Structured light laser three-dimensional scanning method and system based on super eight-position camera
By using an 8-bit industrial camera and high depth image processing technology, the problems of insufficient dynamic range and noise sensitivity of existing systems in complex industrial scenarios are solved, achieving high-precision and robust 3D scanning results.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-08
- Publication Date
- 2026-04-03
AI Technical Summary
Existing structured light 3D scanning systems based on 8-bit depth industrial cameras suffer from insufficient dynamic range, leading to feature loss, limited detail resolution and measurement accuracy, and poor robustness when facing complex industrial scenarios.
Employing an 8-bit ultra-high-resolution industrial camera with a grayscale resolution exceeding 8 bits, combined with a synchronization control module, a supplementary lighting module, and a data processing module, high-precision 3D point cloud reconstruction is achieved through high-bit-depth image preprocessing and sub-pixel edge extraction algorithms.
It significantly improves the integrity and coverage of 3D measurements, reduces data gaps and noise, enhances measurement accuracy and robustness, reduces algorithm complexity, and is suitable for complex industrial environments.
Smart Images
Figure CN121783043A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of structured light measurement technology, specifically to a structured light laser three-dimensional scanning method and system based on an 8-bit camera. Background Technology
[0002] Structured light 3D scanning technology, as an important branch of modern optical non-contact measurement, has been widely applied in industrial inspection, reverse engineering, robot navigation, and quality control due to its high precision, high efficiency, and good field adaptability. The basic principle of this technology is to project an coded structured light pattern onto the surface of the object being measured using a specific projection device, such as a laser. Due to the modulation of the object's three-dimensional shape, these patterns deform. Subsequently, one or more cameras capture the deformed pattern from different angles, and finally, based on the principle of triangulation, image processing algorithms are used to calculate the object's three-dimensional point cloud data.
[0003] In existing industrial solutions, especially those based on laser line scanning, the vast majority of systems use standard 8-bit depth industrial cameras as image acquisition units. An 8-bit camera can only provide image data with 256 grayscale levels, which is sufficient for basic needs when dealing with idealized objects with uniform surface reflectivity. However, when facing real, complex industrial scenarios, such systems suffer from the following technical bottlenecks: First, insufficient dynamic range leads to feature loss: industrial test surfaces often exhibit significant differences in reflectivity. For an 8-bit camera, with a single exposure setting, it is difficult to simultaneously ensure that laser stripes in highly reflective areas are not overexposed and those in dark areas are not underexposed, easily resulting in holes or excessive noise in the 3D data. Second, limited detail resolution and measurement accuracy: the 256 gray levels of an 8-bit camera result in a distinct "step-like" gray-scale distribution curve in the cross-section of the stripes, which is not smooth and continuous. This fundamentally limits the positioning accuracy and repeatability of the sub-pixel algorithm, leading to systematic errors. Finally, it is sensitive to environmental noise and has poor robustness: in low-reflectivity scenarios, the weak laser stripe signal has a low signal-to-noise ratio in 8-bit images, making the stripe extraction algorithm highly susceptible to noise interference, resulting in broken, burred, or positionally drifted stripes.
[0004] To alleviate the above problems, existing technologies typically employ optimization algorithms, multiple scans, or complex multi-exposure image fusion techniques. However, these methods either severely sacrifice measurement efficiency or introduce additional operational complexity and errors.
[0005] Therefore, there is an urgent need in this field for a structured light 3D scanning technology that can capture high dynamic range scene details in a single exposure and achieve higher precision and stronger robustness. Summary of the Invention
[0006] The purpose of this invention is to provide a structured light laser 3D scanning method and system based on an 8-bit camera. By improving grayscale resolution from the image acquisition source, the extraction quality of laser stripes and marker points is fundamentally improved, thereby achieving high-precision and high-completeness 3D measurement.
[0007] To this end, the present invention provides a structured light laser 3D scanning system based on an 8-bit camera, including at least one industrial camera with a grayscale resolution of more than 8 bits, for real-time acquisition of images of the object being measured. A structured light projector is used to project an coded structured light laser pattern onto the surface of a measured object at precise moments. A synchronization control module is electrically connected to the above 8-bit industrial camera and structured light projector to coordinate and control the timing of image acquisition and structured light projection. The supplementary lighting module is electrically connected to the synchronization control module and is used to receive instructions from the synchronization module to provide controllable background lighting; The data transmission module is used to receive and process image data acquired by the over 8-bit industrial camera; The data processing module is used to receive image data and perform 3D reconstruction calculations; The data display module is used to display the final 3D results to the user in real time.
[0008] Preferably, the structured light laser 3D scanning system further includes multiple reflective markers, which are arranged on the object being measured or in the scanning environment to provide spatial reference and point cloud registration.
[0009] Preferably, the 8-bit industrial camera is an image camera with a grayscale resolution of 12 bits or 16 bits.
[0010] Preferably, the system includes two of the aforementioned 8-bit industrial cameras, forming a binocular vision system.
[0011] A structured light laser 3D scanning method based on an 8-bit camera includes the following steps: S1. The structured light projector is triggered by the synchronous control module to project a structured light pattern, and the 8-bit industrial camera is triggered to capture one or more frames of images containing the structured light pattern and reflective markers. S2. Preprocess the raw image data acquired by the 8-bit industrial camera to maximize the dynamic range of useful image information and avoid directly using fixed parameters designed for 8-bit images. S3. From the preprocessed image, candidate regions for marker points are initially identified through Blob analysis, then sub-pixel edge extraction is performed, and the sub-pixel center position of the marker points is determined by ellipse fitting or circle fitting. High bit depth data makes the edge point distribution smoother, the fitting circle center accuracy is higher, and the error can be reduced to 0.01-0.02 pixels. S4. For the sub-pixel center position, extract laser lines of a certain pixel width by row, connect them vertically to form candidate laser connected regions, and use the gray-scale centroid method to extract laser sub-pixels by row. S5. Based on features extracted from images from different perspectives, feature matching and 3D reconstruction are performed to generate a 3D point cloud. For a binocular system, the principle of symmetrical geometry is used to match the laser stripe points and marker points extracted from the left and right camera images. The 3D coordinates of the matching points are calculated using triangulation to generate the 3D point cloud. The reconstruction accuracy and matching correctness can be determined by methods such as reconstructing the distance from the laser point to the light plane.
[0012] Preferably, the specific steps for preprocessing the original image data in step S2 are as follows: First, directly read the camera's 12-bit or 16-bit raw high-bit RAW data, and use a background image with over 8-bit precision to perform image background subtraction in order to retain more shadow details; Then, adaptive contrast stretching or normalization is performed, and the gray-level distribution histogram of the entire high-bit-depth image is analyzed. This is used to maximize the contrast of the image without losing any internal details, making faint stripes clear, while suppressing overly bright stripes without saturating them. High-bit-depth data effectively avoids the interference of saturated pixels on the centroid method and provides a near-ideal input signal for the fitting method, enabling the center extraction accuracy to reach the 0.01 pixel level.
[0013] Preferably, in step S3, at least one of the grayscale centroid method, Gaussian fitting method, or Steger algorithm is used to extract the center line of the structured light stripes at the subpixel level.
[0014] Preferably, for a binocular vision system, the epipolar geometry principle is used to match the corresponding laser stripe points and marker points in the left and right camera images, and their three-dimensional coordinates are calculated by triangulation.
[0015] Preferably, the structured light laser three-dimensional scanning method supports working mode switching, including: In the full-scene 8-bit mode, image acquisition is performed using the 8-bit mode in all scanning scenarios. In selective 8-bit mode, switch to 8-bit mode for data acquisition in special scenarios or for specific areas where fine scanning is required; In scenarios where scanning speed is paramount, use the normal 8-bit mode for data acquisition.
[0016] Preferably, the switching of working modes is achieved through computer-side interaction, device-side interaction, or automatic scene recognition by the system.
[0017] The present invention proposes a structured light laser 3D scanning method and system based on an 8-bit camera, the advantages of which are as follows: This application uses an 8-bit camera for scanning. The 8-bit camera provides images with higher grayscale resolution, far exceeding the 256 levels of traditional 8-bit cameras. This allows the system to capture laser stripe details in both highly reflective and dark areas in a single exposure, avoiding overexposure or underexposure, significantly improving the integrity and coverage of the 3D point cloud, and reducing data holes and noise in the point cloud. High-depth imaging makes the grayscale distribution curves of laser stripes and marker points smoother and more continuous, providing a near-ideal input signal for sub-pixel algorithms. Laser stripe center extraction accuracy can reach the 0.01 pixel level, and marker point center fitting error is reduced to 0.01–0.02 pixels, improving overall 3D measurement accuracy. It maintains a high signal-to-noise ratio even in low-light or low-reflectivity scenarios, effectively suppressing random noise interference. Laser stripe extraction is more stable, reducing breaks, burrs, and positioning drift, thus improving system reliability in complex industrial environments. The system supports full-scene super 8-bit mode, selective super 8-bit mode and normal 8-bit mode. Users can switch modes according to their scanning needs through computer interaction, device interaction or automatic scene recognition by the system, balancing measurement accuracy and efficiency. High bit depth data reduces the need for traditional multi-exposure HDR fusion or complex post-processing, lowers algorithm complexity and computational overhead, and improves the real-time performance and efficiency of 3D reconstruction. This system is suitable for complex surfaces such as highly reflective metals, dark light-absorbing materials, and transparent objects, and has broad application prospects in fields such as industrial inspection, reverse engineering, robot navigation, and quality control. Attached Figure Description
[0018] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0019] Figure 1 This is a system framework diagram of the structured light laser three-dimensional scanning system of the present invention; Figure 2 This is a flowchart of the structured light laser three-dimensional scanning method of the present invention; Figure 3 This is a comparison diagram of laser processing according to the present invention; Figure 4 This is a comparison diagram of the marker point processing of the present invention. Detailed Implementation
[0020] To make the objectives, technical solutions, and beneficial effects of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described in this specification are merely for explaining the invention and are not intended to limit the invention.
[0021] Example 1: Please see Figure 1-2 The present invention provides a structured light laser 3D scanning system based on an 8-bit camera, comprising two 12-bit grayscale industrial cameras, Cam1 and Cam2, forming a binocular vision system; A structured light laser, the structured light laser projection module uses an encoded structured light laser pattern (such as parallel laser lines), the projection angle and density can be adjusted according to the object being measured; Several circular reflective markers are arranged on the surface of the object being measured or in the scanning environment to provide a spatial reference, along with a synchronization control module. Precise timing pulses are generated to coordinate laser projection, camera exposure, and the triggering of the supplementary lighting module. The system also includes a supplementary lighting module using a controllable LED light source, providing adaptive background illumination based on surface reflection characteristics, a data transmission module (such as a gigabit network card), a data processing module, and a data display module.
[0022] The workflow is as follows: the synchronous control module triggers the structured light laser to project a structured light pattern, and simultaneously triggers an 8-bit industrial camera to acquire one or more frames of images. Two 12-bit cameras are exposed simultaneously to acquire images with laser stripes and markers. The cameras send the 12-bit raw image data to the data processing module through the data transmission module. The images contain laser stripes and reflective markers. The raw data is in 12-bit RAW format. The supplementary lighting module provides auxiliary illumination during exposure to ensure that details in the dark areas are visible. The data transmission module receives the raw high-bit-depth image, directly reads the 12-bit RAW data to avoid in-machine compression or down-biting, uses a background image with over 8-bit precision for background subtraction, preserves dark details, and performs adaptive contrast stretching on the 12-bit raw data (range 0-4095) to linearly map the effective grayscale range to the full range, enhancing contrast. Laser line extraction: An adaptive thresholding method is used to locate the laser stripe region, and then Gaussian fitting is used to fit the cross-sectional grayscale distribution curve of each laser line to extract the sub-pixel center line. Thanks to the smooth curve of the 12-bit data, the fitting accuracy is high and stable. Marker point extraction: Candidate marker points are found through Blob analysis, sub-pixel edge extraction is performed, and then ellipse fitting is used to obtain the sub-pixel center of the marker point. Adaptive contrast stretching: The image grayscale histogram is analyzed, and the darkest 1% and brightest 1% of pixels are removed as noise and saturation areas. The middle 98% of pixels are linearly mapped to the full dynamic range, enhancing the visibility of weak stripes while suppressing overly bright stripes without saturation. Blob analysis was used to initially identify candidate regions for marker points. The Canny operator was used for sub-pixel edge extraction. Ellipse or circle fitting was performed on the edge points. High bit depth data made the edge point distribution smoother. The accuracy of the fitted circle center could reach 0.01–0.02 pixel error. The fitting accuracy was verified and unqualified marker points were removed. Laser lines of a certain pixel width are extracted row by row, and connected vertically to form candidate laser connected regions. For each connected region, the sub-pixel center line of the laser is extracted row by row using the gray-scale centroid method. High bit depth data avoids interference from saturated pixels, and the centroid calculation is more accurate. Alternative solutions: Gaussian fitting or Steger's algorithm can be used for center line extraction, both with an accuracy of 0.01 pixels.
[0023] For binocular vision systems, epipolar geometry is used to match corresponding laser stripe points and marker points in the left and right camera images. The three-dimensional coordinates of the matching points are calculated using triangulation to generate a dense three-dimensional point cloud. The reconstruction accuracy and matching correctness are verified by using the distance from the reconstructed laser point to the light plane. Point cloud post-processing includes filtering, stitching, and meshing. Finally, the three-dimensional model is displayed in real time through the data display module.
[0024] Example 2: In another embodiment of the invention, the system supports switching of acquisition modes. Users can select the mode via a computer software interface: High-speed mode: The camera operates in 8-bit output mode to meet the needs of high frame rate scanning; Fine-grained mode: The camera operates in 12-bit raw data output mode, used for scanning critical areas with extremely high precision requirements. The system can also automatically identify scene complexity (such as determining whether there are large areas of high reflectivity or dark areas in the image) through algorithms and automatically switch to over 8-bit mode.
[0025] See Figure 3 To verify the effectiveness of this invention, a simulation experiment was conducted. A laser stripe was simulated using a Gaussian function, with the center position set to x0 = 50.3 (a non-integer pixel). A high-precision (floating-point) grayscale distribution was first generated and mapped to 0–255 (8 bits) and 0–4095 (12 bits), respectively. The stripe center was extracted using the grayscale centroid method or Gaussian fitting method, and the error between the extracted result and the true center was calculated.
[0026] Laser extraction comparison: Simulating a laser stripe centered at 50.3 pixels. Although the extraction errors of 8-bit and 12-bit cameras are similar in this example, in real-world complex scenes, the 12-bit camera, due to its resistance to saturation and smooth curves, exhibits significantly better accuracy and robustness than the 8-bit camera. Marker point extraction comparison: Simulating a circular spot with a real center at (30.30, 30.70). The 8-bit camera fits the center at (30.043, 30.739), with an error of approximately 0.05-0.1 pixels; while the 12-bit camera fits the center at (30.111, 30.667), with the error significantly reduced to 0.01-0.02 pixels, demonstrating the significant improvement in marker point extraction accuracy brought about by higher bit depth data. See Figure 4 A circular light spot was simulated using a two-dimensional Gaussian sphere. The marked point region was obtained by threshold segmentation. The edge coordinates were extracted, and then the extracted edge points were fitted into a circle, where the true center of the circle is (30.30, 30.70). 8-bit camera circle fitting: (30.043, 30.739). 8-bit camera: Due to coarse gray quantization and unstable edge threshold segmentation, the error of the fitted circle center is about 0.05–0.1 pixels.
[0027] 12-bit camera circle fitting: (30.111, 30.667). 12-bit camera: smoother edge point distribution, the fitted circle center is closer to the true value, and the error can be reduced to 0.01–0.02 pixels.
[0028] 8-bit quantized grayscale results in poor edge extraction accuracy and large deviation in the fitted circle center. 12-bit quantization grayscale → smoother edges, fitting results closer to the true value.
[0029] The rich grayscale information makes the grayscale distribution curves of the laser stripes and marker points extremely smooth, significantly improving the accuracy and stability of the sub-pixel localization algorithm, thereby improving the overall 3D measurement accuracy. The high bit depth image has a higher signal-to-noise ratio, enabling the feature extraction algorithm to work stably in low-light or low-reflectivity scenes, effectively resisting noise interference, and outputting continuous and reliable feature centers. Due to the fundamental improvement in the quality of the input image, the subsequent processing algorithm does not need to be too complex to achieve extremely high accuracy, reducing computational complexity and improving system performance.
[0030] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.
Claims
1. A structured light laser 3D scanning system based on an 8-bit camera, characterized in that: include: At least one industrial camera with a grayscale resolution of more than 8 bits is used to acquire images of the object under test in real time; A structured light projector is used to project an coded structured light laser pattern onto the surface of a measured object at precise moments. A synchronization control module is electrically connected to the above 8-bit industrial camera and structured light projector to coordinate and control the timing of image acquisition and structured light projection. The supplementary lighting module is electrically connected to the synchronization control module and is used to receive instructions from the synchronization module to provide controllable background lighting; The data transmission module is used to receive and process image data acquired by the over 8-bit industrial camera; The data processing module is used to receive image data and perform 3D reconstruction calculations; The data display module is used to display the final 3D results to the user in real time.
2. The structured light laser 3D scanning system based on an 8-bit camera according to claim 1, characterized in that: The structured light laser 3D scanning system also includes multiple reflective markers, which are arranged on the object being measured or in the scanning environment to provide spatial reference and point cloud registration.
3. The structured light laser 3D scanning system based on an 8-bit camera according to claim 1, characterized in that: The industrial camera described above uses a 12-bit or 16-bit grayscale resolution image camera.
4. The structured light laser three-dimensional scanning method and system based on an 8-bit camera according to claim 1, characterized in that: The system includes two ultra-8-bit industrial cameras, forming a binocular vision system.
5. A structured light laser 3D scanning method based on an 8-bit camera, applied to the structured light laser 3D scanning system as described in any one of claims 1-4, characterized in that: Includes the following steps: S1. The structured light projector is triggered by the synchronous control module to project a structured light pattern, and the 8-bit industrial camera is triggered to capture one or more frames of images containing the structured light pattern and reflective markers. S2. Preprocess the raw image data acquired by the 8-bit industrial camera; S3. From the preprocessed image, candidate regions for marker points are initially identified through Blob analysis, then sub-pixel edge extraction is performed, and the sub-pixel center position of the marker points is determined through ellipse fitting or circle fitting. S4. For the sub-pixel center position, extract laser lines of a certain pixel width by row, connect them vertically to form candidate laser connected regions, and use the gray-scale centroid method to extract laser sub-pixels by row. S5. Based on the features extracted from images from different perspectives, perform feature matching and 3D reconstruction to generate a 3D point cloud.
6. The structured light laser three-dimensional scanning method based on an 8-bit camera according to claim 5, characterized in that: In step S2, the specific steps for preprocessing the original image data are as follows: First, directly read the camera's 12-bit or 16-bit raw high-bit RAW data, and use a background image with over 8-bit precision to perform image background subtraction in order to retain more shadow details; Then, adaptive contrast stretching or normalization is performed, and the grayscale distribution histogram of the entire high-bit image is analyzed to maximize the contrast of the image without losing any internal details, making faint stripes clear, while suppressing overly bright stripes without saturating them.
7. The structured light laser three-dimensional scanning method based on an ultra-8-bit camera according to claim 5, characterized in that: In step S3, at least one of the gray-scale centroid method, Gaussian fitting method, or Steger algorithm is used to extract the center line of the structured light stripes at the subpixel level.
8. The structured light laser three-dimensional scanning method based on an ultra-8-bit camera according to claim 5, characterized in that: For binocular vision systems, the principle of symmetry geometry is used to match the corresponding laser stripe points and marker points in the left and right camera images, and their three-dimensional coordinates are calculated by triangulation.
9. The structured light laser three-dimensional scanning method based on an ultra-8-bit camera according to claim 6, characterized in that: The structured light laser 3D scanning method supports switching of working modes, including: In the full-scene 8-bit mode, image acquisition is performed using the 8-bit mode in all scanning scenarios. In selective 8-bit mode, switch to 8-bit mode for data acquisition in special scenarios or for specific areas where fine scanning is required; In scenarios where scanning speed is paramount, use the normal 8-bit mode for data acquisition.
10. The structured light laser three-dimensional scanning method based on an ultra-8-bit camera according to claim 9, characterized in that: The switching of the working mode is achieved through computer-side interaction, device-side interaction, or automatic system scene recognition.