A slope-mirror-based correlation light measuring device and control method

Abstract

The application relates to a related light measurement device based on a bevel mirror and a control method, solves the problem that existing three-dimensional detection technology is insufficient in comprehensive adaptability and cannot meet the requirements of stable, accurate and data cooperation for precise detection, and comprises a laser emitting head, double bevel mirrors, a light path receiving module, a zoom camera and a circuit processing unit. The first bevel mirror is provided with a through hole for laser direct irradiation of a target; the light path receiving module converts reflected laser into an electric signal; the second bevel mirror reflects target and spot light to the camera to obtain visual information; and the circuit processing unit controls each component, integrates laser and visual information, and realizes three-dimensional detection of the target. The application has the following effects: through the light path cooperative design of the double bevel mirrors, accurate synchronous collection of laser ranging signals and visual profile / spot information is realized, so that high-precision three-dimensional detection of the distance, flatness and section difference of the target object is efficiently completed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of precision optical engineering, and in particular to a related light measurement device based on a bevel mirror and a control method. BACKGROUND

[0002] In the field of industrial precision manufacturing and part detection, the detection of three-dimensional parameters such as the distance, flatness, and step of the target object is a core requirement for ensuring product precision. With the upgrading of manufacturing processes, the industry increasingly needs to simultaneously obtain multi-dimensional information such as laser geometric data, visual profile, and spot position to achieve efficient and accurate three-dimensional analysis.

[0003] Current mainstream three-dimensional detection technologies can be divided into three categories: laser-only measurement technology (such as triangulation and reflection ranging), which obtains depth parameters through laser signals; pure vision technology, which relies on image acquisition and algorithm to extract object profiles; and laser-vision combined technology, which simultaneously equips laser modules and cameras to try to integrate the two types of information.

[0004] The existing technology has certain limitations: laser-only measurement is easily disturbed by the environment and has low detection efficiency in a large range; pure vision technology cannot directly obtain depth parameters and is limited in precision; and the laser-vision combined scheme has low matching degree of the two types of data due to insufficient light path coordination, and the integration effect is not good, making it difficult to fully meet the actual needs of precision detection. SUMMARY

[0005] In order to achieve precise and synchronous acquisition of laser ranging signals and visual profile / spot information through the light path coordination design of double bevel mirrors, and efficiently complete the high-precision three-dimensional detection of the distance, flatness, and step of the target object, the present application provides a related light measurement device based on a bevel mirror and a control method.

[0006] In a first aspect, the present application provides a related light measurement device based on a bevel mirror, which adopts the following technical solution:

[0007] A related light measurement device based on a bevel mirror, comprising:

[0008] A laser emission head for emitting a laser beam propagating along a predetermined direction and directed at a target object;

[0009] A first bevel mirror arranged on the laser propagation path between the laser emission head and the target object, the bevel of which is a mirror glass surface with a through hole for the laser beam to pass through, the axis of the through hole being collinear with the light emission direction of the laser emission head, so that the laser beam passes through the through hole and directly hits the target object;

[0010] The light path receiving module is arranged adjacent to the laser emitting head, and both are located on the side of the first inclined mirror away from the target object, and is used for receiving the laser beam returned along the original path and passing through the through hole after being reflected by the target object, and converting the reflected laser beam into an electric signal;

[0011] The second inclined mirror is arranged at a preset included angle with the inclined surface of the first inclined mirror, the mirror surface of the second inclined mirror faces the side where the first inclined mirror is located, and the mirror surface is a mirror glass surface, configured to reflect the light from the target object and the light spot formed by the laser beam on the target object;

[0012] The zoom camera has a lens facing the mirror surface of the second inclined mirror, and is used for receiving the light reflected by the second inclined mirror to obtain the visual profile of the target object and the spatial position information of the light spot formed by the laser beam on the target object;

[0013] The circuit processing unit is electrically connected with the laser emitting head, the light path receiving module and the zoom camera respectively, and is used for controlling the emission timing of the laser emitting head, receiving and processing the electric signal output by the light path receiving module to calculate the distance, flatness and step difference of the target object, controlling the zoom camera to work and process the visual information obtained by the zoom camera, and realizing the three-dimensional detection of the target object by integrating the laser signal and the visual information.

[0014] By adopting the above technical scheme, the device realizes the division of labor and cooperation of the two inclined mirrors, the first inclined mirror is used for directly emitting laser to the target, and the second inclined mirror is used for reflecting the visual light, the laser and the visual component are arranged on the same side, the structure is compact and space-saving; the laser ranging and the visual profile information are fused by the circuit processing unit, the distance, the flatness and the step difference of the target can be accurately calculated; the three-dimensional detection of the target is realized, the space utilization rate and the detection accuracy are considered, and the device is suitable for the measurement requirements in multiple scenes such as industry and precision parts.

[0015] In a second aspect, the application provides a control method of a related light measuring device based on an inclined mirror, which adopts the following technical scheme:

[0016] A control method of a related light measuring device based on an inclined mirror, comprising:

[0017] The laser emitting head is controlled to emit a laser beam according to a preset timing, so that the laser beam directly emits to the target object along the through hole axis of the first inclined mirror, and the zoom camera is controlled to collect the target region image reflected by the second inclined mirror, and the initial pixel coordinates of the laser light spot and the initial outline features of the target in the image are extracted to form initial visual data;

[0018] The laser signal reflected by the target and passing through the through hole is collected by the light path receiving module, the laser ranging value is calculated after the laser signal is converted into an electric signal, the mapping relationship between the laser ranging value and the visual pixel coordinates is established by combining the initial pixel coordinates of the light spot in the initial visual data, and the initial outline features are taken as a spatial reference;

[0019] Based on the mapping relationship, the laser signal collected by the light path receiving module in real time is converted into an electrical signal and calculated into target real-time distance data, and a dynamic image reflected by the second inclined mirror is obtained synchronously through the zoom camera, and the real-time pixel coordinates of the light spot and the real-time contour feature of the target are identified;

[0020] The preset data fusion rule is called, the real-time distance data, the real-time pixel coordinates of the light spot and the real-time contour feature are associated, and the target flatness and step difference are calculated based on the initial contour feature as a reference;

[0021] The real-time distance data, the corrected flatness and step difference information are integrated, and the detection result containing the three-dimensional coordinates, the flatness and the step difference parameters of the target object is output.

[0022] By adopting the above technical scheme, the method first synchronously collects initial visual data and laser ranging value, quickly establishes the mapping relationship between the two, and lays a foundation for accurate association; then the distance data and visual features are obtained in real time and synchronously, so as to guarantee the timeliness of detection; the flatness and step difference are accurately calculated through data fusion, so as to improve the parameter accuracy; finally, the three-dimensional detection result is integrated and output, the process is coherent and efficient, and the target three-dimensional detection can be accurately completed, which is suitable for precision measurement and other scenes. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 is a whole use schematic view of a related light measurement device based on an inclined mirror according to an embodiment of the present application.

[0024] Figure 2 is a flowchart of a control method of a related light measurement device based on an inclined mirror according to an embodiment of the present application.

[0025] In the figure, 1 is a laser emitting head; 2 is a first inclined mirror; 3 is a light path receiving module; 4 is a second inclined mirror; 5 is a zoom camera; and 6 is a circuit processing unit. DETAILED DESCRIPTION

[0026] The present application will be further described in detail below with reference to the accompanying drawings.

[0027] Reference Figure 1 A related light measurement device based on an inclined mirror is disclosed in the present application, which mainly comprises a laser emitting head 1, a first inclined mirror 2, a light path receiving module 3, a second inclined mirror 4, a zoom camera 5 and a circuit processing unit 6, and the functions and cooperation relationships of the components are as follows:

[0028] The laser emission head 1 can emit a laser beam along a preset direction, and the propagation path of the laser beam is directed to the target object; the first inclined mirror 2 is arranged on the laser propagation path between the laser emission head 1 and the target object, the inclined surface of which is made of mirror glass, and a through hole for the laser beam to pass through is formed on the inclined surface, the inner wall of the through hole is provided with a black light-absorbing coating to reduce stray light interference, and the diameter of the through hole is 0.5-1mm larger than the spot diameter of the laser beam, which not only ensures the smooth passage of the laser beam, but also avoids excessive ambient light from entering; the axis of the through hole is collinear with the light emission direction of the laser emission head 1, which ensures that the laser beam is directly incident on the target object.

[0029] The light path receiving module 3 is arranged adjacent to the laser emission head 1, and both are located on the side of the first inclined mirror 2 away from the target object, for receiving the laser beam returned along the original path after being reflected by the target object and passing through the through hole, and converting the reflected laser beam into an electrical signal; the second inclined mirror 4 is arranged at a preset angle with the inclined surface of the first inclined mirror 2, the mirror surface faces the side where the first inclined mirror 2 is located and is made of mirror glass, and can reflect the light from the target object and the light forming a spot on the surface of the target object.

[0030] The lens of the zoom camera 5 is directly opposite the mirror surface of the second inclined mirror 4, which can receive the reflected light and further obtain the visual profile of the target object and the spatial position information of the spot; the circuit processing unit 6 is electrically connected with the laser emission head 1, the light path receiving module 3 and the zoom camera 5 respectively, and can control the laser emission timing and the working state of the zoom camera 5, process the electrical signal of the light path receiving module 3 to calculate the target distance, flatness and step difference, and finally realize the three-dimensional detection of the target object by integrating the laser signal and the visual information.

[0031] Referring to Figure 2 A control method of a related light measurement device based on an inclined mirror, comprising:

[0032] In step S100, the laser emission head 1 emits a laser beam according to a preset timing, so that the laser beam is directly incident on the target object along the axis of the through hole of the first inclined mirror 2, and the zoom camera 5 collects the target area image reflected by the second inclined mirror 4 at the same time, and extracts the initial pixel coordinates of the laser spot and the initial outline features of the target in the image to form initial visual data.

[0033] The preset timing refers to the time sequence of the laser emission head 1 emitting the laser beam, which is usually controlled by the circuit processing unit 6 to ensure that the laser emission and subsequent signal collection operations are performed at the appropriate time point. For example, if the preset timing is to emit a laser pulse every 10ms, the circuit processing unit 6 will send a trigger signal to the laser emission head 1 according to this time interval.

[0034] Initial visual data: including the initial pixel coordinates of the laser spot and the initial contour features of the target, which are collected and processed by the zoom camera 5, used for subsequent measurement and calculation.

[0035] The necessary processes are described as follows:

[0036] 1. Preheating and calibration of the laser emitter head 1: Before starting the formal measurement, the laser emitter head 1 needs to be preheated to ensure the stability of the laser beam parameters emitted. At the same time, through the standard reflective target with known distance, the emission parameters of the laser emitter head 1 are adjusted to make it accurately emit laser beams.

[0037] 2. Alignment and verification of the first inclined mirror 2: After the device is assembled, the through-hole axis of the first inclined mirror 2 needs to be aligned and verified. A laser beam can be emitted using a laser collimator, and the alignment condition can be judged by observing whether the laser can smoothly pass through the through-hole and directly shoot at the target object. If deviation is found, the position and angle of the inclined mirror need to be fine-tuned until the laser can accurately pass through the through-hole.

[0038] 3. Parameter setting and image acquisition of the zoom camera 5: According to the size and distance of the target object, the focal length of the zoom camera 5 is adjusted to clearly collect the image of the target area. When collecting images, the exposure time, gain and other parameter settings of the camera need to be reasonable to obtain high-quality images. For example, for darker environments, the exposure time needs to be increased or the gain needs to be increased; for brighter environments, the exposure time needs to be reduced or the gain needs to be reduced.

[0039] 4. Image processing and initial visual data extraction: The collected images usually contain a lot of noise and interference information, which need to be preprocessed through image processing algorithms. First, a filtering algorithm (such as Gaussian filtering) can be used to remove noise in the image. Then, the edge detection algorithm is used to extract the contour features of the target object, and the position of the laser spot is located through threshold segmentation algorithm or region growing algorithm, etc. and the initial pixel coordinates are calculated. For example, assuming that the contour of the target object in the collected image is relatively clear, the contour can be extracted through Canny algorithm, and then the pixel points with brightness higher than a certain threshold in the contour area are searched to determine the position of the light spot.

[0040] Assuming the target object is a flat metal plate with a high surface reflectivity, the laser spot formed on the target object surface has a spot diameter of about 2mm. In step S100, first, the laser emitter head 1 is preheated and calibrated to enable it to stably emit a laser beam with a wavelength of 650nm and a power of 5mW. Then, the through-hole axis of the first inclined mirror 2 is aligned and verified by the laser collimator to ensure that the laser beam can accurately pass through the through-hole and directly hit the target object. Next, according to the distance of the target object, the focal length of the zoom camera 5 is adjusted to 50mm, and the exposure time is set to 1 / 1000s and the gain to 1.5 to capture a clear image of the target area. After the image is captured, the image is denoised using a Gaussian filter algorithm, and then the contour features of the target object are extracted using a Canny edge detection algorithm. In the contour area, the position of the laser spot is located using a threshold segmentation algorithm (set threshold value to 128), and the initial pixel coordinates of the spot are calculated to be (300, 200), and the initial contour features of the target object are extracted as a rectangular area with the upper left corner coordinates of (100, 100) and the lower right corner coordinates of (500, 400). These initial visual data will serve as the basis for subsequent measurement and calculation.

[0041] In step S200, the laser signal reflected by the target and passing through the through-hole is collected by the optical path receiving module 3, and after being converted into an electrical signal, the laser ranging value is calculated, and combined with the initial visual data of the initial pixel coordinates of the spot, a mapping relationship between the laser ranging value and the visual pixel coordinates is established, and the initial contour features are used as the spatial reference.

[0042] Wherein, the laser ranging value: the distance value between the target object and the measuring device calculated after processing the laser signal received by the optical path receiving module 3. It is usually measured based on the principles of Time-of-Flight (ToF) or phase difference. Mapping relationship: the corresponding relationship between the laser ranging value and the visual pixel coordinates. By establishing this relationship, the distance data obtained by laser ranging can be associated with the pixel position in the image, providing a basis for subsequent three-dimensional reconstruction and measurement. Initial contour features: the contour features of the target object extracted in the initial visual data, used as a spatial reference to determine the position and shape of the target object in the image. It is usually obtained through edge detection and other image processing algorithms.

[0043] The necessary procedures are described as follows:

[0044] 1. Calibration and calibration of the optical path receiving module 3: Before formal measurement, the optical path receiving module 3 needs to be calibrated and calibrated to ensure that it can accurately receive and convert the laser signal. It can be calibrated by a standard reflective target with known distance, and the parameters of the optical path receiving module 3 are adjusted to ensure accurate distance measurement.

[0045] 2. Laser ranging value calculation: Based on the electrical signal collected by the optical path receiving module 3, the laser ranging value is calculated through a specific algorithm. For example, when using the time-of-flight method, the distance is calculated by measuring the time of the laser round trip. Assuming that the electrical signal collected by the optical path receiving module 3 represents the laser round trip time as t, the distance d can be calculated by the following formula: ; where c is the speed of light.

[0046] 3. Establishment of mapping relationship: The mapping relationship is established by associating the laser ranging value with the initial pixel coordinates of the light spot in the initial visual data. The specific method is to match the laser ranging value with the pixel coordinates of the light spot in the image. For example, through the calibration process, the laser ranging value corresponding to each pixel coordinate can be determined. Assuming that in the calibration process, the initial pixel coordinates of the light spot in the image are , and the corresponding laser ranging value is , then the mapping relationship can be established as .

[0047] Matching of initial contour features: The initial contour features are used as a spatial reference to determine the position and shape of the target object in the image. This can be achieved by matching the contour features with the pixel coordinates in the image. For example, assuming that the initial contour features are a rectangular region with the upper left corner coordinates (x1, y1) and the lower right corner coordinates (x2, y2), the position of the target object in the image can be determined through these coordinates.

[0048] Step S300, based on the above mapping relationship, the laser signal collected by the optical path receiving module 3 is converted into electrical signal and calculated into target real-time distance data, and the dynamic image reflected by the second inclined mirror 4 is obtained through the zoom camera 5, and the real-time pixel coordinates of the light spot and the target real-time contour features are identified.

[0049] Wherein, dynamic image: refers to the image sequence collected by the zoom camera 5, reflecting the real-time state of the target object in the measurement process. Real-time data synchronization: refers to the alignment of the real-time distance data collected by the optical path receiving module 3 and the dynamic image data collected by the zoom camera 5 in time, ensuring that the two correspond at the same time, providing accurate data basis for subsequent three-dimensional reconstruction.

[0050] The necessary process is described as follows:

[0051] 1. Real-time laser signal acquisition and conversion: The optical path receiving module 3 receives the laser signal and converts it into an electrical signal, which is converted into a digital signal after amplification and filtering processing.

[0052] 2. Real-time distance data calculation: The circuit processing unit 6 calculates real-time distance data based on the digital signal using the time-of-flight method. For example, if the laser round-trip time is 12 ns, the calculated distance is 1.8 m.

[0053] 3. Dynamic image acquisition and processing: The zoom camera 5 captures dynamic images, and Gaussian filtering is used to remove noise, ensuring clear images.

[0054] 4. Real-time pixel coordinates extraction of the light spot: The threshold segmentation algorithm is used to extract the real-time pixel coordinates of the light spot. For example, setting the threshold value to 128, the extracted light spot coordinates are (320, 220).

[0055] 5. Real-time contour feature extraction of the target: The Canny algorithm is used to extract the real-time contour feature of the target object. For example, the extracted contour feature is a rectangular region, with the upper left corner coordinates (110, 110) and the lower right corner coordinates (510, 410).

[0056] 6. Data association and synchronization: The real-time distance data and the real-time pixel coordinates of the light spot are associated through the mapping relationship, ensuring data synchronization and providing accurate data for subsequent three-dimensional detection.

[0057] Step S400, call the preset data fusion rule, associate the real-time distance data, real-time pixel coordinates of the light spot and real-time contour feature, and calculate the target flatness and step difference based on the initial contour feature.

[0058] Among them, the data fusion rule refers to the rule and method of integrating data from different sensors (such as the optical path receiving module 3 and the zoom camera 5) to improve the accuracy and reliability of measurement. Data fusion: call the preset data fusion rule to integrate real-time distance data, real-time pixel coordinates of the light spot and real-time contour feature. For example, use Kalman filter (Kalman Filter) or particle filter (Particle Filter) algorithm for data fusion to improve the accuracy and reliability of measurement.

[0059] Flatness: refers to the flatness of the surface of the target object, usually evaluated by measuring the height difference of each point on the surface. Flatness calculation: based on the initial contour feature, calculate the height difference of each point on the surface of the target object. For example, use the least squares method to fit the plane equation of the target object surface, then calculate the distance from each point to the fitted plane to evaluate the flatness.

[0060] Step difference: refers to the height difference between different regions of the target object surface, used to evaluate the flatness and consistency of the object surface. Step difference calculation: by comparing the height values of different regions of the target object surface, calculate the step difference. For example, use region segmentation algorithm to divide the target object surface into multiple regions, then calculate the height difference between regions.

[0061] The necessary procedures are described as follows:

[0062] 1. Data fusion: call the preset data fusion rule to integrate the real-time distance data, real-time pixel coordinates of the light spot, and real-time contour features. For example, use a Kalman filter to fuse the real-time data to improve the accuracy and reliability of the measurement.

[0063] 2. Flatness calculation: use the least squares method to fit the plane equation of the target object surface based on the initial contour features. Calculate the distance of each point to the fitted plane to evaluate the flatness. For example, assuming the fitted plane equation is z=ax+by+c, calculate the distance of each point (x i ,y i ,z i ) to the plane :

[0064] ; by statistical distribution of these distances, evaluate the flatness.

[0065] 3. Step difference calculation: use region segmentation algorithm to divide the target object surface into multiple regions. Calculate the average height of each region, then calculate the height difference between regions to evaluate the step difference. For example, assuming the target object surface is divided into two regions, the average height of region 1 is h1, and the average height of region 2 is h2, then the step difference Δh is: .

[0066] Step S500, integrate the real-time distance data, corrected flatness and step difference information, and output the detection results containing the three-dimensional coordinates of the target object, flatness and step difference parameters.

[0067] Three-dimensional coordinates: refer to the position coordinates of each point on the surface of the target object in three-dimensional space, usually represented as (x, y, z). Detection results: refer to the integrated measurement data, including the three-dimensional coordinates of the target object, flatness and step difference parameters, used to describe the shape and surface characteristics of the target object.

[0068] Detection result output: integrate the calculated three-dimensional coordinates, flatness and step difference parameters into a data structure, and output as the detection result.

[0069] A control method of a related light measurement device based on a bevel mirror further includes an updating method of a mapping relationship, specifically as follows:

[0070] Step S301, every interval preset period, extract the target real-time contour feature, record the device state and environmental variable related to the contour collection synchronously, wherein the device state includes the real-time emission power of the laser emission head 1, the focal length locking state of the zoom camera 5, the signal gain value of the light path receiving module 3, and the environmental variable includes the real-time illumination intensity of the target area.

[0071] Wherein, the preset period: the time interval of the specified period, used to perform certain operations periodically, such as data collection or state check. Device state: refers to the current working state of each component in the measuring device, including the real-time emission power of the laser emission head 1, the focal length locking state of the zoom camera 5, the signal gain value of the light path receiving module 3, etc. Environmental variable: refers to external factors in the measurement environment that may affect the measurement results, such as the real-time illumination intensity of the target area.

[0072] The necessary process is described as follows:

[0073] 1. Extraction of target real-time contour feature: every interval preset period (such as every 10 seconds), the zoom camera 5 collects the image of the target object. The Canny edge detection algorithm is used to extract the real-time contour feature of the target object. For example, the extracted contour feature is a rectangular area with the upper left corner coordinates (110, 110) and the lower right corner coordinates (510, 410).

[0074] 2. Record the device state: record the device state related to the contour collection synchronously, including the real-time emission power of the laser emission head 1, the focal length locking state of the zoom camera 5, the signal gain value of the light path receiving module 3. For example, the real-time emission power of the laser emission head 1 is recorded as 5mW, the focal length locking state of the zoom camera 5 is 50mm, and the signal gain value of the light path receiving module 3 is 1.5.

[0075] 3. Record the environmental variable: record the real-time illumination intensity of the target area synchronously. For example, the illumination intensity of the target area is recorded as 500 lux.

[0076] Step S302, using the iterative closest point algorithm to match the real-time contour feature and the initial contour feature, calculating the root mean square of the coordinate offset of the feature point pair as the basic contour deviation.

[0077] Wherein, the iterative closest point algorithm: an algorithm for matching two sets of point clouds or contour features, finding the closest point pair by iteration and calculating the optimal transformation matrix to align the two sets of points. Basic contour deviation: refers to the deviation between the real-time contour feature and the initial contour feature, which is usually evaluated by calculating the root mean square of the coordinate offset of the feature point pair.

[0078] The necessary process is described as follows:

[0079] 1. Real-time contour feature matching with initial contour feature: Use ICP algorithm to match real-time contour feature with initial contour feature. ICP algorithm finds the nearest point pair by iteration and calculates the optimal transformation matrix, so that the two sets of contour features are aligned. In short, ICP algorithm will constantly adjust the position and direction of real-time contour feature, so that it is as close as possible to the initial contour feature.

[0080] 2. Calculate the basic contour deviation: Calculate the root mean square (RMS) of the coordinate offset of the feature points as the basic contour deviation. Specifically, calculate the coordinate offset of each feature point, and then calculate the root mean square value of these offsets. The smaller the root mean square value, the closer the real-time contour is to the initial contour.

[0081] Step S303, based on the obtained device state, the preset device state deviation calculation method is used to calculate the device state deviation, and based on the obtained environment variable, the preset environment deviation calculation method is used to calculate the environment variable deviation.

[0082] Device state deviation: refers to the difference between the current working state of each component in the measuring device and the standard or initial state, such as the power change of laser emitter 1, the focal length change of camera, etc. Environment variable deviation: refers to the difference between the external factors that may affect the measurement results in the measurement environment and the standard or initial state, such as the change of light intensity.

[0083] Device state deviation calculation is as follows:

[0084] 1. Laser emitter 1 power deviation: The power of laser emitter 1 is monitored in real time by circuit processing unit 6, and the difference between the standard power and the current power is calculated.

[0085] 2. Zoom camera 5 focal length deviation: The current focal length is obtained through the control interface of the camera, and the difference between the standard focal length and the current focal length is calculated.

[0086] 3. Gain deviation of light path receiving module 3: The current gain value is obtained through the system preset signal processing unit, and the difference between the standard gain value and the current gain value is calculated. The signal processing unit is usually a system-level independent module (or circuit, chip) for centralized processing of electrical signals of multiple modules (including light path receiving module).

[0087] Environment variable deviation calculation is as follows:

[0088] Light intensity deviation: The light intensity of the target area is measured in real time by the environment sensor (such as light sensor), and the difference between the standard light intensity and the current light intensity is calculated.

[0089] Step S304, according to the device state deviation and the environment variable deviation, respectively give the correction coefficient corresponding to the basic contour deviation, get the corrected contour deviation.

[0090] wherein, correction coefficient: a coefficient used to adjust the base profile deviation according to the equipment state deviation and the environmental variable deviation, for reflecting the influence degree of the deviation on the measurement result. Corrected profile deviation: the base profile deviation adjusted by the correction coefficient, which more accurately reflects the actual deviation of the target object profile.

[0091] The necessary process is described as follows:

[0092] Equipment state correction coefficient: calculate the correction coefficient according to the equipment state deviation. For example, assuming that the equipment state deviation is , the correction coefficient can be expressed as: wherein, is a preset proportional coefficient.

[0093] Environmental variable correction coefficient: calculate the correction coefficient according to the environmental variable deviation. For example, assuming that the environmental variable deviation is , the correction coefficient can be expressed as: wherein, is a preset proportional coefficient.

[0094] Calculate the corrected profile deviation: adjust the base profile deviation using the equipment state correction coefficient and the environmental variable correction coefficient. Assuming that the base profile deviation is , the corrected profile deviation can be expressed as: .

[0095] Step S305, if the corrected profile deviation exceeds the preset threshold, adjust the mapping relationship conversion parameters of the laser ranging value and the visual pixel coordinate based on the profile transformation matrix output by the iterative closest point algorithm, combined with the correction coefficients of the equipment state and the environmental variable, through the least square method.

[0096] wherein, profile transformation matrix: output by the iterative closest point algorithm (ICP), used to describe the spatial transformation relationship between the real-time profile features and the initial profile features. Mapping relationship conversion parameters: parameters used to convert the laser ranging value into the visual pixel coordinate, which may need to be adjusted according to the profile transformation matrix and the correction coefficient.

[0097] The necessary process is described as follows:

[0098] 1. Determine the corrected profile deviation: check whether the corrected profile deviation exceeds the preset threshold. If it exceeds, proceed to the next adjustment; if it does not exceed, keep the current mapping relationship conversion parameters unchanged.

[0099] 2. Calculate the mapping relationship conversion parameter: use the contour transformation matrix and the correction coefficient, adjust the mapping relationship conversion parameter through the least square method. For example, if the contour transformation matrix is , the correction coefficient is and , the new mapping relationship conversion parameter can be calculated as follows: , where d is the distance function, P is the original mapping relationship conversion parameter, is the visual pixel coordinate.

[0100] Step S306, if the corrected contour deviation is less than or equal to the preset threshold, it is maintained unchanged.

[0101] Wherein, the preset threshold: a value set in advance, used to judge whether the mapping relationship conversion parameter needs to be adjusted.

[0102] The necessary process is described as follows:

[0103] 1. Determine the contour deviation: compare the corrected contour deviation with the preset threshold, and determine whether the mapping relationship conversion parameter needs to be adjusted.

[0104] 2. Maintain the current parameter: if the corrected contour deviation is less than or equal to the preset threshold, keep the current mapping relationship conversion parameter unchanged.

[0105] According to the device state deviation and the environmental variable deviation, the corresponding correction coefficient of the basic contour deviation is given, and the corrected contour deviation is obtained, including:

[0106] Step S3041, based on the preset angle parameter of the first inclined mirror 2 and the second inclined mirror 4, the real-time contour feature, the spot pixel coordinate and the laser ranging data are geometrically corrected by the preset space coordinate conversion algorithm, the inherent space deviation caused by the double inclined reflection is eliminated, and the corrected real-time contour feature is obtained.

[0107] Wherein, geometric correction: refers to adjusting the measurement data through algorithm to eliminate the system error caused by the geometric layout of the measuring device. Space coordinate conversion algorithm: an algorithm for converting measurement data from one coordinate system to another coordinate system to correct the influence caused by the geometric layout of the device.

[0108] The necessary process is described as follows:

[0109] 1. Determine the geometric transformation parameter: determine the angle, position and direction of the inclined mirror and other geometric transformation parameters through the system calibration process.

[0110] 2. Apply spatial coordinate transformation algorithm: Use these parameters to correct real-time contour features, spot pixel coordinates, and laser ranging data. The algorithm includes rotation, translation, and scaling operations to adapt to different geometric transformation needs.

[0111] 3. Perform geometric transformation:

[0112] Rotation correction: According to the angle θ of the bevel mirror, perform rotation correction on the contour features. For example, if the contour feature coordinates are (x, y), the rotated coordinates (x', y') are calculated as follows:

[0113] ; .

[0114] Translation correction: If the position of the bevel mirror causes image offset, perform translation correction on the contour features. The translation vector can be obtained by measuring system calibration, and then applied as follows: ; , is the two-dimensional coordinate value after translation compensation.

[0115] Scaling correction (if necessary): If the reflection of the bevel mirror causes image scale change, perform scaling correction on the contour features. The scaling factor S can be obtained by measuring system calibration, and then applied as follows: ; S , represents the final target coordinates after scaling transformation.

[0116] 4. Integrate correction data: Integrate the corrected real-time contour features, spot pixel coordinates, and laser ranging data for subsequent steps.

[0117] Step S3042, based on the corrected real-time contour features, combined with the initial contour features, the corrected basic contour deviation is obtained through the preset feature matching algorithm.

[0118] where, feature matching algorithm: an algorithm for identifying and corresponding similar or identical feature points in two sets of feature data (such as contour features).

[0119] The necessary process is described as follows:

[0120] 1. Feature extraction: Extract key feature points from the corrected real-time contour features and initial contour features. These feature points can be inflection points, vertices or other significant geometric features on the contour.

[0121] 2. Apply feature matching algorithm: Use a pre-defined feature matching algorithm, such as Iterative Closest Point (ICP), to match the feature points in the real-time contour feature with the feature points in the initial contour feature. For example, if there is a point in the real-time contour feature at (150, 250), the algorithm will find the closest point in the initial contour feature to match.

[0122] 3. Calculate deviation: For each pair of matched feature points and , calculate their difference in spatial coordinates. This can be done by calculating the Euclidean distance:

[0123] .

[0124] 4. Aggregate deviation evaluation: Calculate an aggregate deviation value, such as the average deviation, from all matched feature points: ; where n is the total number of matched feature points, is the deviation of the i-th pair of feature points.

[0125] Extract key feature points from the corrected real-time contour feature and the initial contour feature. These feature points can be inflection points, vertices, or other significant geometric features on the contour.

[0126] Step S3043, based on the obtained device state deviation value, divide it into a pre-set deviation level interval through a pre-set interval mapping algorithm, and correspondingly assign a pre-set device correction coefficient.

[0127] where interval mapping algorithm: an algorithm used to map the actual measured deviation value to a pre-set deviation level interval. Deviation level: according to the size of the deviation value, the deviation is divided into different levels, so as to apply different correction strategies. Correction coefficient: the coefficient corresponding to the deviation level, used to adjust the measurement result to compensate for the influence of deviation.

[0128] The necessary process is described as follows:

[0129] 1. Determine the deviation level interval: pre-set different deviation level intervals, for example: [0, 0.01], (0.1, 0.2], (0.2, 0.3], etc., each interval corresponds to a different deviation level.

[0130] 2. Apply interval mapping algorithm: use the interval mapping algorithm to map the actual device state deviation value to the pre-set deviation level interval. For example, if the device state deviation value is 0.15, it will be mapped to the (0.1, 0.2] interval.

[0131] 3. Assign a correction coefficient: Assign a correction coefficient for each deviation level. For example: For the interval [0, 0.01], the correction coefficient may be 1.00 (no deviation or very small deviation, no correction needed). For the interval (0.1, 0.2], the correction coefficient may be 1.01 (slight deviation, slight correction needed). For the interval (0.2, 0.3], the correction coefficient may be 1.02 (moderate deviation, moderate correction needed).

[0132] 4. Calculate the correction coefficient: According to the deviation level obtained by mapping, select the corresponding correction coefficient from the pre-set correction coefficient table. For example, if the deviation value 0.15 is mapped to the interval (0.1, 0.2], select the correction coefficient 1.01.

[0133] Step S3044, by pre-set coefficient fusion algorithm, the pre-set device correction coefficient and the pre-set environment correction coefficient are multiplied to obtain the comprehensive correction coefficient.

[0134] Wherein, the pre-set device correction coefficient: the correction coefficient corresponding to the deviation level obtained by interval mapping algorithm according to the device state deviation value, used to compensate the influence of device state change on measurement results. The pre-set environment correction coefficient: the correction coefficient obtained by the corresponding calculation method according to the environment variable deviation value, used to compensate the influence of environmental factors change on measurement results. The comprehensive correction coefficient: the coefficient obtained by multiplying the pre-set device correction coefficient and the pre-set environment correction coefficient, used to consider the influence of device state and environmental factors on measurement results, and to make more comprehensive correction to the basic profile deviation. The coefficient fusion algorithm: the algorithm used to fuse the pre-set device correction coefficient and the pre-set environment correction coefficient, in this step, the multiplication operation is used.

[0135] The necessary process is described as follows:

[0136] 1. Determine the pre-set device correction coefficient: According to the device state deviation value, use the pre-set interval mapping algorithm to divide it into the corresponding deviation level interval, and then select the corresponding device correction coefficient from the pre-set correction coefficient table. For example, if the laser emitter 1 power deviation value is 0.15, it is mapped to the interval (0.1, 0.2] by interval mapping algorithm, and the corresponding device correction coefficient is 1.01.

[0137] 2. Determine the pre-set environment correction coefficient: According to the environment variable deviation value, obtain the environment correction coefficient by the pre-set calculation method. Assuming that the environment variable is light intensity, its deviation value is 50 lux, and through the pre-set relationship model between light intensity deviation and correction coefficient, the environment correction coefficient is calculated to be 1.005.

[0138] 3. Apply the coefficient fusion algorithm: multiply the preset device correction coefficient 1.01 and the preset environment correction coefficient 1.005, i.e. 1.01 x 1.005 = 1.01505, to obtain the comprehensive correction coefficient 1.01505.

[0139] Step S3045, multiply the corrected base profile deviation by the comprehensive correction coefficient to obtain the initial corrected profile deviation.

[0140] Step S3046, apply a preset upper limit constraint to the comprehensive correction coefficient: if the comprehensive correction coefficient exceeds the preset threshold, replace the comprehensive correction coefficient with the threshold and recalculate the corrected profile deviation.

[0141] Wherein, the preset upper limit threshold: a preset maximum value used to limit the maximum possible value of the comprehensive correction coefficient. When the comprehensive correction coefficient exceeds this threshold, the threshold will be used to replace the comprehensive correction coefficient to prevent inaccurate or unstable measurement results caused by excessive correction coefficient. Corrected profile deviation: the recalculated profile deviation value after applying the upper limit constraint. This value is used for subsequent measurement and analysis to ensure the accuracy and reliability of the measurement results.

[0142] The necessary process is described as follows:

[0143] 1. Set the preset upper limit threshold: according to experiments and experience, set the preset upper limit threshold of the comprehensive correction coefficient to 1.1. This value is obtained through multiple experiments and data analysis to ensure the stability and accuracy of the measurement results in various situations.

[0144] 2. Check if the comprehensive correction coefficient exceeds the threshold: compare the comprehensive correction coefficient with the preset upper limit threshold. If the comprehensive correction coefficient exceeds 1.1, replace the comprehensive correction coefficient with 1.1. For example, assume that the comprehensive correction coefficient calculated in step S3044 is 1.12. Because 1.12 is greater than 1.1, replace the comprehensive correction coefficient with 1.1.

[0145] 3. Recalculate the corrected profile deviation: use the new comprehensive correction coefficient (i.e. the preset upper limit threshold 1.1) to recalculate the corrected profile deviation. Assume that the corrected base profile deviation is 0.5 mm, recalculate the corrected profile deviation: 0.5 x 1.1 = 0.55 mm.

[0146] Step S3047, output the corrected profile deviation adjusted by the upper limit constraint.

[0147] The modified contour deviation after the upper limit constraint adjustment: after the upper limit constraint is imposed on the comprehensive correction coefficient in step S3046, the contour deviation value is recalculated. If the comprehensive correction coefficient exceeds the preset upper limit threshold, the threshold is used to replace the comprehensive correction coefficient, and the modified contour deviation is recalculated.

[0148] A control method of a slope mirror-based correlation light measurement device further includes a step after synchronously acquiring a dynamic image reflected by the second slope mirror 4 through the zoom camera 5, identifying the real-time pixel coordinates of the light spot and the target real-time contour feature, and specifically as follows:

[0149] Step S3A0, control the spectral acquisition submodule connected with the zoom camera 5 and the light path receiving module 3 through the circuit processing unit 6, and real-time acquisition of the target area image spectrum distribution and laser signal spectrum.

[0150] Among them, the spectral acquisition submodule: a component connected with the zoom camera 5 and the light path receiving module 3, used for real-time acquisition of the target area image spectrum distribution and laser signal spectrum. It can provide spectral information about the target area, help identify and process stray light. Image spectrum distribution: the spectral information of different positions in the target area image, reflecting the spectral characteristics of the region. Laser signal spectrum: the spectral characteristics of the laser signal emitted by the laser emitter 1, used for comparison and analysis with the image spectrum distribution of the target area.

[0151] The necessary process is described as follows:

[0152] 1. Initialize the spectral acquisition submodule: the circuit processing unit 6 sends an initialization signal to start the spectral acquisition submodule. Ensure that the spectral acquisition submodule is connected normally with the zoom camera 5 and the light path receiving module 3, and prepare to start acquisition.

[0153] 2. Real-time acquisition of image spectrum distribution: the spectral acquisition submodule acquires the image of the target area through the zoom camera 5, and simultaneously acquires the spectrum distribution of each pixel point in the image. These spectral data reflect the spectral characteristics of the target area, including the light intensity of different wavelengths.

[0154] 3. Real-time acquisition of laser signal spectrum: the spectral acquisition submodule simultaneously acquires the spectrum of the laser signal emitted by the laser emitter 1. The laser signal spectrum is known, but the collected data can be used for comparison with the image spectrum distribution to identify stray light.

[0155] 4. Data transmission and preliminary processing: the acquired image spectrum distribution and laser signal spectrum data are transmitted to the processing module through the circuit processing unit 6. The circuit processing unit 6 performs preliminary processing on these data, such as filtering and normalization, for subsequent analysis.

[0156] Step S3B0, analyze the collected image spectral distribution and laser signal spectrum by a preset feature matching algorithm to identify the stray light type, wherein the stray light type includes outdoor direct sunlight and indoor high-frequency flicker light.

[0157] Wherein, the feature matching algorithm: an algorithm used to compare and match the collected image spectral distribution and laser signal spectrum to identify the stray light type. This algorithm is based on a pre-set feature library and determines the type of stray light by comparing spectral features. Stray light type: refers to non-target light sources that interfere with measurements, such as outdoor direct sunlight and indoor high-frequency flicker light. These stray lights can affect the accuracy and reliability of the measurement. Outdoor direct sunlight: refers to light directly from the sun, with a wide spectral distribution and high intensity, which can cause strong interference in the measurement. Indoor high-frequency flicker light: refers to high-frequency flicker light generated by indoor lighting devices (such as fluorescent lamps), with different spectral distribution and frequency characteristics from outdoor sunlight.

[0158] The necessary process is described as follows:

[0159] 1. Obtain the collected data: obtain the image spectral distribution and laser signal spectrum of the target area from the spectral acquisition submodule.

[0160] 2. Preprocess the collected data: preprocess the collected spectral data, such as filtering, normalization, etc., to eliminate noise and improve data quality.

[0161] 3. Feature extraction: extract key features from the preprocessed image spectral distribution, such as spectral peak, spectral width, etc.

[0162] 4. Feature matching: use a pre-set feature matching algorithm to compare the extracted features with a pre-set stray light feature library to identify the stray light type.

[0163] 5. Identify the stray light type: according to the matching result, determine the stray light type. If the feature of outdoor direct sunlight is matched, it is identified as outdoor direct sunlight; if the feature of indoor high-frequency flicker light is matched, it is identified as indoor high-frequency flicker light.

[0164] Step S3C0, for outdoor direct sunlight, the control circuit processing unit 6 switches the emission wavelength band of the laser emission head 1, so that the overlap degree of the switched emission wavelength band and the sunlight interference wavelength band is ≤ a pre-set proportion, and the peak reflection wavelength band of the narrow-band reflection film plated on the surface of the second inclined mirror 4 is matched, the white balance parameter of the zoom camera 5 is adjusted synchronously, and the gain of the laser wavelength channel signal is improved by a pre-set proportion.

[0165] Wherein, sunlight interference band: the spectral distribution range of outdoor sunlight, usually wide and high intensity. Second slope mirror 4 narrow-band reflection film peak reflection band: the central wavelength range of the narrow-band reflection film coated on the second slope mirror 4, used for efficiently reflecting specific band of laser. Zoom camera 5 white balance parameter: parameter used to adjust the color balance of the camera to ensure the image maintains accurate color performance under different lighting conditions. Laser wavelength channel signal gain: gain parameter used to enhance specific laser wavelength signal to improve signal strength and clarity.

[0166] The necessary processes are described as follows:

[0167] 1. Identify outdoor sunlight direct: through the feature matching algorithm of step S3B0, it has been identified that there is outdoor sunlight direct interference in the current environment.

[0168] 2. Switch laser emission band: circuit processing unit 6 controls laser emission head 1 to switch to a new emission band, ensuring that the overlap of the switched emission band and the sunlight interference band does not exceed the preset proportion. For example, if the preset proportion is 10%, the overlapping part of the switched emission band and the sunlight interference band should not exceed 10%. At the same time, the switched emission band should match the peak reflection band of the second slope mirror 4 narrow-band reflection film to ensure that the laser signal can be efficiently reflected and received.

[0169] 3. Adjust zoom camera 5 white balance parameter: circuit processing unit 6 synchronously adjusts the white balance parameter of zoom camera 5 to adapt to the new lighting conditions, ensuring the color accuracy of the image.

[0170] 4. Increase laser wavelength channel signal gain: circuit processing unit 6 increases the signal gain of the laser wavelength channel by a preset proportion to enhance the strength and clarity of the laser signal. For example, if the preset proportion is 20%, the signal gain is increased by 20%.

[0171] Step S3D0, for indoor high-frequency flickering light, control circuit processing unit 6 to make the laser emission timing reverse synchronization with the flickering light period, ensuring to avoid the peak time and the timing deviation ≤ preset time length, at the same time control zoom camera 5 to enable anti-flicker mode, frame rate set to a preset multiple of the flickering light period.

[0172] Wherein, laser emission timing: the time sequence of laser emission head 1 emitting laser, which can be adjusted by circuit processing unit 6. Flickering light period: the period of indoor high-frequency flickering light, usually generated by lighting equipment (such as fluorescent lamp). Timing deviation: the deviation between the laser emission timing and the flickering light period, which needs to be controlled within the preset range. Anti-flicker mode: a working mode of zoom camera 5, used to reduce the impact of high-frequency flickering light on image acquisition. Frame rate: the number of image frames per second collected by zoom camera 5, which can be adjusted by circuit processing unit 6.

[0173] The necessary procedures are as follows:

[0174] 1. Identify indoor high-frequency flickering light: Through the feature matching algorithm in step S3B0, it has been identified that the current environment has indoor high-frequency flickering light interference.

[0175] 2. Adjust the laser emission timing: The circuit processing unit 6 controls the emission timing of the laser emission head 1 to be inversely synchronized with the flickering light period. Specifically, the laser emission timing should avoid the peak time of the flickering light, ensuring that the timing deviation does not exceed the preset time length.

[0176] 3. Enable the anti-flicker mode of the zoom camera 5: The circuit processing unit 6 controls the zoom camera 5 to enable the anti-flicker mode to reduce the impact of high-frequency flickering light on image acquisition.

[0177] 4. Set the frame rate of the zoom camera 5: The circuit processing unit 6 sets the frame rate of the zoom camera 5 to a preset multiple of the flickering light period to ensure stable operation in the anti-flicker mode.

[0178] Step S3E0, mark the real-time contour features and spot real-time pixel coordinates collected after adjustment for credibility. The outdoor direct sunlight area is marked as low credibility, and the indoor non-high-frequency flickering area and other areas without significant stray light interference are marked as high credibility.

[0179] Among them, real-time contour features: real-time contour information of the target object obtained by the zoom camera 5. Spot real-time pixel coordinates: real-time pixel position of the laser spot in the image obtained by the zoom camera 5. Credibility marking: quality assessment of the collected data, marked as high credibility or low credibility, in order to distinguish the reliability of the data in subsequent processing.

[0180] The necessary procedures are as follows:

[0181] 1. Obtain adjusted data: Obtain the adjusted real-time contour features and spot real-time pixel coordinates from the previous steps.

[0182] 2. Evaluate environmental conditions: Evaluate the credibility of the data based on the current environmental conditions (such as whether there is outdoor direct sunlight or indoor high-frequency flickering light).

[0183] 3. Mark the credibility: For outdoor direct sunlight areas, mark as low credibility, because direct sunlight may cause large data errors. For indoor non-high-frequency flickering areas and other areas without significant stray light interference, mark as high credibility, because the data in these areas is relatively reliable.

[0184] The system invokes preset data fusion rules, associates real-time distance data, real-time pixel coordinates of the light spot, and real-time contour features, and calculates the target flatness and step difference based on the initial contour features, including:

[0185] Step S410: Based on the labeled high-confidence data and low-confidence data, assign a preset high weight to the high-confidence data and a preset low weight to the low-confidence data. Then, use a weighted fusion algorithm to associate and fuse real-time distance data, real-time pixel coordinates of light spots, and real-time contour features to generate a weighted 3D point cloud dataset with confidence weights.

[0186] The dataset includes: Preset high weights: Pre-defined weight values ​​used for high-confidence data to enhance its influence during the fusion process. Preset low weights: Pre-defined weight values ​​used for low-confidence data to reduce its influence during the fusion process. Weighted 3D point cloud dataset: A 3D point cloud dataset generated using a weighted fusion algorithm, where each point contains location information and a corresponding confidence weight.

[0187] The necessary process is described below:

[0188] 1. Data Acquisition: Obtain real-time distance data, real-time pixel coordinates of the light spot, real-time contour features, and corresponding confidence labels from the previous steps. This data will serve as input to the weighted fusion algorithm. For example, suppose we have the following data: 1. Real-time distance data: d1, d2, d3; 2. Real-time pixel coordinates of the light spot: (x1, y1), (x2, y2), (x3, y3); 3. Real-time contour features: (c1, c2, c3); 4. Confidence labels: high confidence, low confidence, high confidence.

[0189] 2. Weight Allocation: Based on the confidence level, assign a pre-defined high weight to high-confidence data and a pre-defined low weight to low-confidence data. For example, assume the pre-defined high weight is 0.9 and the pre-defined low weight is 0.1.

[0190] 3. A weighted fusion algorithm is used to correlate and fuse real-time distance data, real-time pixel coordinates of light spots, and real-time contour features. Specifically, for each data point, a weighted 3D point cloud data point is generated by combining its position information, distance information, and weight.

[0191] 4. Integrate all generated weighted 3D point cloud data points into a single dataset. The final weighted 3D point cloud dataset contains the location information, distance information, and corresponding weight for each point.

[0192] Step S420: Using the initial contour features as a spatial reference, extract point cloud data of high-confidence regions from the weighted 3D point cloud dataset, and obtain a reference plane model by fitting a preset plane fitting algorithm.

[0193] Pre-set plane fitting algorithm: a pre-defined algorithm used to fit a plane model from point cloud data. Common algorithms include least squares method, etc. Reference plane model: a plane model obtained by fitting point cloud data from a high-confidence area, used as a reference for subsequent measurements and calculations.

[0194] Extracting high-confidence area point cloud data: extracting data points with weights higher than a pre-set threshold from the weighted three-dimensional point cloud data set. For example, assuming the pre-set threshold is 0.8, extract point cloud data with weights greater than or equal to 0.8 from the data set.

[0195] Applying the pre-set plane fitting algorithm: using the pre-set plane fitting algorithm (such as the least squares method) to fit the extracted high-confidence point cloud data to obtain the reference plane model. The least squares method fits a plane by minimizing the sum of the squares of the perpendicular distances from the points to the plane. Assuming the extracted high-confidence point cloud data is: ; The reference plane model obtained by least squares fitting can be represented as: where a, b, c, d are the fitted plane parameters.

[0196] Step S430, with reference to the reference plane model, calculate the vertical deviation value of the low-confidence area point cloud data from the reference plane model by the pre-set spatial distance algorithm, and count the average deviation value; if the average deviation value exceeds the pre-set deviation threshold, call the multi-frame historical data of the area for cross-validation, and take the deviation value after verification as the effective deviation.

[0197] Pre-set spatial distance algorithm: a pre-defined algorithm used to calculate the vertical distance between point cloud data and the reference plane model. Common algorithms include the distance formula from a point to a plane. Vertical deviation value: the vertical distance between point cloud data and the reference plane model, reflecting the deviation of the data point relative to the reference plane. Pre-set deviation threshold: a pre-set upper limit of the deviation value, used to determine whether the deviation is within an acceptable range. Multi-frame historical data: data collected at multiple time points, used to cross-validate the accuracy of the current data. Cross-validation: comparing data at multiple time points to verify the reliability and accuracy of the current data. Effective deviation: the deviation value confirmed after cross-validation, used for subsequent calculations and analysis.

[0198] The necessary process is described as follows:

[0199] 1. Calculate the vertical deviation value: use the pre-set spatial distance algorithm to calculate the vertical deviation value of the low-confidence area point cloud data from the reference plane model. Assuming the reference plane model is For each low-confidence data point its vertical deviation value can be calculated by the following formula:

[0200] ;

[0201] For example, assuming the reference plane model is 2x + 3y + 4z + 5 = 0, and the point cloud data of the low-confidence area is (1, 2, 3), the vertical deviation value is:

[0202] ;

[0203] 2. Calculate the average deviation value by statistically analyzing the vertical deviation values of all point cloud data in the low-confidence area. Assuming that there are three points of point cloud data in the low-confidence area, and their vertical deviation values are D1 = 4.64, D2 = 3.21, and D3 = 2.78, the average deviation value is:

[0204] Average deviation value = (D1 + D2 + D3) / 3 .

[0205] For the part of determining whether the deviation exceeds the preset deviation threshold, calling multiple historical data for cross-validation, and determining the effective deviation, please refer to steps S431 to S435, which will not be repeated here.

[0206] Step S440, integrate the reference plane model and the effective deviation, and calculate the target overall flatness by the preset flatness algorithm.

[0207] The preset flatness algorithm is an algorithm that is preset and used to calculate the overall flatness of the target object. Common algorithms include the least squares method. The target overall flatness is a parameter that reflects the flatness of the surface of the target object, which is evaluated by calculating the deviation of each point on the surface relative to the reference plane.

[0208] The necessary process is described as follows:

[0209] 1. Integrate the reference plane model and the effective deviation: combine the reference plane model and the effective deviation to form a complete data set for subsequent flatness calculation. For example, assuming the reference plane model is 2x + 3y + 4z + 5 = 0, and the effective deviation is 3.48, the integrated data set includes the reference plane model and the effective deviation value of all points.

[0210] 2. Apply the preset flatness algorithm: use the preset flatness algorithm to calculate the overall flatness of the target object. The commonly used algorithm is the least squares method, which fits the plane by minimizing the sum of the squares of the vertical distances of the points to the plane, and calculates the distances of the points to the fitted plane to evaluate the flatness. For example, assuming that the integrated data set includes the reference plane model and the effective deviation value of all points, the plane model fitted by the least squares method can be represented as: ax + by + cz + d = 0. Where a, b, c, d are the fitted plane parameters.

[0211] 3. Calculate the overall flatness of the target: For each data point , calculate its perpendicular distance to the fitted plane : .

[0212] Statistically sum up the perpendicular distances of all data points, and calculate the mean, standard deviation, etc. to evaluate the overall flatness of the target object. For example, assuming the perpendicular distances of all data points are D1=3.48, D2=3.50, D3=3.45, then the average flatness is:

[0213] Average flatness= . The smaller the standard deviation, the more uniform the flatness.

[0214] Step S450, extract the adjacent boundary points of the real-time contour feature from the weighted three-dimensional point cloud data set, associate the corresponding real-time distance data, calculate the perpendicular distance difference of the adjacent boundary points in the normal direction of the reference plane model, and obtain the initial segment difference value.

[0215] wherein the adjacent boundary points of the real-time contour feature: the boundary points adjacent to the real-time contour feature in the weighted three-dimensional point cloud data set, used for calculating the segment difference. The normal direction of the reference plane model: the normal vector direction of the reference plane model, used for determining the calculation direction of the perpendicular distance. The initial segment difference value: the perpendicular distance difference of the adjacent boundary points in the normal direction of the reference plane model, reflecting the segment difference of the target object surface.

[0216] The necessary procedures are described as follows:

[0217] 1. Extract the adjacent boundary points of the real-time contour feature: extract the boundary points adjacent to the real-time contour feature from the weighted three-dimensional point cloud data set. These boundary points are used for calculating the segment difference. For example, assuming that the real-time contour feature is a rectangular region, extract the four boundary points of the rectangular region.

[0218] 2. Associate the corresponding real-time distance data: associate the extracted adjacent boundary points with the corresponding real-time distance data.

[0219] 3. Calculate the normal direction of the reference plane model: obtain the normal direction of the reference plane model from step S420. Assuming that the reference plane model is ax+by+cz+d=0, its normal vector is (a, b, c).

[0220] 4. Calculate the perpendicular distance difference of the adjacent boundary points: for each pair of adjacent boundary points, calculate their perpendicular distance difference in the normal direction of the reference plane model. For example, for boundary points and , calculate their vertical distance difference :

[0221] .

[0222] Repeat the above process to calculate the vertical distance difference of all adjacent boundary points.

[0223] 5、Get the initial segment difference: integrate all the calculated vertical distance differences to get the initial segment difference.

[0224] Step S460, for the initial segment difference of the low confidence area, superimpose the preset compensation coefficient according to the preset low weight ratio to get the corrected segment difference.

[0225] Where, the initial segment difference of the low confidence area: calculated in step S450, the segment difference located in the low confidence area, which may be affected by measurement error or environmental interference. Preset low weight ratio: preset weight ratio, used to adjust the initial segment difference of the low confidence area to reduce its influence on the final result. Preset compensation coefficient: preset compensation value, used to correct the initial segment difference of the low confidence area to improve the accuracy of the measurement result. Corrected segment difference: segment difference after weight adjustment and compensation, closer to the true value, used for subsequent analysis and application.

[0226] The necessary process is described as follows:

[0227] 1、Get the initial segment difference of the low confidence area: get the initial segment difference of the low confidence area from step S450. For example, assume the initial segment difference is [0.5, 0.6, 0.4].

[0228] 2、Apply the preset low weight ratio: adjust the initial segment difference of the low confidence area according to the preset low weight ratio. Assume the preset low weight ratio is 0.3, the adjusted segment difference is: [0.5x0.3, 0.6x0.3, 0.4x0.3] = [0.15, 0.18, 0.12].

[0229] 3、Superimpose the preset compensation coefficient: superimpose the preset compensation coefficient to the adjusted segment difference. Assume the preset compensation coefficient is 0.2, the corrected segment difference is as follows:

[0230] [0.15+0.2, 0.18+0.2, 0.12+0.2] = [0.35, 0.38, 0.32].

[0231] 4、Get the corrected segment difference: the corrected segment difference [0.35, 0.38, 0.32] is closer to the true value, used for subsequent analysis and application.

[0232] If the average deviation value exceeds the preset deviation threshold, the multi-frame historical data of the region is called for cross-validation, and the deviation value after validation is taken as the effective deviation.

[0233] Step S431, when the average deviation value of the low-confidence region exceeds the preset threshold, the preset number of frames of historical data of the region is extracted from the historical data record, the environmental features of the historical and current data are compared through the preset time series similarity algorithm, and the effective historical data that meets the preset standard of matching degree is screened out, the environmental features including illumination, temperature.

[0234] Among them, the historical data record: the stored multi-frame historical data, used for cross-validation of the accuracy of the current data. The preset number of frames: the number of data frames extracted from the historical data record. The preset time series similarity algorithm: the algorithm preset for comparing the environmental features of the historical data and the current data to evaluate their similarity. Environmental features: including illumination, temperature and other external conditions that affect the measurement results. Effective historical data: the historical data that meets the preset standard of matching degree with the environmental features of the current data screened out through similarity comparison.

[0235] The necessary process is described as follows:

[0236] 1. Determine whether the average deviation value exceeds the preset threshold: compare the average deviation value of the low-confidence region with the preset threshold. If the average deviation value exceeds the preset threshold, it means that the current data may have a large error or interference, which needs to be further verified.

[0237] 2. Extract the preset number of frames of historical data: extract the preset number of frames of historical data of the region from the historical data record. For example, assuming that the preset number of frames is 5, the last 5 frames of data are extracted from the historical data record.

[0238] 3. Compare the environmental features of the historical and current data: use the preset time series similarity algorithm to compare the environmental features of the historical and current data, including illumination and temperature. The similarity algorithm can be based on Euclidean distance, cosine similarity, etc., to evaluate the similarity of the historical and current data.

[0239] 4. Screen effective historical data: according to the comparison result, screen out the historical data that meets the preset standard of matching degree with the environmental features of the current data. For example, assuming that the preset matching degree standard is 0.8, only the historical data with a similarity greater than or equal to 0.8 is retained.

[0240] Step S432, spatial coordinate correction is performed on the effective historical data to ensure consistency with the current coordinate system, and then time series filtering processing is performed to obtain stable historical deviation data set.

[0241] Where, spatial coordinate correction: coordinate transformation is performed on historical data to make it consistent with the current coordinate system, to ensure the comparability of data. Time series filtering processing: filtering processing is performed on the corrected historical data to reduce noise and fluctuations, to obtain stable historical bias data set. Historical bias data set: after correction and filtering processing, the historical data set used for subsequent analysis, containing the bias value of each time point.

[0242] The necessary procedures are described as follows:

[0243] 1. Obtain valid historical data: obtain the filtered valid historical data from step S431. These data have been compared with the similarity of environmental characteristics, to ensure that they have similar environmental conditions with the current data.

[0244] 2. Perform spatial coordinate correction: perform spatial coordinate correction on the valid historical data to ensure its consistency with the current coordinate system. This step usually involves coordinate transformation algorithms such as translation, rotation and scaling. For example, assuming that the current coordinate system is , the coordinate system of the historical data is , the correction needs to be performed by the following steps:

[0245] Translation correction: translate the coordinate origin of the historical data to the origin of the current coordinate system.

[0246] ;

[0247] Where, .

[0248] Rotation correction: adjust the coordinate axis direction of the historical data as needed to make it consistent with the current coordinate system.

[0249] ; where R is the rotation matrix.

[0250] Scaling correction: if necessary, scale the historical data to make it consistent with the scale of the current data.

[0251] .

[0252] Where, is the scaling factor.

[0253] 3. Perform time series filtering processing: perform time series filtering processing on the corrected historical data to reduce noise and fluctuations. Common filtering algorithms include moving average filtering, Kalman filtering, etc. For example, use moving average filtering to process the historical bias data:

[0254] Where D(t) is the bias value at time point t, and N is the window size of the moving average.

[0255] 4. Obtain stable historical deviation dataset: After correction and filtering, a stable historical deviation dataset is obtained. These data will be used for subsequent analysis and verification.

[0256] Step S433, dynamically allocate the fusion weight of the current deviation value and the historical deviation dataset based on the external interference weight, which is calculated by a preset analytic hierarchy process algorithm: classify the environmental interference factors and device state deviation factors according to the preset influence factor, and add up the deviation values of each factor and the corresponding preset weight coefficient after operation, to obtain the weight value of the quantitative external factor comprehensive interference degree.

[0257] Among them, the external interference weight: the weight value of the quantitative external factor (such as environmental interference and device state deviation) affecting the measurement result, used to dynamically adjust the fusion weight of the current deviation value and the historical deviation dataset. Analytic hierarchy process algorithm: a multi-criteria decision analysis method, used to determine the relative importance of each factor and calculate the comprehensive weight. Environmental interference factors: external environmental factors that affect measurement results, such as light intensity, temperature, etc. Device state deviation factors: device state factors that affect measurement results, such as laser emitter head 1 power deviation, camera focal length deviation, etc. Preset influence factor: the relative importance level of each factor preset. Preset weight coefficient: the weight coefficient of each factor preset, used to calculate the comprehensive weight. Fusion weight: used to dynamically adjust the weight of the current deviation value and the historical deviation dataset to improve the accuracy of the fusion result.

[0258] The necessary process is described as follows:

[0259] 1. Obtain environmental interference factors and device state deviation factors: obtain the current environmental interference factors (such as light intensity, temperature) and device state deviation factors (such as laser emitter head 1 power change, camera focal length change) from the measurement environment and device state monitoring module.

[0260] 2. Build a hierarchical model: use the analytic hierarchy process algorithm to build a hierarchical model, classify the environmental interference factors and device state deviation factors according to the preset influence factor. For example, light intensity and temperature are first-level factors, and laser emitter head 1 power change and camera focal length change are second-level factors.

[0261] 3. Calculate the relative importance of each factor: calculate the relative importance of each factor by the analytic hierarchy process algorithm. For example, assume that the relative importance of light intensity is 0.4, the temperature is 0.3, the laser emitter head 1 power change is 0.2, and the camera focal length change is 0.1.

[0262] 4、Calculate the bias value of each factor: Obtain the actual bias value of each factor. For example, assume that the current light intensity bias is 0.1, the temperature bias is 0.2, the laser emitter head 1 power change is 0.15, and the camera focal length change is 0.05.

[0263] 5、Apply the preset weighted sum formula to calculate the bias value of each factor and the corresponding preset weight coefficient, and then accumulate the results to obtain the weight value of the quantized external factor comprehensive interference degree.

[0264] 6、Dynamic allocation of fusion weight: According to the calculated external interference weight, dynamically allocate the fusion weight of the current bias value and the historical bias data set. For example, assume that the external interference weight is 0.135, and the preset fusion weight adjustment formula is:

[0265] Fusion weight 当前 = 1 - external interference weight = 1 - 0.135 = 0.865;

[0266] Fusion weight 历史 = external interference weight = 0.135.

[0267] Step S434, calculate the preliminary fusion bias value by the preset weighted average algorithm.

[0268] The preset weighted average algorithm is an algorithm that calculates a more accurate bias value by weighting the data with a given weight.

[0269] The necessary process is described as follows:

[0270] 1、Obtain the current bias value and the historical bias data set: obtain the current bias value from step S430 or S433, and obtain the historical bias data set from step S432. For example, assume that the current bias value is D current = 0.5, and the historical bias data set is [0.45, 0.48, 0.52].

[0271] 2、Obtain the fusion weight: obtain the fusion weight from step S433. Assume that the weight of the current bias value is W current = 0.865, and the weight of the historical bias data set is W history = 0.135.

[0272] 3、Calculate the weighted average: calculate the preliminary fusion bias value using the preset weighted average algorithm. The weighted average formula is:

[0273] ; where N is the number of data points in the historical bias data set.

[0274] 4、Obtain the preliminary fusion bias value: use the calculation result as the preliminary fusion bias value for subsequent verification and analysis.

[0275] Step S435, the preliminary fusion deviation value is verified for a preset number of iterations, if the fluctuation amplitude ≤ preset stable threshold, it is determined as the effective deviation after verification; otherwise, the screening to fusion step is re-executed until the standard is met and the fluctuation reason is marked.

[0276] Wherein, the preset number of iterations: the maximum number of iterations in the verification process, used to control the termination condition of the verification process. Fluctuation amplitude: the degree of change of the preliminary fusion deviation value in multiple iterations, used to evaluate the stability of the deviation value. The preset stable threshold: the upper limit of the fluctuation amplitude, used to judge whether the deviation value is stable. The effective deviation after verification: the deviation value within the preset stable threshold after multiple iterations of verification, which can be used as a reliable measurement result. Fluctuation reason: the reason for the fluctuation of the deviation value, such as environmental interference, equipment state change, etc., used for subsequent analysis and improvement.

[0277] The necessary process is described as follows:

[0278] 1. Initialize verification parameters: set the current iteration number to 0, and record the preliminary fusion deviation value as the initial verification value. For example, assume that the preliminary fusion deviation value is 0.4977.

[0279] 2. Enter the iteration verification loop: compare the current iteration number with the preset iteration number. If the current iteration number is less than the preset iteration number, continue the verification process; otherwise, terminate the verification and mark it as unqualified.

[0280] 3. Calculate the fluctuation amplitude: in each iteration, recalculate the preliminary fusion deviation value and compare it with the deviation value of the last time to calculate the fluctuation amplitude. For example, assume that the preset stable threshold is 0.01, the deviation value after the first iteration is 0.4980, and the fluctuation amplitude is:

[0281] Fluctuation amplitude = |0.4980 - 0.4977| = 0.0003.

[0282] 4. Determine whether the fluctuation amplitude is within the preset stable threshold: if the fluctuation amplitude is less than or equal to the preset stable threshold, determine the current deviation value as the effective deviation after verification, and terminate the verification process. Otherwise, increase the iteration number and re-execute the screening to fusion step.

[0283] 5. Record the fluctuation reason: if the deviation value does not reach the stable threshold, record the possible fluctuation reasons, such as environmental interference, equipment state change, etc., for subsequent analysis and improvement.

[0284] Before calling the cross-validation of multiple frames of historical data in the region, the re-inspection process is preferentially executed, which specifically includes:

[0285] Step S43A, when the average deviation value of the low-confidence region exceeds the preset threshold, control the laser emitter head 1 to emit laser beams for a preset number of times, and simultaneously collect multiple frames of target images of the region through the zoom camera 5.

[0286] Wherein, the average deviation value of the low-confidence region: the average deviation value of the low-confidence region calculated in step S430, used to evaluate the measurement error of the region.

[0287] Preset threshold: a preset upper limit of deviation value, used to determine whether the deviation is within an acceptable range. Preset number of times: a preset number of times of laser emission, used to control the amount of data collection in the re-inspection process. Multiple frames of target images: multiple frames of images of the target region collected through the zoom camera 5 in the re-inspection process.

[0288] The necessary process is described as follows:

[0289] 1. Determine whether the average deviation value exceeds the preset threshold: compare the average deviation value of the low-confidence region with the preset threshold. If the average deviation value exceeds the preset threshold, it means that the current data may have large errors or interference, and needs to be re-inspected.

[0290] 2. Control the laser emitter head 1 to emit laser beams: when the average deviation value exceeds the preset threshold, control the laser emitter head 1 to emit laser beams for a preset number of times. For example, assuming that the preset number of times is 5, the laser emitter head 1 will emit 5 laser beams.

[0291] 3. Synchronously collect multiple frames of target images: while the laser emitter head 1 emits laser beams, synchronously collect multiple frames of target images of the region through the zoom camera 5. These images will be used for subsequent fusion and noise reduction processing. For example, assuming that the 5 frames of images collected by the zoom camera 5 are .

[0292] Step S43B, fuse and reduce noise of the real-time contour features and spot pixel coordinates of the multiple frames of images, assign weights according to the clarity of each frame and process through a preset weighted average algorithm.

[0293] Wherein, the spot pixel coordinates: the pixel coordinates of the laser spot extracted from the multiple frames of images. Fusion and noise reduction: fuse the data of the multiple frames of images through an algorithm to reduce noise and error and improve data quality. Clarity: a quality indicator of an image, used to evaluate the clarity of the image. Preset weighted average algorithm: an algorithm that calculates a more accurate deviation value by weighting the data with given weights.

[0294] The necessary process is described as follows:

[0295] 1. Extract real-time contour features and spot pixel coordinates: Extract real-time contour features and spot pixel coordinates from multiple frames of images. For example, assume the data extracted from 5 frames of images are as follows:

[0296] Frame 1: Spot pixel coordinates (100, 200); Frame 2: Spot pixel coordinates (102, 202); Frame 3: Spot pixel coordinates (101, 201); Frame 4: Spot pixel coordinates (103, 203); Frame 5: Spot pixel coordinates (102, 202).

[0297] 2. Calculate the sharpness of each frame of image: Calculate the sharpness of each frame of image using image processing algorithms. Assume the sharpnesses are respectively:

[0298] Frame 1: 0.9; Frame 2: 0.85; Frame 3: 0.88; Frame 4: 0.92; Frame 5: 0.87.

[0299] 3. Assign weights: Assign weights according to the sharpness of each frame of image. Assume the weights are respectively: Frame 1: 0.20; Frame 2: 0.19; Frame 3: 0.20; Frame 4: 0.21; Frame 5: 0.20.

[0300] 4. Apply a pre-set weighted average algorithm: Use a weighted average algorithm to fuse and denoise the spot pixel coordinates. The calculation result is:

[0301] . .

[0302] 5. Obtain the fused data: The fused spot pixel coordinates are (101.6, 201.6).

[0303] Step S43C, associate the fused data with real-time distance data, and recalculate the deviation value of the region from the reference plane model as the re-inspection deviation value.

[0304] Wherein, the reference plane model: the plane model fitted in step S420, used for subsequent deviation calculation. Deviation value: the vertical distance between the fused data and the reference plane model, used to evaluate the deviation of the measurement region. Re-inspection deviation value: the recalculated deviation value, used for subsequent verification and analysis.

[0305] A control method of a related light measurement device based on a bevel mirror further includes a step after recalculating the deviation value of the region from the reference plane model as the re-inspection deviation value, as follows:

[0306] Step S43D, extract multiple frames of historical data of the region, and calculate the degree of agreement between the re-inspection deviation value and the historical data using a pre-set degree of agreement calculation algorithm.

[0307] The goodness of fit algorithm is an algorithm used to evaluate the similarity or consistency between the retest bias value and historical data. The goodness of fit is a quantitative indicator that measures the similarity between the retest bias value and historical data, usually a value between 0 and 1, with a higher value closer to 1 indicating a higher goodness of fit.

[0308] The necessary process is described as follows:

[0309] 1. Extract multiple frames of historical data: Extract multiple frames of historical data for the region from the data storage module. These historical data contain the bias values and other related information recorded in previous measurement processes.

[0310] 2. Calculate the goodness of fit: Use the pre-set goodness of fit calculation algorithm to compare the retest bias value with the bias values in each frame of historical data. The algorithm may be based on the difference between the bias values, statistical correlation or other similarity measures. For example, assuming the retest bias value is 15.24, and the bias values in the historical data are [15.0, 15.1, 15.3, 15.2, 15.4]. A simple difference measure can be used to calculate the goodness of fit:

[0311] Goodness of fit = 1 - (max-min) / N , where N is the number of frames of historical data, and max-min is the difference between the maximum and minimum bias values in the historical data.

[0312] 3. Determine the goodness of fit result: The calculated goodness of fit value will be used in the subsequent steps to determine whether the retest bias value can be considered as a valid bias.

[0313] Step S43E, if the goodness of fit is greater than or equal to the pre-set threshold, take the retest bias value as the valid bias.

[0314] Step S43F, otherwise, superimpose the statistical correction amount of the historical data on the basis of the retest bias value to obtain the valid bias.

[0315] Wherein, the retest bias value: the bias value recalculated in step S43C, used for subsequent verification and analysis. The statistical correction amount of the historical data: the correction amount calculated based on the historical data, used to adjust the retest bias value to improve its accuracy.

[0316] The embodiments of the present specific embodiment are the preferred embodiments of the present application, but do not limit the protection scope of the present application, therefore: any equivalent changes made according to the structure, shape, principle of the present application should be covered within the protection scope of the present application.​​

Claims

1. A control method of a slope-mirror-based correlation light measuring device, characterized by, The related light measurement device based on the slope mirror comprises: a laser emitting head (1) for emitting a laser beam propagating along a preset direction and shooting at a target object; a first slope mirror (2) arranged on the laser propagation path between the laser emitting head (1) and the target object, the slope of which is a mirror glass surface and has a through hole for the laser beam to pass through, the axis of the through hole being collinear with the light emitting direction of the laser emitting head (1), so that the laser beam passes through the through hole and directly shoots at the target object; a light path receiving module (3) arranged adjacent to the laser emitting head (1), both of which are located on the side of the first slope mirror (2) away from the target object, for receiving the laser beam returned along the original path after being reflected by the target object and passing through the through hole, and converting the reflected laser beam into an electric signal; a second slope mirror (4) arranged at a preset angle with the slope of the first slope mirror (2), the mirror surface of which faces the side where the first slope mirror (2) is located, and the mirror surface is a mirror glass surface, configured to reflect the light from the target object and the light spot formed by the laser beam on the target object; a zoom camera (5) with its lens facing the mirror surface of the second slope mirror (4), for receiving the light reflected by the second slope mirror (4) and obtaining the visual profile of the target object and the spatial position information of the light spot formed by the laser beam on the target object; a circuit processing unit (6) electrically connected with the laser emitting head (1), the light path receiving module (3) and the zoom camera (5) respectively, for controlling the emission timing of the laser emitting head (1), receiving the electric signal output by the light path receiving module (3) and processing it to calculate the distance, flatness and step difference of the target object, controlling the zoom camera (5) to work and processing the visual information obtained by it, and realizing the three-dimensional detection of the target object by integrating the laser signal and the visual information; The control method comprises: controlling the laser emitting head (1) to emit the laser beam according to the preset timing, so that the laser beam directly shoots at the target object along the axis of the through hole of the first slope mirror (2), and simultaneously controlling the zoom camera (5) to collect the target area image reflected by the second slope mirror (4), extracting the initial pixel coordinates of the laser spot and the initial contour features of the target in the image to form initial visual data; collecting the laser signal reflected by the target and passing through the through hole by the light path receiving module (3), converting it into an electric signal and calculating the laser ranging value, combining the initial pixel coordinates of the light spot in the initial visual data to establish the mapping relationship between the laser ranging value and the visual pixel coordinates, and taking the initial contour features as the spatial reference; based on the above mapping relationship, converting the laser signal collected by the light path receiving module (3) into an electric signal and calculating the target real-time distance data, synchronously obtaining the dynamic image reflected by the second slope mirror (4) by the zoom camera (5), and identifying the real-time pixel coordinates of the light spot and the real-time contour features of the target; calling the preset data fusion rule, associating the real-time distance data, the real-time pixel coordinates of the light spot and the real-time contour features, and calculating the flatness and step difference of the target based on the initial contour features as the reference; integrating the real-time distance data, the corrected flatness and step difference information, and outputting the detection results containing the three-dimensional coordinates, the flatness and the step difference parameters of the target object; and further comprising an updating method of the mapping relationship, which is as follows: Every interval preset period, the target real-time contour feature is extracted, and the device state and environmental variable related to the contour collection are recorded synchronously, wherein the device state includes the real-time emission power of the laser emission head (1), the focal length locking state of the zoom camera (5), and the signal gain value of the light path receiving module (3), and the environmental variable includes the real-time illumination intensity of the target area; The real-time contour feature and the initial contour feature are matched by using an iterative closest point algorithm, and the root mean square of the coordinate offset of the feature point pair is calculated as the basic contour deviation; Based on the obtained device state, the device state deviation is calculated by using a preset device state deviation calculation method, and based on the obtained environmental variable, the environmental variable deviation is calculated by using a preset environmental deviation calculation method; According to the device state deviation and the environmental variable deviation, a correction coefficient corresponding to the basic contour deviation is given, and a corrected contour deviation is obtained; If the corrected contour deviation exceeds a preset threshold, the mapping relationship conversion parameters of the laser ranging value and the visual pixel coordinates are adjusted by using the least square method based on the contour transformation matrix output by the iterative closest point algorithm, combined with the correction coefficients of the device state and the environmental variable; If the corrected contour deviation is less than or equal to the preset threshold, it is maintained unchanged.

2. The method of claim 1, wherein, The inner wall of the through hole of the first inclined mirror (2) is provided with a black light-absorbing coating, and the diameter of the through hole is 0.5-1mm larger than the diameter of the laser beam spot.

3. The method of claim 1, wherein, According to the device state deviation and the environmental variable deviation, a correction coefficient corresponding to the basic contour deviation is given, and a corrected contour deviation is obtained, including: Based on the preset included angle parameters of the first inclined mirror (2) and the second inclined mirror (4), the real-time contour feature, the spot pixel coordinates and the laser ranging data are geometrically corrected by using a preset space coordinate conversion algorithm, so as to eliminate the inherent space deviation caused by the double-inclined-surface reflection, and obtain the corrected real-time contour feature; Taking the corrected real-time contour feature as a reference, the initial contour feature is combined, and the corrected basic contour deviation is obtained by using a preset feature matching algorithm; Based on the obtained device state deviation value, the preset deviation level is divided by using a preset interval mapping algorithm, and a preset device correction coefficient is correspondingly given; The preset device correction coefficient and the preset environmental correction coefficient are multiplied by using a preset coefficient fusion algorithm to obtain a comprehensive correction coefficient; The corrected basic contour deviation is multiplied by the comprehensive correction coefficient to obtain an initial corrected contour deviation; A preset upper limit constraint is applied to the comprehensive correction coefficient: if the comprehensive correction coefficient exceeds a preset threshold, the threshold is replaced by the comprehensive correction coefficient, and the corrected contour deviation is recalculated; The corrected contour deviation adjusted by the upper limit constraint is output.

4. The method of claim 3, wherein Further comprising the steps after synchronously acquiring the dynamic image reflected by the second inclined mirror (4) through the zoom camera (5), identifying the real-time spot pixel coordinates and the target real-time contour feature, specifically as follows: The spectral acquisition submodule electrically connected with the zoom camera (5) and the light path receiving module (3) is controlled by the circuit processing unit (6) to acquire the image spectral distribution and the laser signal spectrum of the target area in real time; The collected image spectrum distribution and laser signal spectrum are analyzed by a preset feature matching algorithm to identify stray light types, wherein the stray light types include outdoor direct sunlight and indoor high-frequency flickering light; For outdoor direct sunlight, the control circuit processing unit (6) switches the emission wavelength band of the laser emission head (1), so that the overlap degree of the switched emission wavelength band and the sunlight interference wavelength band is ≤ a preset proportion, and the switched emission wavelength band matches the peak reflection wavelength band of the narrow-band reflection film coated on the surface of the second inclined mirror (4), and the white balance parameter of the zoom camera (5) is adjusted synchronously, so that the gain of the laser wavelength channel signal is improved by a preset proportion; For indoor high-frequency flickering light, the control circuit processing unit (6) reversely synchronizes the laser emission timing with the flickering light period to ensure that the peak time is avoided and the timing deviation is ≤ a preset time length, and simultaneously controls the zoom camera (5) to enable the anti-flash mode, and sets the frame rate to a preset multiple of the flickering light period; The real-time profile features and real-time pixel coordinates of the collected light spots after adjustment are marked for credibility, and the outdoor direct sunlight area is marked as low credibility, and the indoor non-high-frequency flickering area and other areas without significant stray light interference are marked as high credibility.

5. The method of claim 4, wherein, The preset data fusion rule is called to associate the real-time distance data, real-time pixel coordinates of light spots and real-time profile features, and the target flatness and step difference are calculated based on the initial profile features, including: Based on the marked high credibility data and low credibility data, a preset high weight is assigned to the high credibility data, and a preset low weight is assigned to the low credibility data, and the real-time distance data, real-time pixel coordinates of light spots and real-time profile features are associated and fused by a weighted fusion algorithm to generate a weighted three-dimensional point cloud data set containing credibility weight; Taking the initial profile feature as a spatial reference, the point cloud data of the high credibility area is extracted from the weighted three-dimensional point cloud data set, and a reference plane model is obtained by a preset plane fitting algorithm; Taking the reference plane model as a reference, the vertical deviation value of the point cloud data of the low credibility area from the reference plane model is calculated by a preset spatial distance algorithm, and the average deviation value is counted; if the average deviation value exceeds a preset deviation threshold, the multi-frame historical data of the area is called for cross-validation, and the deviation value after validation is taken as the effective deviation; The reference plane model and the effective deviation are integrated, and the target overall flatness is calculated by a preset flatness algorithm; The adjacent boundary points of the real-time profile feature are extracted from the weighted three-dimensional point cloud data set, the corresponding real-time distance data is associated, the normal direction of the reference plane model is taken as the reference, the vertical distance difference of the adjacent boundary points in the normal direction is calculated, and the initial step difference value is obtained; For the initial step difference value of the low credibility area, a preset compensation coefficient is added according to a preset low weight proportion to obtain a corrected step difference value.

6. The method of claim 5, wherein, If the average deviation value exceeds the preset deviation threshold, the multi-frame historical data of the area is called for cross-validation, and the deviation value after validation is taken as the effective deviation, including: When the average deviation value of the low credibility area exceeds the preset threshold, a preset number of historical data of the area is extracted from the historical data record, the environmental features of the historical and current data are compared by a preset time sequence similarity algorithm, the effective historical data with a matching degree reaching a preset standard are selected, and the environmental features include illumination and temperature. The effective historical data is corrected in spatial coordinates to ensure consistency with the current coordinate system, and then is subjected to time series filtering to obtain a stable historical deviation data set; The external interference weight is used to dynamically allocate the fusion weight of the current deviation value and the historical deviation data set, and the external interference weight is calculated by a preset analytic hierarchy process algorithm: the environmental interference factors and the equipment state deviation factors are classified according to a preset influence factor, the deviation values of the factors and the corresponding preset weight coefficients are calculated and accumulated through a preset weighted summation formula, and a weight value of the quantitative external factor comprehensive interference degree is obtained; The preliminary fusion deviation value is calculated by a preset weighted average algorithm; The preliminary fusion deviation value is verified for a preset number of iterations, if the fluctuation amplitude is less than or equal to a preset stable threshold, the verified effective deviation is determined; otherwise, the screening and fusion steps are re-executed until the standard is met and the fluctuation reason is marked.

7. The method of claim 5, wherein, Before cross-verification of multiple frames of historical data of the region is called, a re-inspection process is preferentially performed, specifically including: When the average deviation value of the low-confidence region exceeds a preset threshold, the laser emitting head (1) emits a laser beam for a preset number of times, and simultaneously the zoom camera (5) collects multiple frames of target images of the region; The real-time contour features and spot pixel coordinates of the multiple frames of images are fused and denoised, the weight is allocated according to the definition of each frame, and the preset weighted average algorithm is used for processing; The fused data are associated with real-time distance data, the deviation value of the region from the reference plane model is recalculated, and the recalculated deviation value is taken as the re-inspection deviation value.

8. The method of claim 7, wherein, Further including a step after the recalculated deviation value of the region from the reference plane model is taken as the re-inspection deviation value, specifically as follows: The multiple frames of historical data of the region are extracted, and a preset coincidence calculation algorithm is used to calculate the coincidence between the re-inspection deviation value and the historical data; If the coincidence is greater than or equal to a preset threshold, the re-inspection deviation value is taken as the effective deviation; Otherwise, the re-inspection deviation value is taken as a reference, and the statistical correction amount of the historical data is superimposed to obtain the effective deviation.

Images