe patterns with a preset phase difference into the red, green, and blue color channels of a color image frame, respectively.

[0013] As a further embodiment of the present invention, the phase differences of the three sinusoidal fringe patterns are 0°, 120°, and 240°, respectively.

[0014] As a further embodiment of the present invention, the structured light projection module is a 3LCD projector, which includes liquid crystal panels corresponding to the red, green and blue channels respectively, for independently modulating the light intensity of each channel and combining them into a colored striped pattern for projection.

[0015] As a further embodiment of the present invention, the imaging module includes:

[0016] Dichroic mirrors are used to separate the reflected composite light into three monochromatic lights: red, green, and blue.

[0017] Three photodetectors are used to receive monochromatic light of the corresponding color and to acquire images.

[0018] The three photodetectors achieve synchronous exposure through hardware triggering.

[0019] As a further embodiment of the present invention, the dichroic mirror separation group includes two-stage dichroic mirrors:

[0020] The first-order dichroic mirror is used to reflect red light and transmit green and blue light.

[0021] The second-order dichroic mirror is used to reflect green light and transmit blue light.

[0022] As a further embodiment of the present invention, the active specular suppression unit includes:

[0023] Electric rotary table;

[0024] Linear polarization analyzer installed on the electric rotary table;

[0025] The host computer control and processing module controls the rotation of the electric rotary table to adjust the polarization axis angle of the analyzer.

[0026] As a further embodiment of the present invention, the host computer control and processing module executes a predictive-tracking gradient descent algorithm to dynamically search for and track the optimal polarization angle for suppressing specular highlights.

[0027] As a further embodiment of the present invention, the prediction-tracking gradient descent algorithm includes:

[0028] Coarse localization stage: Quickly locate the angle range with the lowest highlight brightness through sparse sampling;

[0029] Fine positioning stage: Gradient descent is performed within the specified angle range to find the optimal suppression angle in a step-by-step manner;

[0030] Tracking phase: Utilizing the continuity of polarization angles between adjacent time moments, the optimal angle of the previous time moment is used as the starting point for the current search.

[0031] The present invention also provides a three-dimensional defect detection method based on the system, comprising the following steps:

[0032] Three sinusoidal fringe patterns with a preset phase difference are fused into a composite color fringe pattern and projected onto the surface of the object being measured.

[0033] The image of the monochrome channel deformed stripes after beam splitting and highlight suppression is received. In the highlight suppression step, the optimal polarization angle for suppressing highlights is determined based on a prediction-tracking gradient descent algorithm.

[0034] The absolute phase is calculated using phase solution and phase expansion algorithms;

[0035] The three-dimensional shape is reconstructed based on the absolute phase, and defects are detected.

[0036] Compared with existing technologies, the advantages of this invention are: by projecting multiple sets of sinusoidal fringes with a fixed phase difference onto the surface of the object under test in a single exposure, the composite color light reflected after modulation by the object surface is separated in real time and the corresponding color light is received, enabling the simultaneous acquisition of multiple deformed fringe patterns with phase shifts in a single exposure. This design transforms the traditional three-step phase shift operation in time series into a one-time concurrent processing in color space, greatly improving the data acquisition speed and enabling its effective application in the detection of fast-moving targets.

[0037] An active and dynamic specular suppression scheme is proposed, which upgrades the "passive fixed extinction" to "active dynamic suppression". It can match the constantly changing specular polarization direction on the moving workpiece in real time, ensuring that the best specular suppression effect is maintained at all times during dynamic detection, thereby obtaining complete and accurate surface three-dimensional data. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention.

[0039] Figure 1 This is a schematic diagram of the structure of a three-dimensional defect detection system based on polychromatic stripe projection, provided in an embodiment of the present invention.

[0040] Figure 2 This is a flowchart of a three-dimensional defect detection system based on polychromatic stripe projection, provided for an embodiment of the present invention.

[0041] Figure 3 A flowchart illustrating stripe generation and projection provided for an embodiment of the present invention.

[0042] Figure 4 The flowchart of active highlight removal control provided in the embodiment of the present invention.

[0043] Figure 5 The red 0° phase sinusoidal stripe light provided for the embodiments of the present invention.

[0044] Figure 6 The green 120° phase sinusoidal stripe light provided in the embodiments of the present invention.

[0045] Figure 7 The blue 240° phase sinusoidal stripe light provided in the embodiments of the present invention.

[0046] Figure 8 The composite stripe light projected onto the sheet metal part is provided in the embodiments of the present invention.

[0047] Figure 9 This is a schematic diagram showing a highlight effect provided for an embodiment of the present invention.

[0048] Figure 10 This is a schematic diagram illustrating the effect of removing highlights according to an embodiment of the present invention.

[0049] The components include: 1. Host computer control and processing module; 2. Structured light projection module; 3. Object under test; 4. Convex lens; 5. First electrically controlled rotating polarizer; 6. Red filter; 7. First photodetector; 8. Dichroic mirror dichroic filter group; 9. Second electrically controlled rotating polarizer; 10. Green filter; 11. Second photodetector; 12. Third electrically controlled rotating polarizer; 13. Blue filter; 14. Third photodetector. Detailed Implementation

[0050] To make the technical problems to be solved, the technical solutions, and the beneficial effects of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present invention and are not intended to limit the present invention.

[0051] like Figures 1 to 10 In this embodiment of the invention, a three-dimensional defect detection system based on polychromatic fringe projection includes:

[0052] Upper computer control and processing module 1; structured light projection module 2; imaging module;

[0053] The host computer control and processing module 1 is used to generate a composite color stripe pattern encoded with phase difference information and to coordinate and control the synchronous operation of the entire system.

[0054] The structured light projection module 2 is used to receive the composite color stripe pattern and project it onto the surface of the object 3 under test in one go.

[0055] The imaging module is used to receive composite color light reflected after being modulated by the surface of the object under test 3, and to separate it into monochromatic light signals of different colors for synchronous acquisition.

[0056] The imaging module also includes an active specular highlight suppression unit, which dynamically adjusts the polarization direction according to the instructions of the host computer control and processing module 1 to suppress specular highlights caused by specular reflection.

[0057] In this embodiment, the structured light projection module 2 includes a projector, and the object under test 3 is an automotive sheet metal part.

[0058] The entire system's workflow begins with the host computer. The host computer control and processing module 1, which includes the host computer, first generates three sinusoidal fringe patterns with preset phase differences and encodes them into the red, green, and blue color channels of a single-frame color image, combining them into a single-frame composite fringe pattern. It is also responsible for coordinating and controlling the synchronous operation of all hardware modules. Subsequently, a 3LCD projector receives this composite pattern and projects it onto the surface of the object being measured 3. The surface morphology modulates the fringe, causing it to deform. The reflected polychromatic deformed fringe light is received in real-time by a dichroic mirror beam splitter group and separated into three independent monochromatic light paths (red, green, and blue), which are then guided to three photodetectors for acquisition, thus simultaneously acquiring three monochromatic channel deformed fringe images. During this imaging process, a high-performance analyzer driven by a high-precision electric rotary stage acts as an active specular highlight suppression unit. It adjusts its polarization axis angle in real-time according to instructions from the host computer, independently optimizing light of different wavelengths. A specially designed predictive-tracking gradient descent optimal angle search algorithm actively suppresses the changing specular highlights caused by object movement, ensuring image acquisition quality. Finally, all image data is transmitted back to the host computer, which executes phase calculation and 3D reconstruction algorithms to reconstruct the 3D shape of the object's surface, thereby enabling the detection of scratches on the surface of automotive sheet metal parts.

[0059] like Figures 5 to 10 In a preferred embodiment of the present invention, the composite color stripe pattern is obtained by encoding three sinusoidal stripe patterns with a preset phase difference into the red, green and blue color channels of a frame of color image respectively.

[0060] The phase differences of the three sinusoidal fringe patterns are 0°, 120°, and 240°, respectively.

[0061] In this embodiment, the host computer first generates three grayscale sinusoidal stripe patterns with fixed phase differences of 0°, 120°, and 240°, and then places them into the red, green, and blue channels of a color image frame, respectively, to synthesize a single-frame composite color stripe pattern. This composite pattern is then transmitted completely to the 3LCD projector via an HDMI video interface.

[0062] like Figure 3 As shown, in a preferred embodiment of the present invention, the structured light projection module 2 is a 3LCD projector, which includes liquid crystal panels corresponding to the red, green and blue channels respectively, for independently modulating the light intensity of each channel and combining them into a colored striped pattern for projection.

[0063] In this embodiment, after the projector receives the frame signal, its internal image processing circuit routes the brightness data of the three channels R, G, and B to the driving circuits of the corresponding red, green, and blue liquid crystal panels (LCDs) in parallel.

[0064] The high-intensity white light emitted by the projector's light source is separated into three primary color beams—red, green, and blue—by a beam splitter system, and each beam illuminates its corresponding LCD panel. The image on each LCD panel is independently modulated by the intensity of the primary color light passing through it—specifically, the red LCD panel is driven by the R channel data of the composite pattern, modulating the red light intensity and carrying 0° phase information; the green panel is driven by the G channel data, modulating the green light intensity and carrying 120° phase information; and the blue panel is driven by the B channel data, modulating the blue light intensity and carrying 240° phase information.

[0065] The modulated beams of the three primary colors of light are precisely recombined in the color combining prism of the projector, thereby projecting a composite color stripe pattern simultaneously from a projection lens. This pattern physically contains red, green, and blue sinusoidal stripes with different phase shifts.

[0066] Changes in the object's height modulate the fringe phase. The intensity distribution of the three fringes using the three-step phase-shifting method is as follows:

[0067] ;

[0068] in, , , These are the light intensity values ​​collected from the red, green, and blue channels, respectively. , , These represent the background light intensity of each channel, indicating the influence of ambient light and the reflectivity of the object's surface. , , These represent the stripe contrast for each channel; The absolute phase, modulated by the height of the object, is the solution to be obtained.

[0069] like Figure 1 As shown, in a preferred embodiment of the present invention, the imaging module includes:

[0070] Dichroic mirror dichroic filter group 8 is used to separate the reflected composite light into three monochromatic lights: red, green, and blue.

[0071] Three photodetectors are used to receive monochromatic light of the corresponding color and to acquire images.

[0072] The three photodetectors achieve synchronous exposure through hardware triggering.

[0073] In this embodiment, the three photodetectors include a first photodetector 7, a second photodetector 11, and a third photodetector 14. The active high-light suppression unit includes a first electrically controlled rotating polarizer 5, a second electrically controlled rotating polarizer 9, and a third electrically controlled rotating polarizer 12. A convex lens 4 is also provided on the reflected light path.

[0074] A red filter 6 is provided between the first electrically controlled rotating polarizer 5 and the first photodetector 7, a green filter 10 is provided between the second electrically controlled rotating polarizer 9 and the second photodetector 11, and a blue filter 13 is provided between the third electrically controlled rotating polarizer 12 and the third photodetector 14.

[0075] like Figure 1 As shown, in a preferred embodiment of the present invention, the dichroic mirror separation group 8 includes two-stage dichroic mirrors:

[0076] The first-order dichroic mirror is used to reflect red light and transmit green and blue light.

[0077] The second-order dichroic mirror is used to reflect green light and transmit blue light.

[0078] In this embodiment, the reflected light, which is deformed after being modulated by the surface height of the sheet metal part, is first collected and collimated by a main imaging system, namely convex lens 4.

[0079] The collimated polychromatic beam enters the dichroic mirror beam splitter group. This group consists of two-stage dichroic mirrors:

[0080] The first-order dichroic mirror is designed to reflect long-wavelength light (red light with wavelengths > 630nm) while transmitting medium- and short-wavelength green and blue light. The reflected red light stripe image is received by the first photodetector 7.

[0081] The second-stage dichroic mirror receives the transmitted green / blue mixed light. Its film is designed to reflect mid-wavelength light (green light with wavelengths between 500nm and 560nm) while transmitting short-wavelength blue light. The reflected green stripe image is received by the second photodetector 11.

[0082] The remaining blue light (wavelength <480nm) is transmitted directly and received by the third photodetector 14.

[0083] The three photodetectors are connected by a hardware trigger cable and are controlled by a host computer to ensure that they are exposed and acquire images synchronously at exactly the same time, so that pure red, green and blue deformed stripe patterns can be obtained in a single exposure.

[0084] In a preferred embodiment of the present invention, the active specular suppression unit includes:

[0085] Electric rotary table;

[0086] Linear polarization analyzer installed on the electric rotary table;

[0087] The host computer control and processing module 1 controls the rotation of the electric rotary table to adjust the polarization axis angle of the analyzer.

[0088] In this embodiment, the linear polarization analyzer installed on the electric rotary table includes a first electrically controlled rotating polarizer 5, a second electrically controlled rotating polarizer 9, and a third electrically controlled rotating polarizer 12.

[0089] The host computer control and processing module 1 executes a predictive-tracking gradient descent algorithm to dynamically search for and track the optimal polarization angle for suppressing specular highlights.

[0090] The predictive-tracking gradient descent algorithm includes:

[0091] Coarse localization stage: Quickly locate the angle range with the lowest highlight brightness through sparse sampling;

[0092] Fine positioning stage: Gradient descent is performed within the specified angle range to find the optimal suppression angle in a step-by-step manner;

[0093] Tracking phase: Utilizing the continuity of polarization angles between adjacent time moments, the optimal angle of the previous time moment is used as the starting point for the current search.

[0094] In this embodiment, before entering each photodetector, an active specular suppression unit is provided in the optical path, consisting of a high-precision motorized rotary stage (e.g., Thorlabs PRM1Z8) and a visible light broadband high-performance linear polarization analyzer mounted on it (e.g., Thorlabs LPVIZ series).

[0095] The mechanical rotation scheme was chosen based on the pursuit of ultimate measurement accuracy. Compared to liquid crystal devices, high-quality linear polarizers (analyzers) can provide ultra-high extinction ratios far exceeding 100,000:1, but require the selection of polarizers with high transmittance for different wavelengths, and their polarization performance is very stable with changes in wavelength and temperature. This provides the most reliable physical basis for eliminating highlights and performing subsequent precise chromaticity response correction. DC servo motor rotary stages like the PRM1Z8 offer positioning accuracy and repeatability better than 0.1°, ensuring precise angle control.

[0096] According to Malus's law, the intensity of linearly polarized light after passing through an analyzer can be expressed as:

[0097] ;

[0098] in, For the incident light intensity, It is the angle between the polarization axis of the analyzer and the polarization direction of the incident light.

[0099] For polychromatic light (such as R / G / B trichromatic light), the polarization characteristics of different wavelengths may differ, especially since reflection from metal surfaces can introduce wavelength-dependent phase delays. Therefore, the optimal polarization angles required for highlight suppression in different color channels may vary slightly. This invention employs a channel-specific optimization strategy to control the polarization of the R, G, and B light paths separately, further enhancing the suppression effect.

[0100] Traditional Malus's law describes the intensity attenuation of linearly polarized light after passing through an analyzer, but it does not consider the wavelength λ. For specular reflection (highlights) from a metal surface, the polarization state of the reflected light is related to both the wavelength and the angle of incidence.

[0101] Therefore, we can propose a modified model of Malus's law that depends on multiple wavelengths:

[0102] For each color channel c (c∈{R, G, B}), the light intensity after passing through the analyzer is:

[0103] ;

[0104] in, It is the final output light intensity of channel c (red / green / blue). It is the original incident light intensity of the specular highlight in channel c. It is the transmittance coefficient that is wavelength- and angle-dependent. This is the preset analyzer rotation angle for channel c, which is the system control variable. It is the optimal suppression polarization angle for the incident light in channel c.

[0105] Predictive-Tracking Gradient Descent Optimal Angle Search Algorithm:

[0106] First, the system performs pre-sampling at a default angle, quickly locating the highlight area using a brightness thresholding method. If it's the first detection or the target has changed drastically, the system performs a rapid sparse scan for global coarse localization. The rotating stage is commanded to sample the highlight area brightness every 30°, requiring only a few sampling points to pinpoint a general angle range with the lowest brightness. .

[0107] Then the system immediately Centered on the target, gradient descent is used for precise positioning. Specifically, the controller commands the rotary table to... Fine-tuning sampling is performed on both sides at small angles (±5°). The gradient direction (i.e., the direction of the fastest decrease in brightness) is determined based on the brightness change. Then, the iterative search continues along this direction with smaller step sizes (1°) until the optimal suppression angle where the brightness of the highlight area no longer decreases is found. This process only requires a few steps to converge and takes very little time.

[0108] The mathematical description of this process is as follows:

[0109] Define brightness function For angle The intensity of the highlight at that location. The formula for gradient descent is:

[0110] ;

[0111] in, Let the angle be the angle at step k. It is the step size (1°). It is the gradient of the brightness function, representing the rate of change of brightness.

[0112] Since the workpiece being measured moves continuously, the change in the polarization direction of its surface specular highlights must also be continuous. Therefore, at the next moment... Optimal angle It must be at the optimal angle at the current moment. Within the neighborhood of [the location]. Utilizing this prior knowledge, starting from the second measurement, the algorithm skips the global coarse localization step and directly applies the previous time step's [locality]. This serves as the starting point for precise local gradient localization, thereby enabling real-time and efficient tracking of the target's polarization direction.

[0113] The mathematical model of this process can be represented by the following formula:

[0114] ;

[0115] in, It is the change in polarization direction between adjacent time points, which is estimated by the brightness difference between two adjacent frames.

[0116] This active suppression process transforms a high-precision but speed-limited mechanical system into a dynamic system capable of intelligently predicting and precisely matching the constantly changing polarization direction of highlights on a moving workpiece, ensuring optimal highlight suppression at all times. The flowchart of the highlight suppression algorithm is shown below. Figure 4 As shown.

[0117] A photodetector detects the deformed fringes modulated by an object. The phase shift of the fringes varies depending on the height of a point on the same object. This phase shift is used in the phase calculation section to reconstruct the 3D model of the automotive sheet metal parts.

[0118] When the change in surface height causes a phase difference exceeding 2π, the light intensity distribution becomes repetitive, making it impossible to directly distinguish phases of different periods. Therefore, the phase calculation result naturally presents an enclosed state ([-π, π]). By using the light intensity information of the three stripes to eliminate background and contrast, the principal phase value is solved.

[0119] Three-step phase shift method for phase resolution:

[0120] ;

[0121] in, It is the arctangent of the four quadrants, and the output range is [-π, π]. It can be converted to the range [0, 2π] by adding 2π.

[0122] Due to the properties of the arctangent function, the phase can be directly calculated. It is enclosed and discontinuous. To obtain the true phase that corresponds one-to-one with the object's height, phase unwrapping must be performed to obtain the true phase. .

[0123] The core of phase unrolling is detecting and compensating for 2π-integer multiples of transitions in the wrapped phase map. Its basic principle can be stated as follows: Integrating along the spatial path of the phase data, when the phase difference between adjacent pixels exceeds a π-transition threshold, the transition is eliminated by adding or subtracting 2kπ (where k is an integer), restoring phase continuity. Its mathematical expression is:

[0124] ;

[0125] in The integer number of transitions (i.e., the order) required for each pixel. In practical applications, robust algorithms such as quality map-guided paths, least squares, or multi-frequency heterodyne can be used to complete this process to ensure correct unfolding even in the presence of noise and occlusion.

[0126] Finally, using pre-calibrated system parameters, the continuous absolute phase obtained after expansion is... Convert to the object's true height information The three-dimensional shape is reconstructed, and finally the generated three-dimensional point cloud is processed by the host computer software to realize the surface inspection of automotive sheet metal parts.

[0127] The present invention also provides a three-dimensional defect detection method for the system, characterized by comprising the following steps:

[0128] Three sinusoidal fringe patterns with a preset phase difference are fused into a composite color fringe pattern and projected onto the surface of the object under test 3;

[0129] The image of the monochrome channel deformed stripes after beam splitting and highlight suppression is received. In the highlight suppression step, the optimal polarization angle for suppressing highlights is determined based on a prediction-tracking gradient descent algorithm.

[0130] The absolute phase is calculated using phase solution and phase expansion algorithms;

[0131] The three-dimensional shape is reconstructed based on the absolute phase, and defects are detected.

[0132] This invention proposes an efficient single-snapshot 3D measurement method, which solves the problem of low detection efficiency caused by the need for multiple projections in traditional structured light technology to ensure accuracy.

[0133] Three sets of sinusoidal fringes—red (R), green (G), and blue (B)—with fixed phase differences (0°, 120°, and 240°) are projected onto the surface of the object under test in a single exposure using a 3LCD projector. At the imaging end, an optical beam splitting system composed of dichroic mirrors is used to separate the composite light reflected after modulation by the object surface into independent red, green, and blue beams in real time. Three photodetectors are used to receive the corresponding colored light, enabling the simultaneous acquisition of three phase-shifted deformed fringe patterns in a single exposure.

[0134] This design transforms the traditional three-step phase-shift operation in time series into a one-time concurrent processing in color space, greatly improving the data acquisition speed and enabling it to be effectively applied to the detection of fast-moving targets.

[0135] This invention addresses the problem of strong reflective (highlight) interference on metal surfaces such as automotive sheet metal parts by proposing an active and dynamic highlight suppression scheme that overcomes the limitations of traditional fixed polarizers.

[0136] Traditional fixed-angle polarizers cannot cope with the synchronous changes in the position and polarization direction of the highlights caused by workpiece movement or surface changes, resulting in reduced suppression efficiency or failure.

[0137] This invention employs an electrically controlled rotating polarizer and a dynamic polarization matching algorithm that tracks the polarization angle of specular light in real time to rapidly adjust the polarization direction of the polarizer. This process upgrades "passive fixed extinction" to "active dynamic suppression".

[0138] This technology can match the constantly changing polarization direction of the specular light on the moving workpiece in real time, ensuring that the best specular light suppression effect is maintained at all times during dynamic detection, thereby obtaining complete and accurate three-dimensional surface data.

[0139] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A three-dimensional defect detection system based on polychromatic fringe projection, characterized in that, include: The system comprises a host computer control and processing module, a structured light projection module, and an imaging module. The host computer control and processing module is used to generate composite color stripe patterns encoded with phase difference information and to coordinate and control the synchronous operation of the entire system. The structured light projection module is used to receive the composite color stripe pattern and project it onto the surface of the object being measured in one go; The imaging module is used to receive composite color light reflected after being modulated by the surface of the object under test, and to separate it into monochromatic light signals of different colors for synchronous acquisition. The imaging module also includes an active specular highlight suppression unit, which dynamically adjusts the polarization direction according to the instructions of the host computer control and processing module to suppress specular highlights caused by specular reflection. The active specular suppression unit includes: Electric rotary table; Linear polarization analyzer installed on the electric rotary table; The host computer control and processing module controls the rotation of the electric rotary table to adjust the polarization axis angle of the analyzer; The host computer control and processing module executes a predictive-tracking gradient descent algorithm to dynamically search for and track the optimal polarization angle for suppressing specular highlights.

2. The three-dimensional defect detection system based on polychromatic fringe projection according to claim 1, characterized in that, The composite color stripe pattern is obtained by encoding three sinusoidal stripe patterns with a preset phase difference into the red, green, and blue color channels of a color image frame, respectively.

3. The three-dimensional defect detection system based on polychromatic fringe projection according to claim 2, characterized in that, The phase differences of the three sinusoidal fringe patterns are 0°, 120°, and 240°, respectively.

4. The three-dimensional defect detection system based on polychromatic fringe projection according to claim 1, characterized in that, The structured light projection module is a 3LCD projector, which includes liquid crystal panels corresponding to the red, green, and blue channels respectively. These panels are used to independently modulate the light intensity of each channel and combine them into a colored striped pattern for projection.

5. A three-dimensional defect detection system based on polychromatic fringe projection according to claim 1, characterized in that, The imaging module includes: Dichroic mirrors are used to separate the reflected composite light into three monochromatic lights: red, green, and blue. Three photodetectors are used to receive monochromatic light of the corresponding color and to acquire images. The three photodetectors achieve synchronous exposure through hardware triggering.

6. A three-dimensional defect detection system based on polychromatic fringe projection according to claim 5, characterized in that, The dichroic mirror separation group includes two levels of dichroic mirrors: The first-order dichroic mirror is used to reflect red light and transmit green and blue light. The second-order dichroic mirror is used to reflect green light and transmit blue light.

7. A three-dimensional defect detection system based on polychromatic fringe projection according to claim 1, characterized in that, The predictive-tracking gradient descent algorithm includes: Coarse localization stage: Quickly locate the angle range with the lowest highlight brightness through sparse sampling; Fine positioning stage: Gradient descent is performed within the specified angle range to find the optimal suppression angle in a step-by-step manner; Tracking phase: Utilizing the continuity of polarization angles between adjacent time moments, the optimal angle of the previous time moment is used as the starting point for the current search.

8. A three-dimensional defect detection method based on the system according to any one of claims 1-7, characterized in that, Includes the following steps: Three sinusoidal fringe patterns with a preset phase difference are fused into a composite color fringe pattern and projected onto the surface of the object being measured. The image of the monochrome channel deformed stripes after beam splitting and highlight suppression is received. In the highlight suppression step, the optimal polarization angle for suppressing highlights is determined based on a prediction-tracking gradient descent algorithm. The absolute phase is calculated using phase solution and phase expansion algorithms; The three-dimensional shape is reconstructed based on the absolute phase, and defects are detected.

Images