Anti-interference structured light measurement method based on full-high-frequency binary projection
By combining full-frequency binary projection and illumination separation technology with Gray code and four-step edge-moving method, the problem of insufficient decoding accuracy of traditional structured light measurement on complex reflective surfaces is solved, and high-precision and efficient 3D reconstruction is achieved.
Patent Information
- Application Number
- CN202511740564.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-25
- Publication Date
- 2026-03-10
AI Technical Summary
Traditional structured light measurement methods are easily interfered with in complex reflective surface environments, resulting in insufficient decoding accuracy and difficulty in meeting the requirements for high precision and real-time performance.
A high-frequency coding pattern is designed by combining full-frequency binary projection with Gray code and a four-step edge-moving method. Robust acquisition and high-precision decoding of high-frequency structured light signals are achieved through illumination separation and complementary projection strategies.
It significantly improves the accuracy and robustness of 3D reconstruction under complex lighting conditions, reduces measurement time and computational load, and is suitable for high-precision applications such as industrial inspection and precision manufacturing.
Smart Images
Figure CN121639604A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of optical three-dimensional measurement and computer vision, and particularly relates to an anti-interference structured light measurement method based on full high-frequency binary projection. The method combines structured light coding, binocular stereo vision and high-frequency light separation technology, and aims at the problems of traditional structured light measurement being easily disturbed and decoding accuracy being insufficient in complex light reflection surface environment. Through designing a full high-frequency binary coding pattern, an improved four-step edge moving method, and a light separation and complementary projection strategy, the robust collection and high-precision decoding of high-frequency structured light signals in light reflection scenes are realized. The present application can be widely applied in the fields of industrial detection, cultural heritage digitization, virtual reality modeling and precision manufacturing, and belongs to a high-precision and anti-interference three-dimensional reconstruction measurement technology. BACKGROUND
[0002] In recent years, with the continuous upgrading of China's manufacturing industry and the in-depth promotion of the "intelligent manufacturing" strategy, industrial production is accelerating towards high precision and automation. The geometric size control and appearance quality detection of products have become the key link affecting the performance, assembly precision and production efficiency of products. Whether in high-end equipment manufacturing, precision mold processing, automobile part detection, or in the fields of electronic packaging, biomedical devices and cultural heritage digitization, fast, non-contact and high-precision three-dimensional appearance measurement of the measured object has important engineering significance. Three-dimensional reconstruction technology obtains the spatial geometric information of the object surface, realizes the conversion from entity to digital model, and provides key data support for subsequent size analysis, assembly positioning, defect detection and reverse design, and has become an indispensable basic technology in the intelligent manufacturing system.
[0003] Among many three-dimensional measurement technologies, structured light measurement has become one of the most widely used active optical measurement methods due to its high resolution, high precision and real-time performance. The method projects an encoded light pattern onto the surface of the measured object through a projector, and captures the reflected image using a camera. According to the stereo vision and phase decoding algorithm, the depth information of the object surface is recovered. Compared with contact measurement, structured light system realizes efficient and non-destructive three-dimensional data acquisition, and can adapt to complex workpieces and online detection scenes. However, in actual industrial environment, the surface of the measured object often has mirror reflection or non-uniform reflection characteristics, such as metal, glass and bright plastic materials. These high light reflection characteristics can cause overexposure, underexposure and other problems in the imaging process of structured light pattern, resulting in loss of encoded stripe information, which seriously affects the decoding accuracy and measurement integrity. Such complex lighting conditions with high brightness and dark areas are called high dynamic range (High Dynamic Range, HDR) scenes. Traditional structured light systems are difficult to ensure the stability and precision of three-dimensional reconstruction under such conditions.
[0004] To deal with the problem of anti-interference, the prior art mainly improves from two aspects of optical hardware and image processing. The hardware optimization scheme such as polarization filter method, phase measurement deflection method, etc. can weaken the mirror reflection or improve the lighting condition to a certain extent, but the system is complex, the cost is high, and it is subject to specific materials and environment; the image-based processing method such as multiple exposure, adaptive projection, etc. can improve the imaging quality by fusing different exposure images or adjusting the light intensity, but it needs multiple shooting or high complexity calculation, which is difficult to meet the real-time measurement demand. Although the deep learning method in recent years has made progress in anti-interference area repair and phase prediction, its training depends on a large number of labeled data, and the generalization ability is limited, which is still difficult to deal with complex industrial scenes.
[0005] In view of the above problems, the present application provides an anti-interference structured light measurement method based on full high-frequency binary projection. The method is based on the traditional structured light three-dimensional reconstruction, and a full high-frequency coding strategy combining binary Gray code stripes and high-frequency checkerboard patterns is designed, which can suppress low-frequency components from the coding level by means of combinatorial mathematics principle, and avoid the interference of anti-reflection on decoding. At the same time, through the illumination separation characteristics of high-frequency coded pattern, the direct light and global light are distinguished, and the time domain binary algorithm robust to anti-reflection scene is proposed combined with complementary projection, so as to obtain high-precision decoding result in complex lighting environment. Further, by using complementary line shift checkerboard projection and edge information, sub-pixel level edge positioning is realized, and the coding accuracy and detail restoration ability are improved.
[0006] In summary, the present application combines algorithm-level coding optimization with illumination separation mechanism, which significantly improves the three-dimensional reconstruction accuracy and robustness of anti-reflection surface and high dynamic range scene without increasing the complexity of the system hardware, and has the advantages of simple structure, high efficiency and strong adaptability, which is suitable for high-precision measurement fields such as industrial detection, precision manufacturing and virtual modeling. SUMMARY
[0007] The present application relates to an anti-interference structured light measurement method based on full high-frequency binary projection, and a method for separating direct light and global light to improve the robustness of structured light decoding in high dynamic range scene. The method combines structured light coding, binocular stereo vision technology and high-frequency light separation technology, and proposes an innovative scheme based on full high-frequency binary coding pattern, improved four-step edge moving method, and light separation and complementary projection strategy, aiming at the problem that traditional structured light measurement is easily disturbed in complex anti-reflection surface environment and the decoding accuracy is insufficient. Through these technical means, the present application can realize stable acquisition and high-precision decoding of high-frequency structured light signal in anti-reflection scene, and significantly improve the anti-interference ability. The present application is suitable for multiple fields such as industrial detection, digitalization of cultural relics, virtual reality modeling and precision manufacturing, and belongs to a kind of three-dimensional reconstruction measurement technology with high precision and strong anti-interference.
[0008] To solve the above technical problems, the present application adopts the following technical solutions:
[0009] A full high-frequency binary projection-based anti-interference structured light measurement method, characterized by comprising the following main steps:
[0010] Step 1: Binocular camera calibration of the measurement system: a binocular camera is used to form an imaging system, and internal and external parameter calibration is performed to establish the mapping relationship between the binocular camera and the projector projection points. A unified world coordinate system is established through the calibration process to ensure consistency of coordinate transformation in subsequent three-dimensional reconstruction calculations. A checkerboard or coded target is used for imaging and matching during the calibration process.
[0011] Step 2: Full high-frequency pattern generation: based on the pattern of Gray code combined with the four-step edge movement method, and performing XOR logical operation encryption between each pattern and an 8-pixel wide checkerboard pattern, a group of stripe patterns changing from low frequency to high frequency is converted into a full high-frequency coded pattern for projection.
[0012] Step 3: Coded pattern projection and decoding: the encrypted pattern and the checkerboard pattern are used for actual projection. After binarizing the captured image, the high-frequency pattern and the checkerboard pattern are subjected to XOR operation again, and the original stripe pattern is restored, ensuring that all patterns are composed of 4-pixel wide squares, which has better robustness to high light interference. The improved four-step edge movement method is used to replace the line movement method to improve the edge preservation property.
[0013] Step 4: Edge movement method optimization based on light separation: for high-frequency checkerboard pattern threshold segmentation under reflection interference, a series of high-frequency projection patterns are used to quantitatively decompose the collected light into global light and direct light Two parts, according to and Determine the threshold segmentation direction, and combine the light intensity variation law in complementary projection to obtain a threshold segmentation result robust to reflection interference.
[0014] Step 5: Phase unwrapping and binocular stereo matching: for the optimized wrapped phase obtained in step four, finally perform sub-pixel level phase matching according to the unwrapped phases of the left and right cameras, and perform triangulation calculation with the calibration parameters to obtain the final point cloud information.
[0015] Further, in step 1, the imaging device in the system is a binocular measurement system composed of two cameras, and a DLP programmable projection light machine.
[0016] Further, in step 2, the design scheme of the full high-frequency pattern is based on a set of black and white stripe patterns with Gray code and four-step edge movement method, and each pattern is subjected to XOR logical operation with an 8-pixel-wide black and white chessboard pattern, which is called encryption, and after encryption, a set of stripe patterns changing from low frequency to high frequency is converted into full high-frequency coding pattern for projection.
[0017] Further, the full high-frequency coding pattern designed by the method has the characteristic that the minimum unit of all patterns is composed of 4-pixel-wide squares, which has better robustness to high light interference.
[0018] Further, in step 3, the set of encrypted patterns and the chessboard pattern are used for actual projection, and after the captured picture is binarized, the high-frequency pattern and the chessboard pattern are subjected to XOR operation again, so that the original stripe pattern can be restored.
[0019] Further, in step 3, the improved four-step edge movement method is used instead of the line movement method, which improves the edge-preserving property compared with the line movement method.
[0020] Further, in step 4, the improved four-step edge movement method of the high-frequency chessboard uses the edge information of the black and white block boundary for coding, and through projection of two complementary patterns of the light wall, the robust positioning of the edge information is realized according to the symmetry of the reflected light interference in the collected double-frame complementary projection.
[0021] Further, the symmetry of the reflected light interference in the complementary image is used to perform difference operation on the collected light, so as to eliminate the influence of the reflected light on edge detection.
[0022] The present application has the following beneficial effects due to the adoption of the above technical scheme:
[0023] 1. The present application adopts full high-frequency binary coding pattern, and designs a set of structured light stripe patterns with higher frequency and anti-interference characteristics by combining Gray code with four-step edge movement method. Compared with the traditional low-frequency binary coding pattern, the full high-frequency coding pattern of the present application can effectively reduce the interference of reflected light through the encryption process. In the surface of high-reflective objects, especially the reflection of metal, glass and other materials often causes overexposure or underexposure of the image, affecting the measurement accuracy. The traditional method cannot handle the interference in such high dynamic range scenes, which is easy to cause decoding error or information loss. The high-frequency coding strategy and improved projection method of the present application effectively improve the anti-interference ability of the system in complex lighting conditions. Through this optimization scheme, stable three-dimensional reconstruction effect can still be maintained in high-reflective environment, greatly improving the decoding accuracy and reliability of the result.
[0024] 2、The application replaces the traditional line shift method with an improved four-step edge shift method, significantly reducing the number of projected patterns required for a single measurement compared to methods such as multiple exposure that require repeated projection of multiple patterns. The improved scheme of the application optimizes the projection process by reducing the number of projected patterns, maintains the edge information of the patterns, and avoids the problem of edge loss in traditional methods. Reducing the number of patterns not only reduces the burden on the projection equipment, but also shortens the measurement time and improves the efficiency of the system. In addition, reducing the number of projected patterns reduces the amount of data acquisition and processing, ensuring that the measurement process is more efficient, thereby significantly improving the overall system's computing efficiency, enabling more accurate three-dimensional reconstruction in a shorter time.
[0025] 3、In the application, the light separation technology is innovatively applied to the design and decoding process of high-frequency coded patterns. Traditional structured light measurement methods often cannot effectively distinguish between direct light and global light in complex lighting environments, especially in scenes with strong reflection, resulting in a decline in image quality and affecting decoding accuracy. To address this problem, the application uses a high-frequency coded pattern-based light separation strategy to effectively distinguish between direct light and global light, thereby avoiding the interference of overexposure and underexposure problems on the decoding process. In addition, the application introduces a sub-pixel level edge positioning algorithm to accurately capture the edge information of the reflected image, further optimizing the decoding process. In a strong reflection interference environment, it can more accurately restore edge features to ensure the accuracy and integrity of the decoding. Through the application of this technology, the application significantly improves the robustness and accuracy of the measurement results in complex lighting conditions, especially in scenes with strong reflection.
[0026] 4、The application ensures high-precision three-dimensional reconstruction while avoiding the problem of excessive computational complexity caused by increasing the number of projected patterns or complex algorithms in traditional methods. Traditional structured light three-dimensional measurement methods usually require multiple measurements and multiple image processing to overcome reflection interference and occlusion problems, which often requires a large amount of computing resources and time, limiting real-time performance and efficiency. The application realizes higher precision three-dimensional reconstruction without increasing additional computational load through the encryption design of full high-frequency binary projection patterns and high-frequency decoding methods, suitable for industrial production lines and other high-precision application scenarios. BRIEF DESCRIPTION OF DRAWINGS
[0027] The application will be further described below in conjunction with the accompanying drawings:
[0028] Figure 1 The application is an anti-interference structured light measurement method based on full high-frequency binary projection.
[0029] Figure 2 The application is a full high-frequency binary projection pattern encryption process.
[0030] Figure 3 This is a schematic diagram of the full high-frequency binary projection pattern decoding process of the present invention.
[0031] Figure 4 This invention is compared with other high-frequency structured light methods for suppressing reflective surfaces. Detailed Implementation
[0032] The present invention will now be described in further detail with reference to the accompanying drawings.
[0033] Figure 1 This diagram illustrates the anti-interference structured light measurement method based on full-frequency binary projection according to the present invention. This method is based on a pattern derived from Gray code combined with a four-step edge-moving method. Each pattern is encrypted by performing an XOR operation with an 8-pixel-wide checkerboard pattern, transforming a set of stripe patterns varying from low to high frequency into a full-frequency coded pattern for projection. During actual projection, this encrypted pattern and the checkerboard pattern are used. After binarizing the captured image, the high-frequency pattern and the checkerboard pattern are XORed again to restore the original stripe pattern. This method ensures that the smallest unit of all graphics consists of 4-pixel-wide squares, providing better robustness against specular interference. This method uses an improved four-step edge-moving method instead of a line-shifting method, reducing the number of projected patterns and improving edge preservation compared to the line-shifting method.
[0034] Figure 2 This diagram illustrates the encryption process of the full high-frequency binary projection pattern of the present invention. The proposed high-frequency binary projection pattern is obtained by performing a Boolean XOR operation on the traditional low-frequency binary pattern sequentially with the high-frequency reference pattern (the pattern with the green border in the diagram).
[0035] Figure 3 This is a schematic diagram of the full high-frequency binary projection pattern decoding process of the present invention. The encrypted and captured pattern sequence is first binarized and segmented, and then sequentially XORed with the high-frequency reference pattern. The decoded pattern sequence is then restored from the high-frequency pattern to a low-frequency binary pattern using the traditional binary encoding method.
[0036] Figure 4 This paper compares the 3D reconstruction results of this invention with other structured light methods. Compared to existing methods, the method proposed in this invention, when used to photograph a reflective metallic object, demonstrates a greater suppression effect on highlight areas.
[0037] The measurement method relates to a typical binocular structured light system. The system mainly comprises a digital light processing (DLP) projector and two industrial cameras. The projector is used to project the coded pattern to the measured object, and the two cameras synchronously collect the pattern images modulated by the object surface. Before implementing the measurement method, the system calibration needs to be completed first: the internal and external parameters of the binocular camera are obtained through a mature calibration method (such as using a checkerboard or a coded target), and the mapping relationship between the projector and the camera coordinate systems is established, laying a foundation for subsequent three-dimensional reconstruction.
[0038] On this basis, as Figure 1 indicated, the technical scheme of the present application comprises the following steps:
[0039] 1. Step 1: full high-frequency binary projection pattern design, the workflow is as follows:
[0040] In order to obtain a full high-frequency coded pattern, the core of this method is to decompose the traditional single low-frequency binary pattern into two high-frequency patterns through Boolean exclusive or, as Figure 2 indicated. Specifically, the traditional low-frequency stripe pattern such as Gray code can be represented as , which can be regarded as being composed of two high-frequency patterns and through the exclusive or operation of each pixel, and the above decomposition process can be represented as: .
[0041] Reference checkerboard pattern. Select the reference checkerboard pattern In the present application, for a projection surface with a length of 1280*720, a black and white checkerboard pattern with a side length of 8 pixels is used as the reference checkerboard pattern.
[0042] Full high-frequency binary projection pattern encryption process, as Figure 2 indicated. The traditional low-frequency binary pattern is sequentially subjected to Boolean exclusive or operation with the designed high-frequency reference pattern to obtain the proposed high-frequency binary projection pattern . Wherein M represents the number of traditional binary coded patterns to be decomposed, in the present application, a 10-bit Gray code pattern is used in combination with four edge moving stripe patterns.
[0043] 2. Step 2: coded pattern acquisition stage. The projector projects a series of high-frequency binary projection patterns obtained through step 1 and high-frequency reference patterns to the surface of the measured object.The high-frequency pattern has anti-glare characteristics and carries the encoding information of Gray code and four-step edge movement method, which can ensure that each spatial position point has unique encoding characteristics. At the same time, the binocular camera and the projection process are triggered to take pictures synchronously, and the direct reflection image sequence from the object surface and the indirect image sequence formed by the reflection of the half-sphere mirror are captured in real time. By constructing the high-resolution phase distribution and the corresponding relationship of the stripes, reliable data support is provided for subsequent three-dimensional reconstruction.
[0044] 3. Step 3: Full high-frequency binary projection pattern decoding process, as shown in Figure 3 . The specific steps are as follows:
[0045] 3.1 Binaryzation segmentation is performed on the images collected by all cameras, and the pattern sequence is denoted as and .
[0046] 3.2 Perform Boolean XOR operation between , and the high-frequency reference pattern in turn. The decoded pattern sequence is restored from the high-frequency pattern to the low-frequency binary pattern using the traditional binary encoding method.
[0047] 4. Step 4: Decode a series of binary patterns obtained in step 3 to obtain the unwrapped phase. The specific steps are as follows:
[0048] 4.1 Direct and indirect illumination calculation. The present application uses a series of high-frequency projection patterns in the time domain to quantitatively decompose the collected illumination into global illumination and direct illumination , which can be represented as:
[0049]
[0050] In the above formula, is obtained by projecting a full black pattern and a full white pattern and calculating the ratio of the two pattern acquisition results, and can be estimated by collecting high-frequency checkerboard images and line-shifted checkerboard images.
[0051] 4.2 If a pixel is directly illuminated by the projector, the collected light intensity range should be ; if a pixel is not directly illuminated by the projector, the collected light intensity range should be . Let the collected image to be threshold segmented be , and the complementary collected image be .
[0052] 4.3 If , the range of collected light intensity in the complementary projection does not overlap, and the encoded information can be directly obtained from and ; , the global illumination caused by the reflection makes the range of light intensity of and overlap, at this time, only when and , or symmetrically, and , the range of collected light intensity in the complementary projection does not overlap, thereby realizing the determination of the encoded information interfered by the reflection according to and .
[0053] 4.4 After the encoded information of the reflection interference area is determined, the obtained series of binary patterns are decoded to obtain the wrapped phase and the unwrapped phase, and the decoding principle can refer to the conventional binary method, which will not be described here. This method can accurately recover the area where the information is missing due to the self-occlusion or high reflection of the object in the conventional structured light measurement, thereby obtaining more complete and continuous three-dimensional topographic data of the object.
[0054] Although the present application has been disclosed with the preferred embodiments as above, it is not intended to limit the present application. Those skilled in the art having ordinary knowledge can make various modifications and decorations without departing from the spirit and scope of the present application. Therefore, the protection scope of the present application shall be subject to the definition of the claims.
Claims
1. A full high frequency binary projection based anti-interference structured light measurement method, characterized in that The method comprises the following main steps: Step 1: Binocular camera calibration of the system: a binocular camera is used to form an imaging system, internal and external parameters are calibrated to establish the mapping relationship between the binocular camera and the projection points of the projector, and a unified world coordinate system is established through the calibration process, so that the coordinate transformation in the subsequent three-dimensional reconstruction calculation is consistent, and a chessboard or a coded target is used for imaging and matching in the calibration process; Step 2: Full high-frequency pattern generation: based on the pattern of Gray code combined with four-step edge movement method, and each pattern is encrypted by performing XOR logical operation with an 8-pixel-wide chessboard pattern, a group of high-frequency to low-frequency changing stripe patterns is converted into a full high-frequency coded pattern for projection; Step 3: Coded pattern projection and decoding: the encrypted pattern and the chessboard pattern are used for actual projection, the captured image is binarized, and the high-frequency pattern and the chessboard pattern are subjected to XOR operation again, so that the original stripe pattern is restored, and the minimum unit of all patterns is composed of 4-pixel-wide squares, which has better robustness to high light interference, and the improved four-step edge movement method is used to replace the line movement method to improve the edge preservation characteristic; Step 4: Edge moving method based on light separation optimization: high-frequency chessboard image threshold segmentation method under the interference of reflection light, using a series of high-frequency projection patterns in time domain, the collected light is quantitatively decomposed into global light and direct light Two parts, according to and Determine the direction of threshold segmentation, and combine the light intensity change rule in the complementary projection to obtain a threshold segmentation result robust to reflection light interference; Step 5: Phase unwrapping and binocular stereo matching: the wrapped phase obtained in step 4 is finally subjected to sub-pixel level phase matching according to the unwrapped phases of the left and right cameras, and triangular measurement calculation is performed with the calibration parameters to obtain the final point cloud information.
2. The anti-interference structured light measurement method based on full high frequency binary projection according to claim 1, characterized in that: In step 1, the imaging device in the system is a binocular measurement system composed of two cameras, and a DLP programmable projection light machine.
3. The anti-interference structured light measurement method based on full high frequency binary projection according to claim 1, characterized in that: In step 2, the design scheme of the full high-frequency pattern is based on a group of black and white stripe patterns of Gray code combined with four-step edge movement method, and each pattern is subjected to XOR logical operation with an 8-pixel-wide black and white chessboard pattern, which is called encryption.
4. The anti-interference structured light measurement method based on full high frequency binary projection according to claim 1, characterized in that: The encrypted full high-frequency coded pattern designed by the method has the pattern characteristic that the minimum unit of all patterns is composed of 4-pixel-wide squares, which has better robustness to high light interference.
5. The anti-interference structured light measurement method based on full high frequency binary projection according to claim 1, characterized in that: In step 3, the encrypted pattern and the chessboard pattern are used for actual projection, the captured image is binarized, and the high-frequency pattern and the chessboard pattern are subjected to XOR operation again, so that the original stripe pattern is restored.
6. The anti-interference structured light measurement method based on full high frequency binary projection according to claim 1, characterized in that: In step 3, the improved four-step edge movement method is used instead of the line movement method, which improves the edge preservation characteristic compared with the line movement method.
7. The anti-interference structured light measurement method based on full high frequency binary projection according to claim 1, characterized in that: In step 4, the improved four-step edge movement method of the high-frequency chessboard uses the edge information of black and white blocks to encode, projects two complementary patterns of light walls, and realizes robust positioning of edge information according to the symmetry of the reflected light interference in the collected double frames of complementary projection.
8. The anti-interference structured light measurement method based on full high frequency binary projection according to claim 7, characterized in that: The symmetry of the reflected light interference in the complementary images is used to perform difference operation on the collected light, so as to eliminate the influence of the reflected light on edge detection.