Space position measuring method based on heading machine space position measuring system

Abstract

The invention relates to a spatial position measuring method based on a heading machine spatial position measuring system. The spatial position measuring system comprises a light source, a camera, a control module and a calculation module, the light source is arranged on a heading machine body, the control module adjusts the light source to periodically flicker within a specific time interval, the camera is arranged on a roadway wall of a heading working face, and the calculation module is arranged on the roadway wall. The camera collects continuous video frames and transmits the continuous video frames to the calculation module, and the calculation module processes the continuous video frames and outputs spatial position information of the heading machine. According to the spatial position measuring method based on the spatial position measuring system of the heading machine, the accuracy of spatial positioning of the heading machine in roadway operation can be improved, the requirements of high-precision position tracking and automatic control are met, and the intelligent level of fully-mechanized excavation is improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent spatial position measurement technology in coal mines; specifically, this invention relates to a spatial position measurement method based on a tunneling machine spatial position measurement system. Background Technology

[0002] Intelligent mining equipment is a core component of intelligent coal mine construction. It is not only an important measure to implement my country's strategy of "replacing manpower with mechanization and reducing manpower with automation" for safety through science and technology, but also a fundamental path to achieving safe, green, and efficient coal mining. Through continuous efforts in multiple stages, including technology introduction, assimilation, and independent innovation, my country's intelligent fully mechanized mining face technology has developed rapidly. In recent years, the research and development of unmanned or minimally manned automatic cutting control systems and their supporting technologies has become a focus of attention both domestically and internationally. Over the past decade, automatic face straightening technology based on inertial navigation systems and "transparent face" intelligent mining technology based on geological modeling and digital coal seam construction have represented the development directions of fully mechanized mining technology at different stages. These advanced technological paths have accumulated rich technical foundations and practical experience for my country to achieve intelligent unmanned mining.

[0003] Although significant progress has been made in the intelligent technology of fully mechanized mining faces, the production equipment and processes of fully mechanized tunneling faces remain relatively outdated, making it difficult to meet the development requirements of high-yield and high-efficiency mines, leading to an increasingly prominent problem of "mining-tunneling imbalance." Compared with fully mechanized mining systems, most roadway excavation still relies on manually operated cantilever roadheaders, which is not only labor-intensive but also has low production efficiency. Achieving autonomous and precise positioning of cantilever roadheaders has become a key core technology for promoting the construction of intelligent tunneling faces and has significant research value. In traditional construction processes, the direction of roadway excavation usually relies on the light spot formed on the cross-section by a laser pointer for manual judgment. Whether the excavation cross-section conforms to the design specifications of the roadway centerline largely depends on the driver's experience and operating skills. Constrained by current production processes, tunneling quality standards, and multiple factors such as high dust levels and complex geological conditions at the working face, the perception of the position and working condition of tunneling equipment is extremely difficult. Among these, dynamic positioning technology for underground roadway tunneling equipment has become a major bottleneck in the development of intelligent tunneling.

[0004] In underground coal mine tunneling operations, accurate position sensing of the tunneling machine is crucial for safe production and intelligent operation. Existing technologies typically rely on single-laser ranging, inertial measurement, or binocular vision ranging. However, relying solely on single-laser ranging or inertial measurement is insufficient to obtain a complete 3D position, while binocular vision measurement has a relatively short range, which is unsuitable for practical needs. Furthermore, the complex underground environment presents challenges such as uneven lighting, dust obstruction, and other issues, leading to insufficient positioning accuracy. Summary of the Invention

[0005] In view of this, the present invention provides a spatial position measurement method based on a tunneling machine spatial position measurement system, thereby solving or at least alleviating one or more of the above-mentioned problems and other problems existing in the prior art.

[0006] To achieve the aforementioned objectives, the present invention provides a spatial position measurement method based on a tunneling machine spatial position measurement system. The spatial position measurement system comprises a light source, a camera, a control module, and a calculation module. The light source is mounted on the tunneling machine body, and the control module adjusts the light source to flash periodically within a specific time interval. The camera is mounted on the tunnel wall at the tunneling face, and the camera acquires continuous video frames and transmits them to the calculation module. The calculation module processes the continuous video frames and outputs the spatial position information of the tunneling machine. The spatial position measurement method includes the following steps: Step 1: Install a light source on the tunneling machine body and a camera on the tunnel wall at the working face. The light source flashes periodically at specific time intervals, with a flashing frequency of [missing information]. The camera captures continuous video frames. The calculation module performs image preprocessing on the consecutive video frames to obtain preprocessed images. ,in, These are the pixel coordinates of the image in a pixel coordinate system, with the origin of the pixel coordinate system located at the top left corner of the image. The axis is horizontal to the right. The axis points vertically downwards, and the units are pixels. Step 2: Calculate the preprocessed image The brightness characteristics are used to extract regions that meet the brightness characteristics requirements as candidate light spot regions, and the centroid coordinates of the candidate light spot regions are used as candidate light spot points. Step 3: Calculate the geometric shape features of the candidate light spot region, filter the candidate light spot regions whose geometric shape features are similar to those of the light source, and use the centroid coordinates of the filtered candidate light spot regions as the light spot extraction points; Step 4: Extract the brightness sequence of the light spot extraction point over time. Perform a Fourier transform on the brightness sequence to obtain the main frequency. Select light spot extraction points whose main frequency information is similar to that of the light source as light spot retention points; Step 5: Input the preserved light spot points into the deep learning target detection model for auxiliary verification; Step Six: Retrieve the pixel coordinates of the spot points that passed the auxiliary verification in the pixel coordinate system. Converted to three-dimensional coordinates in the world coordinate system ; Step 7: Check the three-dimensional coordinates Is it within the permissible space of the tunnel? If so, then the three-dimensional coordinates... As the spatial position output of the tunneling machine, if not, the three-dimensional coordinates Mark the anomaly and remeasure.

[0007] In the spatial position measurement method described above, step one may optionally include: processing the acquired continuous video frames. The image is preprocessed by performing grayscale conversion, filtering, and contrast enhancement. ,in, .

[0008] In the spatial position measurement method described above, step one may optionally include: determining the region of interest based on the camera's orientation and the tunneling machine's estimated position. Limiting spot detection to the region of interest Inside, among them, .

[0009] In the spatial position measurement method described above, step two may optionally include: processing the preprocessed image... Threshold segmentation is performed to obtain a candidate bright spot set. ,in, in, The brightness threshold is set according to the actual situation. The number of candidate light spots. For the candidate bright spot set Connectivity analysis is performed to mark candidate bright spots with adjacent pixel positions as the same connected region, and the connected region constitutes the candidate region of the light spot.

[0010] In the spatial position measurement method described above, step three may optionally include: calculating the roundness characteristics of the candidate light spot region. and firmness characteristics Set the roundness threshold according to the actual situation. and firmness threshold When the candidate region of the light spot satisfies and When the centroid coordinates of the candidate light spot region are used as the light spot extraction point, wherein, , Where A is the pixel area of ​​the candidate light spot region, and P is the perimeter of the candidate light spot region. The area of ​​the convex hull of the candidate light spot region is denoted as .

[0011] In the spatial position measurement method described above, step four may optionally include: processing the brightness sequence. Perform a Fast Fourier Transform to obtain the frequency sequence. Set the frequency deviation threshold according to the actual situation. When the main frequency With respect to the flicker frequency of the light source satisfy At that time, the light spot extraction points are selected as light spot retention points, wherein, FFT stands for Fast Fourier Transform.

[0012] In the spatial location measurement method described above, step five may optionally include: inputting the spot retention point into the YOLO target detection model, setting a confidence threshold according to actual needs, the model outputting the boundary and centroid coordinates of the spot retention area where the spot retention point is located, if the confidence level output by the model is greater than the confidence threshold, the spot retention point is considered to have passed the auxiliary verification, and if the output confidence level is less than or equal to the confidence threshold, the spot retention point is removed.

[0013] In the spatial position measurement method described above, step six optionally includes: based on the principle of the pinhole camera model, determining the pixel coordinates of the spot retention point that has passed the auxiliary verification in the pixel coordinate system. Converted to three-dimensional coordinates in the world coordinate system ,in, in, Scaling factor The inverse of the camera intrinsic parameter matrix. The inverse of the camera's extrinsic rotation matrix. This is the translation vector of the camera's extrinsic parameters.

[0014] In the spatial position measurement method described above, step six optionally includes: determining the pixel coordinates of the spot retention point that has passed the auxiliary verification in the pixel coordinate system. Subpixel localization is performed using the gray-scale centroid method to obtain subpixel coordinates with subpixel-level localization accuracy. The sub-pixel coordinates Converted to three-dimensional coordinates in the world coordinate system .

[0015] In the spatial position measurement method described above, step six optionally includes: using a laser rangefinder or communication rangefinder with the same field of view as the camera, measuring the actual distance between the spot retention point obtained through the auxiliary verification and the camera. For the three-dimensional coordinates After correction, the corrected three-dimensional coordinates are obtained. ,in, in, These are experience-weighted coefficients set based on actual circumstances.

[0016] The spatial position measurement method based on the tunneling machine spatial position measurement system of the present invention uses the flicker frequency of the light source as a priori marker for visual recognition. Based on image analysis and deep learning technology, it reduces the negative impact of uneven lighting, dust obstruction and other problems on the ranging accuracy in the complex underground environment. It can realize high-precision measurement and positioning of the spatial position of the tunneling machine in underground coal mines, improve the accuracy of spatial positioning of the fully mechanized tunneling face, and enhance the safety and intelligence level of the fully mechanized tunneling industry. Attached Figure Description

[0017] The disclosure of this invention will become more apparent from the accompanying drawings. It should be understood that these drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. In the drawings: Figure 1 This is a schematic diagram of an embodiment of the tunneling machine spatial position measurement system according to the present invention.

[0018] Figure 2 This is a schematic block diagram of an embodiment of the tunneling machine spatial position measurement method according to the present invention.

[0019] Figure 3 This is a flowchart illustrating an embodiment of the tunneling machine spatial position measurement method according to the present invention.

[0020] Figure 4 This is a schematic diagram showing the result of extracting the light source from the candidate region of the light spot using the tunneling machine spatial position measurement method according to the present invention.

[0021] Figure 5 The curve showing the variation of lateral error of the measurement results with measurement distance according to the spatial position measurement method of the tunneling machine of the present invention.

[0022] Figure 6 The curve showing the variation of depth ranging error with measurement distance in the spatial position measurement method of the tunneling machine according to the present invention.

[0023] Attached reference numerals: 1-Tunneling machine; 2-Light source; 3-Streetlight in underground tunnel; 4-Control module; 5-Camera; 6-Computing module. Detailed Implementation

[0024] Referring to the accompanying drawings and specific embodiments, the structure, composition, features, and advantages of the spatial position measurement method based on the tunneling machine spatial position measurement system of the present invention will be described below by way of example. However, all descriptions should not be construed as limiting the present invention in any way.

[0025] Furthermore, for any single technical feature described or implied in the embodiments mentioned herein, or any single technical feature shown or implied in the various figures, the present invention still allows for any combination or deletion of these technical features (or their equivalents) without any technical obstacle, and thus these further embodiments according to the present invention should also be considered within the scope of this description.

[0026] Figure 1 This is a schematic diagram of an embodiment of the tunneling machine spatial position measurement system according to the present invention.

[0027] As shown in the figure, the tunneling machine spatial position measurement system includes a light source 2 mounted on the body of the tunneling machine 1 and a camera 5 mounted on the tunnel wall in the underground roadway. The control module 4 adjusts the light source 2 to flash periodically at specific time intervals, and the calculation module 6 processes and outputs the continuous video frames captured by the camera 5. Simultaneously, there are multiple streetlights 3 in the underground roadway at the tunneling machine's working face. Besides the streetlights 3, other interfering light sources or substances not shown in the figure may also exist in the underground environment.

[0028] The flashing frequency of light source 2 is set according to the site conditions. In an optional embodiment, the time interval between two flashes of light source 2 is between 0.05 seconds and 2 seconds. The flashing frequency characteristic of light source 2 is used as a distinguishing feature between light source 2 and streetlights 3 and other interfering light sources in the underground roadway. Light source 2 can be a high-brightness light source depending on the underground environment, with brightness meeting the shooting requirements of camera 5. In an optional embodiment, control module 4 is installed on the body of tunneling machine 1 and integrates an explosion-proof device. In other optional embodiments, control module 4 can also be installed in other locations underground or remotely, as long as it meets the actual working needs. In an optional embodiment, camera 5 is a mine explosion-proof camera with functions such as low-light resistance, dust resistance, and explosion protection, which can ensure stable acquisition of image data in complex underground environments.

[0029] In this embodiment, light source 2 is a single light source. In other optional embodiments, the number of light sources 2 can also be multiple, with a different flashing frequency set for each light source to achieve the differentiation and positioning of multiple targets and multiple points.

[0030] Figure 2 This is a schematic block diagram of an embodiment of the tunneling machine spatial position measurement method according to the present invention. Figure 3 This is a flowchart illustrating an embodiment of the tunneling machine spatial position measurement method according to the present invention.

[0031] As shown in the figure, the spatial position measurement method of the tunneling machine can include seven steps. It uses visual image analysis and deep learning model technology to preprocess the video information collected by the camera, extract the bright candidate region, screen the circular light source features, perform continuous frame brightness sequence spectrum analysis, target detection model-assisted recognition, sub-pixel positioning and three-dimensional coordinate calculation, and finally determine the spatial position coordinates of the light source, which are then regarded as the spatial position coordinates of the tunneling machine.

[0032] Step one involves video acquisition and preprocessing. A light source with circular geometry is installed on the tunneling machine's body. The control module adjusts the flashing frequency of the light source to... This causes it to flash periodically at specific time intervals. Cameras are installed on the tunnel walls to capture continuous video frames. ,in These are the pixel coordinates of the image in the pixel coordinate system. This assigns the frame number to the time series. The pixel coordinates in the pixel coordinate system refer to the coordinates where the origin of the coordinate system is at the top left corner of the image plane. The axis is horizontal to the right. The axis points vertically downwards, and the unit is pixels.

[0033] To improve imaging quality in low-light and dusty environments, the computing module processes the acquired continuous video frames. Image preprocessing is performed. In this embodiment, the preprocessing methods are noise reduction, grayscale conversion, and brightness enhancement. This can be done by processing consecutive video frames. The color image is converted to a grayscale image, noise is removed by Gaussian filtering, and then histogram equalization is used to enhance the image contrast and increase the image brightness, resulting in the preprocessed image. .

[0034] Histogram equalization is an image processing method that alters the grayscale level of individual pixels by adjusting the image's grayscale histogram. In other alternative embodiments, different methods can be used to achieve preprocessing goals that enhance the visual effect of the image. Preprocessing the acquired video images can improve the accuracy of subsequent feature extraction.

[0035] In an optional embodiment, the region of interest can be determined based on the camera's orientation and the tunneling machine's estimated position. , Limit subsequent steps to the region of interest. Analyzing within a certain range can improve the efficiency and accuracy of spatial position measurement for tunneling machines.

[0036] Step two involves highlight candidate region extraction. In this embodiment, threshold segmentation and connected component analysis are used to extract the highlight candidate region from the preprocessed image. Extract candidate regions of light spots that may be light sources. Set brightness thresholds based on equipment specifications and operational requirements. The preprocessed image Brightness of each pixel and brightness threshold The comparison is performed, and pixels with brightness greater than the threshold are extracted as a set of bright spots. , in This represents the number of pixels.

[0037] For candidate highlight set Connectivity analysis is performed to mark candidate bright spots with adjacent pixel positions as the same connected region. These connected regions constitute candidate spot regions, and the centroid coordinates of these candidate spot regions are used as candidate spot points.

[0038] Step three involves feature filtering for circular light sources. In this embodiment, morphological processing is performed on candidate light spot regions to calculate their geometric features and determine whether the shape of the candidate light spot region is close to a circle. If it is close to a circle, the candidate light spot region is retained for the next step of analysis; otherwise, the region is discarded.

[0039] In an optional embodiment, the roundness of each candidate spot region is calculated. and firmness , , in, Let be the pixel area of ​​the candidate region for the light spot. Let the perimeter of the candidate region of the light spot be denoted as . Let be the convex hull area of ​​the candidate light spot region. The convex hull area refers to the area of ​​the smallest convex polygon that can completely enclose the candidate light spot region.

[0040] Set the roundness threshold according to the actual situation. and firmness threshold When the geometric features of the candidate light spot region satisfy and If the candidate region of the light spot is retained, the centroid coordinates of the candidate region of the light spot are used as the light spot extraction point, and the next step of screening and analysis is carried out.

[0041] Roundness threshold and firmness threshold The threshold is set manually based on the actual spatial location measurement requirements. If there are many candidate areas for the light spot, the threshold can be appropriately increased; if there are few candidate areas, the threshold can be appropriately decreased. The threshold should be set to meet the efficiency and accuracy requirements of location measurement.

[0042] Step four involves continuous frame brightness sequence spectral analysis. In this embodiment, by performing cross-frame correlation and tracking of the candidate light spot regions retained in step three, a temporal brightness change sequence of the candidate light spot regions is established. This brightness change sequence over time is analyzed and calculated to obtain the dominant frequency information of the extracted light spot point. The dominant frequency information is then compared with the flicker frequency of the light source. The comparison is performed. If the frequencies are the same, the candidate region of the light spot is retained and proceeds to the next step; if the frequencies are different, the region is discarded.

[0043] In an optional embodiment, temporal brightness analysis and dominant frequency calculation are performed using Fast Fourier Transform. The brightness sequence of the light spot extraction points obtained in step three is extracted over time. For the brightness sequence Perform a Fast Fourier Transform to obtain the frequency sequence. And calculate the main frequency , FFT stands for Fast Fourier Transform.

[0044] Set the frequency deviation threshold based on actual conditions and experience. The dominant frequency of each light spot extraction point With the flicker frequency of the light source When comparing, At that time, the extracted spot is selected as a spot retention point and proceeds to the next step of analysis. In an optional embodiment, It can be set to Or other values ​​that meet the requirements.

[0045] Step five involves target detection model-assisted identification. In this embodiment, the retained light spot points are input into a deep learning model for identification. If the confidence level output by the deep learning model is greater than 0.7, the retained light spot point is considered to have passed the auxiliary verification; if the confidence level is less than or equal to 0.7, the retained light spot point is discarded. Simultaneously, a consistency judgment on the number of light sources is performed on the retained light spots that have passed the auxiliary verification. If the number of retained light spots that have passed the auxiliary verification is consistent with the system preset or historical prior, the retained light spot point is considered as the extracted target light source, and the pixel coordinates of the target light source in the pixel coordinate system are obtained. If the number is inconsistent, return to the step of extracting the highlight candidate region again until the target light source is successfully extracted.

[0046] In an optional embodiment, the YOLO model is used for auxiliary verification. In other optional embodiments, deep learning models such as RT-DETR can also be used. The spot retention point is input into the YOLO object detection model, and the model outputs the bounding box and centroid coordinates of the spot retention region where the spot retention point is located. The confidence level of the model output is then judged to meet the requirements. In this embodiment, the confidence threshold is set to 0.7. In other optional embodiments, other values ​​that meet the accuracy requirements can also be selected.

[0047] Step six involves sub-pixel localization and 3D coordinate calculation. In this embodiment, the target light source is calculated using camera intrinsic parameters. shaft and The distance along the axis, and the measurement of the light source at... The distance along the axis is used to obtain the centroid coordinates of the extracted target light source.

[0048] In an optional embodiment, the pixel coordinates of the target light source in the pixel coordinate system are determined using the gray-scale centroid method. Perform subpixel localization to obtain subpixel coordinates with subpixel-level localization accuracy. The gray-scale centroid method is an image processing method that calculates the weighted average of gray values ​​in an image to determine the centroid position. It achieves a positioning accuracy of 0.05 to 0.1 pixels, further improving the precision of spatial position measurement for tunneling machines. In other optional embodiments, other sub-pixel positioning methods can also be used.

[0049] Based on the pinhole camera model, subpixel coordinates Converted to three-dimensional coordinates in the world coordinate system , in, Scaling factor This is the inverse of the camera intrinsic parameter matrix. Let be the inverse of the camera's extrinsic rotation matrix. This is the camera's extrinsic translation vector. This conversion formula is used to transform the target light source's coordinates from the pixel coordinate system to the world coordinate system.

[0050] In other alternative embodiments, three-dimensional coordinates can be determined by laser ranging or communication ranging. The Z-axis coordinates are corrected to obtain the corrected three-dimensional coordinates. Using a laser rangefinder or communication rangefinder with the same field of view as the camera, the actual distance between the extracted target light source and the camera is measured. , in, This is an empirical weighting coefficient, typically ranging from 0.8 to 1.0.

[0051] Step seven is to output the tunnel boring machine's position. In this embodiment, the calculated three-dimensional coordinates in the world coordinate system are used as the spatial position of the tunnel boring machine and output to the host computer.

[0052] In an optional embodiment, the validity of the three-dimensional coordinates is checked to determine whether they are within the permissible spatial range of the tunnel. If the three-dimensional coordinates are valid, the spatial location information is uploaded to the host computer via a wireless or wired network for real-time position tracking, automatic control, and three-dimensional modeling of the tunneling machine; if the three-dimensional coordinates are invalid, they are marked as abnormal, and the measurement process is repeated.

[0053] Figure 4 This is a schematic diagram showing the result of extracting the light source from the candidate region of the light spot using the tunneling machine spatial position measurement method according to the present invention.

[0054] As shown in the figure, in practical applications, the spatial position measurement method for tunneling machines of this invention can acquire multiple light spot images from the candidate light spot region. These light spot images have different geometric shapes and flicker frequencies. By measuring the shape characteristics and flicker frequency characteristics of the light source according to this invention, the image of the light source 2 can be extracted from the multiple light spot images acquired by the camera 5 with high confidence.

[0055] Figure 5 The curve showing the variation of lateral error in the measurement results of the tunneling machine spatial position measurement method according to the present invention as a function of measurement distance is shown. Figure 6 The curve showing the variation of depth ranging error with measurement distance in the spatial position measurement method of the tunneling machine according to the present invention.

[0056] As can be seen from the figure, within a measurement distance of 50 meters, the error of the tunneling machine spatial position measurement method of the present invention is at the centimeter level in both the lateral and longitudinal directions. The lateral error does not exceed 3.5 cm, the longitudinal distance measurement error does not exceed 25 cm, and the result error does not exceed 0.5%, which shows high accuracy in spatial positioning.

[0057] The technical scope of this invention is not limited to the contents of the above specification. Those skilled in the art can make various modifications and variations to the above embodiments without departing from the technical concept of this invention, and all such modifications and variations should fall within the scope of this invention.

Claims

1. A method for measuring the spatial position of a tunneling machine based on a tunneling machine spatial position measurement system, characterized in that, The spatial position measurement system includes a light source (2), a camera (5), a control module (4), and a calculation module (6). The light source (2) is installed on the body of the tunneling machine (1). The control module (4) adjusts the light source (2) to flash periodically within a specific time interval. The camera (5) is installed on the tunnel wall of the tunneling face. The camera (5) collects continuous video frames and transmits them to the calculation module (6). The calculation module (6) processes the continuous video frames and outputs the spatial position information of the tunneling machine (1). The spatial position measurement method includes the following steps: Step 1: A light source (2) is installed on the body of the tunneling machine (1), and a camera (5) is installed on the tunnel wall at the tunneling face. The light source (2) flashes periodically at specific time intervals, with a flashing frequency of [missing information]. The camera (5) captures continuous video frames. The calculation module (6) performs image preprocessing on the continuous video frames to obtain the preprocessed image. ,in, These are the pixel coordinates of the image in a pixel coordinate system, with the origin of the pixel coordinate system located at the top left corner of the image. The axis is horizontal to the right. The axis points vertically downwards, and the units are pixels. Step 2: Calculate the preprocessed image The brightness characteristics are used to extract regions that meet the brightness characteristics requirements as candidate light spot regions, and the centroid coordinates of the candidate light spot regions are used as candidate light spot points. Step 3: Calculate the geometric shape features of the candidate light spot region, filter the candidate light spot regions whose geometric shape features are similar to those of the light source (2), and use the centroid coordinates of the filtered candidate light spot regions as the light spot extraction points; Step 4: Extract the brightness sequence of the light spot extraction point over time. Perform a Fourier transform on the brightness sequence to obtain the main frequency. Select light spot extraction points whose main frequency information is similar to that of the light source (2) as light spot retention points; Step 5: Input the preserved light spot points into the deep learning target detection model for auxiliary verification; Step Six: Retrieve the pixel coordinates of the spot points that passed the auxiliary verification in the pixel coordinate system. Converted to three-dimensional coordinates in the world coordinate system ; Step 7: Check the three-dimensional coordinates Is it within the permissible space of the tunnel? If so, then the three-dimensional coordinates... As the spatial position output of the tunneling machine (1), if not, the three-dimensional coordinates are used. Mark the anomaly and remeasure.

2. The spatial position measurement method as described in claim 1, characterized in that, Step one includes: processing the acquired continuous video frames. The image is preprocessed by performing grayscale conversion, filtering, and contrast enhancement. ,in, 。 3. The spatial position measurement method as described in claim 1, characterized in that, Step one includes: determining the region of interest based on the orientation of the camera (5) and the estimated position of the tunneling machine (1). Limiting spot detection to the region of interest Inside, among them, 。 4. The spatial position measurement method as described in claim 1, characterized in that, Step two includes: processing the preprocessed image Threshold segmentation is performed to obtain a candidate bright spot set. ,in, in, The brightness threshold is set according to the actual situation. The number of candidate light spots. For the candidate bright spot set Connectivity analysis is performed to mark candidate bright spots with adjacent pixel positions as the same connected region, and the connected region constitutes the candidate region of the light spot.

5. The spatial position measurement method as described in claim 1, characterized in that, Step three includes: calculating the roundness features of the candidate light spot region. and firmness characteristics Set the roundness threshold according to the actual situation. and firmness threshold When the candidate region of the light spot satisfies and When the centroid coordinates of the candidate light spot region are used as the light spot extraction point, wherein, , Where A is the pixel area of ​​the candidate light spot region, and P is the perimeter of the candidate light spot region. The area of ​​the convex hull of the candidate light spot region is denoted as .

6. The spatial position measurement method as described in claim 1, characterized in that, Step four includes: processing the brightness sequence Perform a Fast Fourier Transform to obtain the frequency sequence. Set the frequency deviation threshold according to the actual situation. When the main frequency The flicker frequency of the light source (2) satisfy At that time, the light spot extraction points are selected as light spot retention points, wherein, FFT stands for Fast Fourier Transform.

7. The spatial position measurement method as described in claim 1, characterized in that, Step five includes: inputting the spot retention point into the YOLO target detection model, setting a confidence threshold according to actual needs, and the model outputting the bounding box and centroid coordinates of the spot retention area where the spot retention point is located. If the confidence score output by the model is greater than the confidence threshold, the spot retention point is considered to have passed the auxiliary verification. If the output confidence score is less than or equal to the confidence threshold, the spot retention point is removed.

8. The spatial position measurement method as described in claim 1, characterized in that, Step six includes: based on the principle of the pinhole camera model, determining the pixel coordinates of the light spot retention point that passed the auxiliary verification in the pixel coordinate system. Converted to three-dimensional coordinates in the world coordinate system ,in, in, Scaling factor The inverse of the camera intrinsic parameter matrix. The inverse of the camera's extrinsic rotation matrix. This is the translation vector of the camera's extrinsic parameters.

9. The spatial position measurement method as described in claim 1, characterized in that, Step six includes: retrieving the pixel coordinates of the spot retention points that passed the auxiliary verification in the pixel coordinate system. Subpixel localization is performed using the gray-scale centroid method to obtain subpixel coordinates with subpixel-level localization accuracy. The sub-pixel coordinates Converted to three-dimensional coordinates in the world coordinate system .

10. The spatial position measurement method as described in claim 1, characterized in that, Step six includes: using a laser rangefinder or communication rangefinder with the same field of view as the camera (5), measuring the actual distance between the spot retention point of the auxiliary verification and the camera (5). For the three-dimensional coordinates After correction, the corrected three-dimensional coordinates are obtained. ,in, in, These are experience-weighted coefficients set based on actual circumstances.

Images