Stereoscopic vision detection method and system for internal defects of multilayer transparent material

Abstract

The invention discloses a stereoscopic vision detection method and system for internal defects of a multilayer transparent material, and relates to the technical field of machine vision and optical detection. The method comprises the steps of system calibration, initial positioning, refraction correction and layer number judgment. Performing system calibration to obtain camera internal and external parameters; in initial positioning, defect apparent three-dimensional coordinates are obtained through feature matching and triangulation; in refraction correction, the apparent coordinates are used as initial values, forward ray tracing is executed, the refraction direction is calculated according to the Snell's law, an optimization objective function is constructed, and the real three-dimensional coordinates are solved by adopting an iterative optimization algorithm; and determining the layer of the defect by layer number determination. By considering the light refraction effect, high-precision three-dimensional positioning and layer number judgment of defects are achieved, and the method has the advantages of being non-destructive, wide in applicability and the like and can meet the quality control requirement of the high-end manufacturing industry.

Description

Technical Field

[0001] This invention relates to the field of machine vision and optical inspection technology, specifically for the precise three-dimensional location of internal defects in industrial ultra-thin glass, multi-layer composite glass, and other transparent layered materials. By combining image processing, computer vision, artificial intelligence, and industrial automation technologies with stereoscopic vision and optical refraction correction, this method provides a high-precision, non-contact solution for detecting internal defects in transparent materials. Background Art

[0002] Multi-layer transparent materials are increasingly used in high-end manufacturing applications such as display panels, specialty glass, and semiconductor packaging. During the production process, these materials (e.g., multi-layer glass and glued laminates) inevitably develop microscopic defects such as bubbles, impurities, and cracks. These defects can severely impact the product's optical performance, mechanical strength, and ultimately, service life.

[0003] Currently, optical imaging methods are commonly used to detect defects in transparent materials. However, existing technologies generally face a core challenge: when using cameras for inspection, because light passes through multiple layers of media with different refractive indices, the defects observed in the image are merely their two-dimensional projections. Systems using a single camera or traditional binocular stereo vision can detect the presence of defects, but ignore the refraction of light as it passes through the material surface, resulting in significant errors in three-dimensional positioning. This error prevents the inspection system from accurately determining the specific layer where the defect is located (for example, the protective glass layer, the functional adhesive layer, or the core substrate glass layer). If the depth (Z-axis) coordinate of the defect cannot be determined, it is impossible to assess whether the defect affects the core functional layers of the product, resulting in difficult quality control. This can lead to products that are not harmful to use being misclassified as defective or products with fatal defects entering the market.

[0004] Therefore, the industry urgently needs a technical solution that can penetrate multiple layers of transparent materials, accurately measure the three-dimensional spatial coordinates of internal defects, and determine the physical layer number in which they are located. Summary of the Invention

[0005] The present invention aims to overcome the shortcomings of existing technologies by providing a method and system for high-precision three-dimensional localization of defects within multi-layer transparent materials. By combining stereo vision triangulation with a refraction correction algorithm based on a physical model, this method aims to address positioning errors caused by light refraction, thereby accurately determining the true physical location of the defect and pinpointing its specific material layer.

[0006] The present invention provides a method for stereoscopic vision detection of internal defects in multi-layer transparent materials, comprising the following steps: Step S1: System calibration, obtaining the internal and external parameters of at least two cameras in the stereo vision system, wherein the internal parameters include the intrinsic parameter matrix and the distortion coefficient, and the external parameters include the rotation matrix and the translation vector; by accurately obtaining the internal and external parameters of the camera, an accurate camera imaging geometry model is established to reduce the defect positioning deviation caused by the camera parameter error and ensure the accuracy of the projection conversion from three-dimensional space to two-dimensional image.

[0007] Step S2: Initial positioning: matching the same defect point in the images captured by the at least two cameras, and calculating the apparent three-dimensional coordinates of the defect point through a triangulation algorithm; using the parallax information of multiple camera perspectives, the approximate three-dimensional position of the defect is quickly obtained, providing an initial guess close to the true value for subsequent refraction correction, avoiding the blindness of directly performing complex refraction calculations, and improving the efficiency of the overall detection process.

[0008] Step S3: Refraction correction, using the apparent three-dimensional coordinates as the initial value, based on the thickness and refractive index of each layer of the multi-layer transparent material, perform forward ray tracing, starting from the current guessed true defect position, emit the initial light to the optical center of the camera; calculate the refraction direction of the light after passing through the interface of each layer of the medium according to Snell's law; construct an optimization objective function, which is the reprojection error between the predicted pixel and the actual measured pixel; solve the true three-dimensional coordinates through an iterative optimization algorithm to minimize the reprojection error; simulate the light propagation path through forward ray tracing, combine Snell's law to accurately calculate the refraction direction, use the optimization objective function to quantify the difference between the prediction and the actual, and then eliminate the positioning error caused by light refraction through an iterative optimization algorithm, thereby significantly improving the accuracy of the three-dimensional positioning of the defect and making the positioning result closer to the real physical position.

[0009] Step S4: Layer number determination. The final true 3D coordinates are compared with the physical boundaries of each material layer to determine the layer where the defect is located. Accurately determining the specific physical layer where the defect is located provides a direct and critical basis for product quality assessment, avoiding misjudgments caused by the inability to determine the number of defective layers.

[0010] Furthermore, in the refraction correction step, the initial light direction vector of the forward ray tracing is , where OL is the optical center of the camera and Preal is the actual defect position guessed at present, which ensures the accuracy of the starting point and direction of the forward ray tracing and improves the accuracy of the light propagation path simulation.

[0011] Furthermore, in the refraction correction step, the direction vector after refraction is calculated according to the vector form of Snell's law, and the vector form of Snell's law is: ;in is the direction vector after refraction, η is the ratio of the refractive indices of the two media, that is, η=n1 / n2, is the direction of incident light, is the interface normal vector. Using the vector form of Snell's law can more accurately calculate the refraction direction of light at the interface of different media. This takes into account the spatial vector characteristics of light propagation and further improves the accuracy of refraction direction calculation compared to the scalar form, thereby improving the accuracy of defect location.

[0012] Furthermore, the optimization objective function is , where p L 、p R is the pixel coordinate of the defect point actually measured on the left and right camera imaging planes, p L ’ 、p R ’ The predicted pixel coordinates on the left and right camera imaging planes after refraction from the true 3D coordinates are given. This optimization objective function quantitatively describes the difference between the predicted and measured pixels, providing a clear direction for iterative optimization. By minimizing this function, we can continuously approximate the true 3D coordinates of the defect, ensuring the accuracy of the refraction correction results.

[0013] Furthermore, the iterative optimization algorithm adopts a nonlinear optimization algorithm to minimize the objective function by iteratively updating the guess value of the real three-dimensional coordinates, that is, The nonlinear optimization algorithm can effectively handle the complex nonlinear problems caused by light refraction. By continuously iteratively updating the guess value of the true three-dimensional coordinates, the objective function is minimized, thereby accurately solving the true three-dimensional coordinates of the defect and further improving positioning accuracy.

[0014] Furthermore, in the system calibration step, the camera intrinsic parameter calibration is achieved by the following formula: ; where (u,v) is the pixel coordinate, (X C ,Y C ,Z C ) is the three-dimensional point coordinate in the camera coordinate system, s is the scale factor, (f x ,f y ) is the focal length, (C x ,C y ) is the main point.

[0015] This formula accurately establishes the mapping relationship between three-dimensional points and image pixel coordinates in the camera coordinate system. By calibrating internal parameters such as focal length and principal point, the accuracy of the conversion from three-dimensional points to two-dimensional pixel coordinates is ensured.

[0016] Furthermore, the triangulation algorithm is implemented by solving the following overdetermined equations: and , where s L and sR is the scale factor, (u L ,v L ) and (u R ,v R ) are the pixel coordinates of the defect point in the left and right images respectively, K L and K R where P is the intrinsic parameter matrix for the left and right cameras, R is the rotation matrix, and t is the translation vector. By solving the overdetermined system of equations and leveraging the image information and camera parameters of the left and right cameras, the apparent 3D coordinates of the defect can be calculated more accurately, reducing the error introduced by a single perspective.

[0017] The present invention also provides a stereoscopic vision detection system for internal defects of multi-layer transparent materials, comprising: a light source module: arranged at the bottom layer, which is a programmable LCD or OLED screen for providing uniform backlight upward; a product support module for horizontally placing the multi-layer transparent material to be tested; a stereoscopic vision module: at least two cameras, arranged symmetrically above the product in a "V"-shaped convergent layout, with their optical axes intersecting at or near the plane where the product is located, forming a convergence angle of 30° to 60°, for synchronously capturing product images; a processing module: connected to the stereoscopic vision module. The light source module provides uniform and programmable backlight to ensure clear defect imaging and improve the defect recognition rate; the product support module stably places the material to be tested to prevent its position change from affecting the detection results; the "V"-shaped convergent layout and appropriate convergence angle of the stereoscopic vision module ensure sufficient parallax and measurement range, and synchronous acquisition ensures image time consistency; the processing module executes the corresponding detection method to achieve automated and high-precision detection of defects. The modules work together to improve the overall detection performance and reliability of the system.

[0018] In the present invention, system calibration provides accurate camera parameters for initial positioning. The apparent coordinates of the initial positioning serve as the initial values ​​of refraction correction. The real coordinates obtained by refraction correction provide a basis for layer number determination, and collaboratively realize the complete detection process from coarse positioning to precise positioning and then to layer number determination.

[0019] Refraction correction involves the coordinated efforts of forward ray tracing, the application of Snell's law, the construction of an optimization objective function, and an iterative optimization algorithm. Forward ray tracing provides the path for refraction calculation, while Snell's law ensures accurate calculation of the refraction direction. The optimization objective function provides direction for iterative optimization, and the iterative optimization algorithm efficiently finds the optimal solution. Together, they eliminate refraction errors and achieve high-precision positioning.

[0020] The nonlinear screen appearance defect evaluation method of the present invention has the following technical effects: High-precision positioning: By introducing a refraction correction model based on Snell's law, combined with forward ray tracing, optimization objective function construction and Levenberg-Marquardt iterative optimization algorithm, the measurement error caused by light path refraction is greatly eliminated, achieving submillimeter or even micron-level three-dimensional positioning accuracy for internal defects, which is more accurate than traditional positioning methods.

[0021] Accurate layer determination: Based on high-precision depth coordinates, the specific physical layer where the defect is located can be reliably determined, providing a direct and critical basis for product quality assessment, avoiding misjudgments and missed judgments, and improving the accuracy of quality control.

[0022] Non-destructive testing: The entire testing process is non-contact optical measurement, which will not cause any physical damage to the product. It can be used to test various types of precise and fragile multi-layer transparent materials without affecting the subsequent use of the product.

[0023] Wide applicability: It is suitable for complex layered transparent materials containing multiple layers of glass and multiple layers of adhesive layers with different refractive indices. It can play a role in many fields such as display panels, special glass, semiconductor packaging, etc., and has broad application prospects.

[0024] High degree of automation: Through the processing module, a series of algorithms are executed to realize automated detection from image acquisition to result output, reducing manual intervention, improving detection efficiency, and meeting the batch detection needs in industrial production.

[0025] In order to more clearly illustrate the structural features and effects of the present invention, the present invention is described in detail below with reference to the accompanying drawings and specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 A side view of a stereoscopic vision detection system for internal defects of multi-layer transparent materials according to the present invention; Figure 2 This is a top view of a stereoscopic vision detection system for internal defects of multi-layer transparent materials according to the present invention. DETAILED DESCRIPTION

[0027] The present invention will be further described below with reference to the accompanying drawings and related knowledge, and described clearly and completely. Obviously, the described applications are only part of the embodiments of the present invention, rather than all of the embodiments.

[0028] Example 1: The present invention provides a method for stereoscopic vision detection of internal defects in multi-layer transparent materials, comprising the following steps: System calibration: Obtaining the internal and external parameters of at least two cameras in the stereo vision system. The internal parameters include the intrinsic matrix and distortion coefficients, and the external parameters include the rotation matrix and translation vector. Specifically, camera intrinsic calibration: obtain the intrinsic parameter matrix K and distortion coefficient of each camera. The intrinsic parameter matrix describes the projection relationship from a 3D point to a 2D image: ; Among them, (u, v) is the pixel coordinate, (X C ,Y C ,Z C ) is the three-dimensional point coordinate in the camera coordinate system, s is the scale factor, (f x ,f y ) is the focal length, (C x ,C y ) is the main point.

[0029] External parameter calibration: solve the relative position relationship between the two camera coordinate systems, that is, the rotation matrix R and the translation vector t. The coordinates P of a three-dimensional point in the left and right camera coordinate systems L and P R Satisfaction relationship: P R =R·P L +t; Initial positioning: matching the same defect point in the images captured by the at least two cameras, and calculating the apparent three-dimensional coordinates of the defect point through a triangulation algorithm; Specifically, this step is used to obtain an initial three-dimensional coordinate of the defect. Feature matching: In the images captured by the two cameras, the same physical defect point is matched to obtain its pixel coordinates P in the left and right images. L =(u L ,v L ) and P R =(u R ,v R ); Further triangulation: Based on the calibrated parameters, an overdetermined set of equations is solved to find the spatial point P. When the world coordinate system is set at the left camera, the set of equations is as follows: ; ; Solving the equations by the least squares method, we can obtain the “apparent coordinates” P ignoring refraction. apparent .

[0030] Refraction correction: Using the apparent 3D coordinates as the initial values, forward ray tracing is performed based on the thickness and refractive index of each layer of the multi-layer transparent material. The initial light is emitted from the current estimated true defect location toward the camera's optical center. The refraction direction of the light after passing through each layer of the medium interface is calculated according to Snell's law. An optimization objective function is constructed, which is the reprojection error between the predicted pixels and the actual measured pixels. The true 3D coordinates are solved through an iterative optimization algorithm to minimize the reprojection error. Specifically, this step is the core innovation of the present invention, which solves the real physical position P of the defect through iterative optimization. real Physical model: The thickness d of each layer of the product to be tested is known i and refractive index n i The interfaces of each layer are parallel planes, and their normal vectors are .

[0031] More specifically: Iterative optimization process: Initial guess: The obtained apparent coordinate P apparent As the real physical location P real The initial guess value of .

[0032] Forward ray tracing: From the current guess of the true defect position P real Starting from the optical center of the camera (for example, the left camera O L ) emits an initial ray with a direction vector of .

[0033] Apply Snell's law: Every time a ray crosses a medium interface, its direction vector changes. Use the vector form of Snell's law to calculate the direction vector after refraction. If the incident light direction is , the interface normal vector is , the refractive indices of the two media are n1 and n2 respectively (let η = n1 / n2), then: ; This calculation starts at the layer where the defect is located and traces outward layer by layer until the light enters the air.

[0034] Calculate the predicted pixel coordinates: The final light after refraction through all interfaces will intersect with the camera imaging plane to form a predicted pixel coordinate. This operation is performed on both the left and right cameras to obtain a pair of predicted coordinates (p L ’ ,p R ’ ).

[0035] Construct optimization objective function: The optimization goal is to find a real coordinate P real , so that it minimizes the reprojection error between the predicted pixels and the actual measured pixels. The objective function E(Preal ) is defined as: ; Iterative optimization: Use nonlinear optimization algorithms (such as Levenberg-Marquardt) to minimize the objective function E(P real ) as the goal, continuously iteratively update P real The guess value of . ; Get the final result: When the error converges to the minimum value, the optimization process ends. That is, the true and accurate three-dimensional coordinates of the defect inside the product after taking into account the refraction effect.

[0036] Layer number determination: Compare the final obtained true three-dimensional coordinates with the physical boundaries of each layer of the material to determine the layer where the defect is located. Specifically, the final obtained true depth coordinate Z real By comparing with the known depth boundaries of each layer of material, it can be determined in which physical layer of the product the defect is located.

[0037] By introducing a refraction correction model based on Snell's law and combining it with a nonlinear optimization algorithm, the present invention greatly eliminates the measurement error caused by light path refraction and achieves submillimeter or even micron-level three-dimensional positioning accuracy for internal defects.

[0038] Based on high-precision depth coordinates, the present invention can reliably determine the specific physical layer where the defect is located, providing a direct and critical basis for product quality assessment and avoiding misjudgment and missed judgment.

[0039] The entire inspection process of this invention is non-contact optical measurement, which does not cause any physical damage to the product. It is not only applicable to double-layer glass, but also to complex layered transparent materials containing multiple layers of glass and multiple layers of adhesive layers with different refractive indices.

[0040] Example 2, reference Figure 1 、 Figure 2 As shown, the present invention also provides a stereoscopic vision detection system for internal defects of multi-layer transparent materials, comprising: a light source module: arranged at the bottom layer, which is a programmable LCD or OLED screen for providing uniform backlight upward; a product support module for horizontally placing the multi-layer transparent material to be tested; a stereoscopic vision module: at least two cameras, arranged symmetrically above the product in a "V"-shaped convergent layout, with their optical axes intersecting at or near the plane where the product is located, forming a convergence angle of 30° to 60°, for synchronously capturing product images; a processing module: connected to the stereoscopic vision module; In the present invention, the detection system adopts a transmission stereo deflectometry layout, and its core components are arranged vertically, from top to bottom: Camera layer: Two or more high-resolution industrial cameras are arranged in a "V"-shaped, converging three-dimensional configuration. The cameras are symmetrically positioned about the product center, with their optical axes tilted inward and converging near the surface of the product being measured, forming a convergence angle of 30° to 60°. A fixed horizontal distance, known as the baseline, is maintained between the optical centers of the two cameras.

[0041] Product layer: The multi-layer transparent material product to be tested is placed horizontally on a high-rigidity, optically transparent support platform.

[0042] Light source layer: An LCD or OLED display screen that can display high-frequency stripe patterns is placed horizontally at the bottom of the system and serves as a uniform, programmable backlight source.

[0043] Specific reference Figure 1 : Side view of the inspection system of the present invention, showing the vertical "V"-shaped converging layout of the camera, product and light source. Figure 2 :Top view of the detection system of the present invention, showing the baseline distance between the two cameras and their overlapping stereo measurement area.

[0044] exist Figure 1 The figure shows the core structure of the system from a side view. At the bottom is a display screen 1, serving as a backlight. A multi-layer glass product 2 to be tested is placed on top, with a defect 3 marked inside the product. At the top are the first and second cameras 4 and 5, arranged in an inverted "V" shape. Their convergence angle 6 ensures their fields of view are aligned with the product. Figure 1 FIG. 7 schematically shows the path of the light path 7 emitted from the defect point to the two cameras.

[0045] Figure 2 The system layout is shown from above. A horizontal baseline distance exists between the first camera 4 and the second camera 5. Each camera has a field of view (FOV1, FOV2). The area 8 where the two fields of view overlap is the valid stereo measurement area, and defects must be located within this area.

[0046] Example 3, a system for detecting internal defects of mobile phone cover glass, its hardware is as follows Figure 1 and Figure 2 Build as shown.

[0047] System components: Camera: Two industrial cameras with a resolution of 5120x5120 pixels, namely the first camera 4 and the second camera 5.

[0048] Light source: a high-brightness OLED screen (display screen 1).

[0049] Product 2: A multi-layer glass panel consisting of two 0.5mm thick layers of glass bonded together with a 0.1mm thick OCA adhesive layer. The refractive index of each layer is known.

[0050] Implementation process: 1. System calibration: First, place a high-precision checkerboard calibration plate on the product support platform.

[0051] Start the first camera and the second camera to capture images of the calibration plate from 15 different poses.

[0052] Use calibration software (such as the functions in the OpenCV library) to calculate the intrinsic parameter matrices K1, K2 and distortion coefficients of the two cameras respectively.

[0053] Using the same captured image, calculate the rotation matrix R and translation vector t from the first camera coordinate system to the second camera coordinate system. Save all calibration parameters.

[0054] 2. Defect image acquisition and initial positioning: Remove the calibration plate and place the multi-layer glass plate to be tested on the supporting platform.

[0055] The OLED screen displays a pure white background, and the two cameras expose synchronously to capture images containing internal defects.

[0056] In the left and right images, the center of the defect is found by image processing algorithms (such as spot detection) and its pixel coordinates P are obtained. L and P R .

[0057] Call the triangulation function, input the calibration parameters and pixel coordinate pairs (P L , P R ), calculate the apparent three-dimensional coordinates of the defect P apparent Assume that the calculation result is (10.25, 20.41, -0.85) mm.

[0058] Refraction correction and precise positioning: Start the refraction correction optimization program. apparent As the real coordinate P real The initial value of .

[0059] The program enters an iterative loop and minimizes the reprojection error function using the Levenberg-Marquardt algorithm .

[0060] In each iteration, the program calculates the current P real The predicted pixel P is calculated by using the vector form of Snell's law to perform forward ray tracing. L ’ and P R , and update P real .

[0061] After the cycle is completed, the final optimization result is obtained For example, the final coordinates are (10.23, 20.39, -0.58) mm.

[0062] Layer determination: It is known that the Z coordinate of the upper surface of the product is 0, the thickness of the first layer of glass is 0.5mm, the thickness of the glue layer is 0.1mm, and the thickness of the second layer of glass is 0.5mm.

[0063] The Z coordinates of the boundaries of each layer are: 0 mm (upper surface), -0.5 mm (glass 1-glue layer), -0.6 mm (glue layer-glass 2), and -1.1 mm (lower surface).

[0064] The final defect depth coordinate Z real =-0.58mm. Because -0.6<-0.58<-0.5, the system determines that the defect is located in the middle OCA layer. This result will be recorded and used for subsequent quality control decisions.

[0065] The technical principles of the present invention have been described above in conjunction with specific embodiments, which are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions based on the principles of the present invention fall within the scope of protection of the present invention. Those skilled in the art will be able to conceive of other specific embodiments of the present invention without inventive effort, and such methods will fall within the scope of protection of the present invention.

Claims

1. A method for detecting internal defects of multi-layer transparent materials using stereoscopic vision, characterized in that: The following steps are involved: Step S1: Acquire internal parameters and external parameters of at least two cameras in a stereo vision system, wherein the internal parameters include an intrinsic parameter matrix and a distortion coefficient, and the external parameters include a rotation matrix and a translation vector; Step S2: matching the same defect point in the images captured by the at least two cameras, and calculating the apparent three-dimensional coordinates of the defect point by a triangulation algorithm; Step S3: Using the apparent 3D coordinates as the initial value, forward ray tracing is performed based on the thickness and refractive index of each layer of the multi-layer transparent material. Starting from the currently estimated true defect location, an initial ray is emitted toward the optical center of the camera. The refraction direction of the ray after passing through each layer of the medium interface is calculated according to Snell's law. An optimization objective function is constructed, which is the reprojection error between the predicted pixels and the actual measured pixels. The true 3D coordinates are solved through an iterative optimization algorithm to minimize the reprojection error. Step S4: Compare the final obtained true three-dimensional coordinates with the physical boundaries of each material layer to determine the layer where the defect is located.

2. The method according to claim 1, characterized in that In step S3, the initial light direction vector of the forward ray tracing is , where O L is the optical center of the camera, P real is the actual defect location currently guessed.

3. The method according to claim 1, characterized in that In step S3, the direction vector after refraction is calculated according to the vector form of Snell's law. The vector form of Snell's law is: ;in is the direction vector after refraction, η is the ratio of the refractive indices of the two media, that is, η=n1 / n2, n1, n2 are the refractive indices of the two media, is the direction of the incident light, is the interface normal vector.

4. The method according to claim 2, characterized in that The optimization objective function is , where P L 、P R is the pixel coordinate of the defect point actually measured on the imaging plane of the left and right cameras, P L ’ 、P R ’ are the predicted pixel coordinates on the left and right camera imaging planes after refraction from the real 3D coordinates.

5. The method according to claim 4, characterized in that The iterative optimization algorithm uses a nonlinear optimization algorithm to minimize the objective function by iteratively updating the guess value of the real three-dimensional coordinates, that is, .

6. The method according to claim 1, wherein In step S1, the camera internal parameter calibration is achieved by the following formula: ; Among them, (u, v) is the pixel coordinate, (X C ,Y C ,Z C ) is the three-dimensional point coordinate in the camera coordinate system, s is the scale factor, K is the camera internal parameter matrix, (f x ,f y ) is the focal length, (C x ,C y ) is the main point.

7. The method according to claim 1, characterized in that The triangulation algorithm is implemented by solving the following overdetermined system of equations: and , where s L and s R is the scale factor, (u L ,v L ) and (u R ,v R ) are the pixel coordinates of the defect point in the left and right images respectively, K L and K R are the intrinsic parameter matrices of the left and right cameras respectively, P is the spatial point, R is the rotation matrix, and t is the translation vector.

8. A stereoscopic vision detection system for internal defects of multi-layer transparent materials, characterized in that: Includes: Light source module: set at the bottom layer, it is a programmable LCD or OLED screen used to provide uniform backlight upwards; The product support module is used to horizontally place the multi-layer transparent material to be tested; the stereoscopic vision module: at least two cameras are symmetrically arranged above the product in a "V"-shaped convergence layout, and their optical axes intersect at or near the plane where the product is located, forming a convergence angle of 30° to 60° for synchronously capturing product images; the processing module: connected to the stereoscopic vision module, used to execute the detection method described in any one of claims 1-7.

Images