Bit blade parameter extraction method, device, equipment, storage medium and program product

Abstract

This application provides a method, apparatus, device, storage medium, and program product for extracting drill bit blade parameters. The method includes: acquiring multi-view image data of the drill bit blade to be detected; performing cutting tooth detection processing based on the multi-view image data to obtain two-dimensional feature points; performing three-dimensional reconstruction processing based on the multi-view image data to obtain a three-dimensional reconstruction result; wherein the three-dimensional reconstruction result includes a sparse three-dimensional point set and a depth map set; performing feature point matching processing based on the two-dimensional feature points and the three-dimensional reconstruction result to determine a set of three-dimensional feature point coordinates; performing fusion processing based on the set of three-dimensional feature point coordinates to obtain target three-dimensional feature points; constructing a drill bit blade curve model based on the target three-dimensional feature points; and extracting parameters of the drill bit blade to be detected based on the drill bit blade curve model. This improves the stability of multi-view data fusion, thereby improving the accuracy of the drill bit blade parameters.

Description

Technical Field

[0001] This application relates to the field of digital design and testing technology for oil drilling equipment, and in particular to a method, apparatus, equipment, storage medium, and program product for extracting drill bit blade parameters. Background Technology

[0002] In oil and gas drilling engineering, PDC (Polycrystalline Diamond Compact) drill bits are widely used in drilling operations in complex formations due to their high rock-breaking efficiency and adaptability. A PDC drill bit consists of multiple blades, each with several cutting teeth. The spatial arrangement and geometric parameters of these cutting teeth directly affect the drill bit's cutting efficiency, stress state, and service life. Therefore, accurately obtaining the blade parameters of a PDC drill bit is a crucial foundation for drill bit design optimization, manufacturing consistency assessment, and service condition analysis.

[0003] Existing technology discloses a method and system for PDC drill bit wear detection based on multi-view 3D reconstruction (Publication No.: CN117808794A). This method acquires images of worn PDC drill bits, uses deep learning multi-view stereo matching to obtain depth data and generate a preliminary point cloud, obtains a point cloud model through feature mapping, and then uses a neural network to locate the cutting tooth area. The worn and unworn point clouds are registered to calculate the wear degree, outputting the wear location and wear level to improve detection accuracy and reconstruction effect. However, the multi-view data fusion process is susceptible to the effects of viewpoint differences, occlusion, and noise interference, leading to unstable positioning of key structures. Furthermore, existing technology lacks precise modeling and parameter extraction for drill bit blade parameters, resulting in insufficient accuracy of drill bit blade parameters and making it difficult to meet the actual needs of drill bit design, manufacturing, and service analysis.

[0004] Therefore, in the existing technology, there is a problem of insufficient accuracy in the parameters of the drill bit blades. Summary of the Invention

[0005] This application provides a method, apparatus, device, storage medium, and program product for extracting drill bit blade parameters, in order to improve the accuracy of drill bit blade parameters.

[0006] In a first aspect, embodiments of this application provide a method for extracting drill bit blade parameters, including:

[0007] Acquire multi-view image data of the drill bit blade to be inspected;

[0008] Based on multi-view image data, cutting tooth detection processing is performed to obtain two-dimensional feature points;

[0009] Based on multi-view image data, 3D reconstruction processing is performed to obtain 3D reconstruction results; the 3D reconstruction results include sparse 3D point sets and depth map sets;

[0010] Based on the two-dimensional feature points and the three-dimensional reconstruction results, feature point matching processing is performed to determine the set of three-dimensional feature point coordinates;

[0011] Based on the set of three-dimensional feature point coordinates, a fusion process is performed to obtain the target three-dimensional feature points;

[0012] Construct a drill bit blade curve model based on the target's three-dimensional feature points;

[0013] Based on the drill bit blade curve model, the parameters of the drill bit blade to be tested are extracted.

[0014] In one possible implementation, a 3D reconstruction process is performed based on multi-view image data to obtain a 3D reconstruction result, including:

[0015] Based on multi-view image data, the number of cutting teeth is statistically processed to determine the baseline detection quantity;

[0016] Based on the baseline detection quantity, multi-view image data is filtered to determine the target view image frame; where the target view image frame refers to multi-view image data with the same number of cutting teeth as the baseline detection quantity.

[0017] For each target view image frame, cutting tooth identification processing is performed to obtain the corresponding cutting tooth feature points;

[0018] Calculate the similarity between target view image frames based on the corresponding cutting tooth feature points;

[0019] If the similarity between target view image frames meets the preset similarity threshold requirement, then geometric consistency screening is performed based on the corresponding cutting tooth feature points to generate a target matching set.

[0020] Based on the matching point pairs in the target matching set and the preset polar geometric constraint relationship, polar geometric constraint processing is performed to obtain the target matching point pairs;

[0021] Calculate the relative pose of the corresponding target view image frame based on the target matching point pair;

[0022] Based on the relative pose, the target matching point pairs are triangulated to generate a sparse 3D point set.

[0023] In one possible implementation, the target matching point pairs are triangulated based on their relative poses to generate a sparse 3D point set, including:

[0024] Based on the relative pose, determine the reference target view image frame and the incremental target view image frame;

[0025] Based on the reference target view image frame, the target matching point pairs corresponding to the reference target view image frame are triangulated to obtain the initial sparse three-dimensional point set.

[0026] Calculate the relative pose of the incremental target viewpoint image frame based on the initial sparse 3D point set and the incremental target viewpoint image frame;

[0027] Based on the relative pose of the incremental target view image frames, the target matching point pairs corresponding to the incremental target view image frames are triangulated to update the sparse 3D point set.

[0028] In one possible implementation, after triangulating the target matching point pairs corresponding to the incremental target viewpoint image frames according to their relative poses to update the sparse 3D point set, the method further includes:

[0029] Obtain camera intrinsic parameters;

[0030] Based on the camera intrinsic parameters, relative pose, and sparse 3D point set, the overall reprojection error is minimized to obtain optimized camera intrinsic parameters, optimized relative pose, and optimized sparse 3D point set.

[0031] In one possible implementation, a 3D reconstruction process is performed based on multi-view image data to obtain a 3D reconstruction result, including:

[0032] Obtain a preset set of candidate depth maps;

[0033] Iterate through all target viewpoint image frames, and for each target viewpoint image frame, perform the following operations:

[0034] Based on a pre-defined set of candidate depth maps, photometric consistency calculations are performed on the target viewpoint image frames to obtain the corresponding cost function.

[0035] Based on the cost function, determine the depth map corresponding to the current target viewpoint image frame;

[0036] Once the traversal is complete, a depth map set is constructed based on the depth map corresponding to the target view image frame; the depth map set includes the target view image frame and the depth map corresponding to the target view image frame.

[0037] In one possible implementation, the two-dimensional feature points include the top feature point of the cutting tooth, the bottom feature point of the cutting tooth, the center feature point of the cutting tooth, the front boundary feature point of the cutting tooth, and the rear boundary feature point of the cutting tooth.

[0038] Accordingly, based on the two-dimensional feature points and the three-dimensional reconstruction results, feature point matching processing is performed to determine the coordinates of the three-dimensional feature points, including:

[0039] Determine the depth values ​​of two-dimensional feature points based on the depth map corresponding to the target view image frame;

[0040] Obtain the pixel coordinates of two-dimensional feature points in the target view image frame;

[0041] Based on the pixel coordinates, depth values, and camera intrinsic parameters of the two-dimensional feature points, back projection processing is performed on the two-dimensional feature points to obtain the corresponding set of three-dimensional feature point coordinates.

[0042] In one possible implementation, a fusion process is performed based on the set of three-dimensional feature point coordinates to obtain the target three-dimensional feature points, including:

[0043] Based on the set of three-dimensional feature point coordinates, target matching processing is performed on the cutting teeth in image frames from different target viewpoints to establish the cutting tooth matching relationship;

[0044] Based on the cutting tooth matching relationship and the set of three-dimensional feature point coordinates, an observation set is constructed;

[0045] Based on the observation set, outlier removal is performed to obtain the target observation point set;

[0046] Based on the target observation point set, mean fusion processing is performed to obtain the target's three-dimensional feature points.

[0047] Secondly, embodiments of this application provide a device for extracting drill bit blade parameters, comprising:

[0048] The acquisition module is used to acquire multi-view image data of the drill bit blades to be detected;

[0049] The detection module is used to perform cutting tooth detection processing based on multi-view image data to obtain two-dimensional feature points;

[0050] The reconstruction module is used to perform 3D reconstruction processing based on multi-view image data to obtain 3D reconstruction results; the 3D reconstruction results include sparse 3D point sets and depth map sets;

[0051] The matching module is used to perform feature point matching processing based on two-dimensional feature points and three-dimensional reconstruction results, and to determine the set of three-dimensional feature point coordinates.

[0052] The fusion module is used to perform fusion processing based on the set of 3D feature point coordinates to obtain the target 3D feature points;

[0053] The building module is used to construct the drill bit blade curve model based on the target's three-dimensional feature points;

[0054] The extraction module is used to extract the parameters of the drill bit blade to be detected based on the drill bit blade curve model.

[0055] In one possible implementation, the reconstruction module can also be used for:

[0056] Based on multi-view image data, the number of cutting teeth is statistically processed to determine the baseline detection quantity;

[0057] Based on the baseline detection quantity, multi-view image data is filtered to determine the target view image frame; where the target view image frame refers to multi-view image data with the same number of cutting teeth as the baseline detection quantity.

[0058] For each target view image frame, cutting tooth identification processing is performed to obtain the corresponding cutting tooth feature points;

[0059] Calculate the similarity between target view image frames based on the corresponding cutting tooth feature points;

[0060] If the similarity between target view image frames meets the preset similarity threshold requirement, then geometric consistency screening is performed based on the corresponding cutting tooth feature points to generate a target matching set.

[0061] Based on the matching point pairs in the target matching set and the preset polar geometric constraint relationship, polar geometric constraint processing is performed to obtain the target matching point pairs;

[0062] Calculate the relative pose of the corresponding target view image frame based on the target matching point pair;

[0063] Based on the relative pose, the target matching point pairs are triangulated to generate a sparse 3D point set.

[0064] In one possible implementation, the reconstruction module can also be used for:

[0065] Based on the relative pose, determine the reference target view image frame and the incremental target view image frame;

[0066] Based on the reference target view image frame, the target matching point pairs corresponding to the reference target view image frame are triangulated to obtain the initial sparse three-dimensional point set.

[0067] Calculate the relative pose of the incremental target viewpoint image frame based on the initial sparse 3D point set and the incremental target viewpoint image frame;

[0068] Based on the relative pose of the incremental target view image frames, the target matching point pairs corresponding to the incremental target view image frames are triangulated to update the sparse 3D point set.

[0069] In one possible implementation, the reconstruction module can also be used for:

[0070] Obtain camera intrinsic parameters;

[0071] Based on the camera intrinsic parameters, relative pose, and sparse 3D point set, the overall reprojection error is minimized to obtain optimized camera intrinsic parameters, optimized relative pose, and optimized sparse 3D point set.

[0072] In one possible implementation, the reconstruction module can also be used for:

[0073] Obtain a preset set of candidate depth maps;

[0074] Iterate through all target viewpoint image frames, and for each target viewpoint image frame, perform the following operations:

[0075] Based on a pre-defined set of candidate depth maps, photometric consistency calculations are performed on the target viewpoint image frames to obtain the corresponding cost function.

[0076] Based on the cost function, determine the depth map corresponding to the current target viewpoint image frame;

[0077] Once the traversal is complete, a depth map set is constructed based on the depth map corresponding to the target view image frame; the depth map set includes the target view image frame and the depth map corresponding to the target view image frame.

[0078] In one possible implementation, the two-dimensional feature points include the top feature point of the cutting tooth, the bottom feature point of the cutting tooth, the center feature point of the cutting tooth, the front boundary feature point of the cutting tooth, and the rear boundary feature point of the cutting tooth.

[0079] Accordingly, the matching module can also be used for:

[0080] Determine the depth values ​​of two-dimensional feature points based on the depth map corresponding to the target view image frame;

[0081] Obtain the pixel coordinates of two-dimensional feature points in the target view image frame;

[0082] Based on the pixel coordinates, depth values, and camera intrinsic parameters of the two-dimensional feature points, back projection processing is performed on the two-dimensional feature points to obtain the corresponding set of three-dimensional feature point coordinates.

[0083] In one possible implementation, the fusion module can also be used for:

[0084] Based on the set of three-dimensional feature point coordinates, target matching processing is performed on the cutting teeth in image frames from different target viewpoints to establish the cutting tooth matching relationship;

[0085] Based on the cutting tooth matching relationship and the set of three-dimensional feature point coordinates, an observation set is constructed;

[0086] Based on the observation set, outlier removal is performed to obtain the target observation point set;

[0087] Based on the target observation point set, mean fusion processing is performed to obtain the target's three-dimensional feature points.

[0088] Thirdly, embodiments of this application provide an electronic device, including: a memory and a processor;

[0089] The memory stores the instructions that the computer executes;

[0090] The processor executes computer execution instructions stored in memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect as described above.

[0091] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible implementations of the first aspect.

[0092] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the first aspect and / or various possible implementations of the first aspect.

[0093] The method, apparatus, device, storage medium, and program product for extracting drill bit blade parameters provided in this application acquire multi-view image data of the drill bit blade to be detected; perform cutting tooth detection processing based on the multi-view image data to obtain two-dimensional feature points; perform three-dimensional reconstruction processing based on the multi-view image data to obtain a three-dimensional reconstruction result; wherein, the three-dimensional reconstruction result includes a sparse three-dimensional point set and a depth map set; perform feature point matching processing based on the two-dimensional feature points and the three-dimensional reconstruction result to determine a set of three-dimensional feature point coordinates; perform fusion processing based on the set of three-dimensional feature point coordinates to obtain target three-dimensional feature points; construct a drill bit blade curve model based on the target three-dimensional feature points; and extract the parameters of the drill bit blade to be detected based on the drill bit blade curve model. Compared with the prior art, the method of this application uses cutting teeth as the basic unit, constructs spatial geometric relationships through multi-view image data, and utilizes the fusion of two-dimensional feature points and three-dimensional reconstruction results to solve the problems of poor stability of multi-view image data fusion and insufficient accuracy due to occlusion. Attached Figure Description

[0094] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0095] Figure 1 A schematic diagram of a system architecture for extracting drill bit blade parameters provided in this application;

[0096] Figure 2A flowchart illustrating a method for extracting drill bit blade parameters provided in this application. Figure 1 ;

[0097] Figure 3 A schematic diagram of the two-dimensional feature points provided in this application;

[0098] Figure 4 A schematic diagram of the drill bit blade curve model provided in this application;

[0099] Figure 5 A flowchart illustrating a method for extracting drill bit blade parameters provided in this application. Figure 2 ;

[0100] Figure 6 The depth map corresponding to the current target viewpoint image frame provided in this application;

[0101] Figure 7 A schematic diagram of the dense point cloud provided for this application;

[0102] Figure 8 Flowchart of the method for extracting drill bit blade parameters provided in this application Figure 3 ;

[0103] Figure 9 A schematic diagram illustrating the matching relationship of the cutting teeth provided in this application;

[0104] Figure 10 A schematic diagram of the target three-dimensional feature points provided in this application;

[0105] Figure 11 A schematic diagram of the target observation set provided in this application;

[0106] Figure 12 A schematic diagram of the structure of the drill bit blade parameter extraction device provided in this application;

[0107] Figure 13 A schematic diagram of the structure of the electronic device provided in this application.

[0108] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0109] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0110] It should be noted that all data involved in this application are information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant laws, regulations and standards, and corresponding operation portals are provided for users to choose to authorize or refuse.

[0111] PDC drill bits are widely used in oil and gas drilling engineering due to their high rock-breaking efficiency and strong adaptability. PDC drill bits typically consist of multiple blades, and the spatial distribution, arrangement curves, and key geometric parameters of the composite blades on each blade directly affect the drill bit's cutting performance, stress state, and service life. Therefore, accurately obtaining the three-dimensional geometric structure and design parameters of the PDC drill bit blades is a crucial foundation for PDC drill bit design optimization, manufacturing consistency assessment, and service condition analysis.

[0112] In existing technologies, the acquisition of PDC drill bit cutter blade parameters mainly relies on manual measurement, contact measurement, two-dimensional image-based analysis methods, three-dimensional scanning equipment, or multi-view reconstruction technology. Manual measurement refers to manually measuring the drill bit using tools such as calipers and micrometers. Contact measurement refers to obtaining the position and geometric parameters of the cutting teeth using contact equipment such as coordinate measuring machines. Two-dimensional image-based analysis methods involve identifying the position and contour of the cutting teeth through single-view images and deriving design parameters based on the two-dimensional image features. Multi-view reconstruction technology involves reconstructing the three-dimensional structure of the PDC drill bit cutter blades from multiple perspective images.

[0113] However, manual or contact-based measurements are cumbersome, inefficient, and difficult to adapt to complex structures such as large obstructed areas in the cutter blade region, resulting in large measurement errors and poor repeatability. Two-dimensional image-based analysis methods cannot reflect the true three-dimensional spatial relationships of the cutting teeth; measurement results are easily affected by shooting angle, lighting conditions, and image noise, making it difficult to meet high-precision design requirements. Three-dimensional scanning equipment is expensive, complex to operate and maintain, and requires a high level of expertise from both the user environment and operators, making widespread deployment in routine inspection scenarios or field operations difficult. Multi-view reconstruction technology is susceptible to differences in viewpoints, obstruction, and noise interference during multi-view data fusion, leading to unstable positioning of key structures; furthermore, it lacks parametric modeling methods for cutter blade design parameters, making it difficult to directly extract three-dimensional parameters suitable for engineering design. This results in insufficient accuracy of drill bit cutter blade parameters, failing to meet the actual needs of drill bit design, manufacturing, and service analysis.

[0114] To address the aforementioned issues, the core concept of this application is to: use cutting teeth as the basic unit, construct spatial geometric relationships through multi-view image data, and utilize the fusion of two-dimensional feature points and three-dimensional reconstruction results to solve the problems of strong equipment dependence, indirect parameter extraction, and poor stability of multi-view data fusion in the prior art, thereby extracting accurate three-dimensional design parameters for drill bit blades to meet the engineering needs of drill bit design, manufacturing, and service analysis.

[0115] Optionally, Figure 1 This is a schematic diagram of a system architecture for extracting drill bit blade parameters provided in this application. Figure 1 As shown, the system architecture for extracting drill bit blade parameters includes at least one of a data acquisition device 101, a processing device 102, and a display device 103.

[0116] It is understood that the structures illustrated in the embodiments of this application do not constitute a specific limitation on the above architecture. In other feasible embodiments of this application, the above architecture may include more or fewer components than illustrated, or combine some components, or split some components, or arrange different components, which can be determined according to the actual application scenario and is not limited here. Figure 1 The components shown can be implemented in hardware, software, or a combination of both.

[0117] In the specific implementation process, the data acquisition device 101 may include an input / output interface or a communication interface. The data acquisition device 101 can be connected to the processing device through the input / output interface or the communication interface to acquire multi-view image data of the drill bit blade to be detected.

[0118] The processing device 102 can be used to perform cutting tooth detection processing based on multi-view image data to obtain two-dimensional feature points; perform three-dimensional reconstruction processing based on multi-view image data to obtain three-dimensional reconstruction results; wherein, the three-dimensional reconstruction results include relative pose, sparse three-dimensional point set, depth map set, and multi-view spatial relationship model; perform feature point matching processing based on two-dimensional feature points and three-dimensional reconstruction results to determine the three-dimensional feature point coordinate set; perform fusion processing based on the three-dimensional feature point coordinate set to obtain target three-dimensional feature points; construct a drill bit wing curve model based on the target three-dimensional feature points; and extract the parameters of the drill bit wing to be detected based on the drill bit wing curve model.

[0119] The display device 103 can also be a touch screen or the screen of a terminal device, used to receive user commands while displaying the above-mentioned content, so as to realize interaction with the user.

[0120] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0121] Figure 2 A flowchart illustrating a method for extracting drill bit blade parameters provided in this application. Figure 1 ,like Figure 2 As shown, the method includes:

[0122] S201. Acquire multi-view image data of the drill bit blade to be detected.

[0123] In this embodiment, the camera continuously captures video of the PDC drill bit blades to be detected, and extracts image frames from the video according to a preset event interval. Since there is relative motion between the PDC drill bit blades and the camera, different image frames correspond to different viewing angles, thus obtaining multi-view image data of the drill bit blades to be detected.

[0124] S202. Based on the multi-view image data, perform cutting tooth detection processing to obtain two-dimensional feature points.

[0125] In some embodiments, the two-dimensional feature points include the top feature point of the cutting tooth, the bottom feature point of the cutting tooth, the center feature point of the cutting tooth, the front boundary feature point of the cutting tooth, and the rear boundary feature point of the cutting tooth.

[0126] Among them, such as Figure 3As shown, the top feature point of the cutting tooth refers to the position closest to the drill bit's outer diameter in the radial direction along the outer contour of the cutting tooth, used to characterize the highest point of the cutting tooth in the radial direction. The bottom feature point of the cutting tooth refers to the position closest to the drill bit's axis in the radial direction along the outer contour of the cutting tooth, used to characterize the lowest point of the cutting tooth in the radial direction. The center feature point of the cutting tooth refers to the region inside the cutting tooth contour, used to characterize the overall spatial position of the cutting tooth, and can be determined by the cutting tooth contour or the distribution of key points. The front boundary feature point of the cutting tooth refers to the boundary position on one side of the outer contour of the cutting tooth along the cutting direction, used to characterize the leading edge position of the cutting tooth in the cutting direction. The rear boundary feature point of the cutting tooth refers to the boundary position on the opposite side of the outer contour of the cutting tooth along the cutting direction, used to characterize the trailing edge position of the cutting tooth in the cutting direction.

[0127] By extracting two-dimensional feature points, the spatial position, contour shape, and orientation information of the cutting teeth in the image plane are characterized, effectively improving the accuracy of the cutting tooth parameters.

[0128] In this embodiment, for each image frame in the multi-view image data, a pre-trained detection model is used to perform cutting tooth detection processing to obtain two-dimensional feature points.

[0129] For example, the pre-trained detection model can be a YOLOv8s-pose (keypoint detection network) model.

[0130] If the multi-view image data I = {I1, I2, ..., I...} n}, for each viewpoint image data I k The YOLOv8s-pose model is used for cutting tooth detection, and two-dimensional feature points are output. Where I is the multi-view image data identifier, n is the total number of multi-view image data, k is the multi-view image data sequence number, P is the two-dimensional feature point identifier, and j is the cutting tooth sequence number. This refers to the feature point at the tip of the cutting tooth. These are feature points at the bottom of the cutting teeth. The central feature point of the cutting tooth. These are the feature points on the front boundary of the cutting teeth. These are the feature points on the rear boundary of the cutting tooth.

[0131] S203. Based on the multi-view image data, perform three-dimensional reconstruction processing to obtain the three-dimensional reconstruction result; wherein, the three-dimensional reconstruction result includes a sparse three-dimensional point set and a depth map set.

[0132] In this embodiment, the number of cutting teeth is counted and abnormal frames are filtered out from the multi-view image data to obtain the target view image frame; through cutting tooth feature point matching, polar geometry constraint, relative pose solution, triangulation processing and parameter optimization processing, a sparse three-dimensional point set and a depth map set are obtained, thereby completing the three-dimensional reconstruction and obtaining the three-dimensional reconstruction result.

[0133] S204. Based on the two-dimensional feature points and the three-dimensional reconstruction results, perform feature point matching processing to determine the set of three-dimensional feature point coordinates.

[0134] In this embodiment, two-dimensional feature points of the cutting teeth are extracted based on the target view image frame. Combined with the relative pose, camera intrinsic parameters and depth map set obtained by three-dimensional reconstruction, the two-dimensional feature points are back-projected to obtain the three-dimensional feature point coordinate set corresponding to each view.

[0135] S205. Based on the set of three-dimensional feature point coordinates, perform fusion processing to obtain the target three-dimensional feature points.

[0136] In this embodiment, outlier removal and mean fusion are performed on the set of three-dimensional feature point coordinates of the same two-dimensional feature point of the same cutting tooth under multiple perspectives to obtain stable and accurate target three-dimensional feature points.

[0137] S206. Construct a drill bit blade curve model based on the target's three-dimensional feature points.

[0138] In this embodiment, the observation radius of cutting tooth a is calculated based on the target three-dimensional feature points; wherein, the calculation formula for the observation radius of cutting tooth a is as follows:

[0139]

[0140] In the formula, r a Let be the observation radius of cutting tooth a, and S be the index of the target three-dimensional feature point set excluding the feature point at the top of the cutting tooth, where S = {1, 2, 3, 4}, representing the indices of the target three-dimensional feature points corresponding to the feature points at the bottom, center, front, and rear boundaries of the cutting tooth, respectively. Let l be the target three-dimensional feature point of cutting tooth a. These are the target 3D feature points corresponding to the feature points at the top of the toothed teeth.

[0141] Calculate the average of the observed radii of all cutting teeth as the total average radius; the formula for calculating the total average radius is as follows:

[0142]

[0143] In the formula, The total average radius is given by N, the number of cutting teeth is given by a, and the cutting tooth identifier is given by r.a Let be the observation radius of cutting tooth a.

[0144] Based on the known composite blade diameter and total average radius of the PDC drill bit blade, the scaling factor is calculated to achieve true scale restoration of the drill bit blade curve model.

[0145] In some embodiments, the scaling factor can also be calculated by statistically analyzing the spacing between target three-dimensional feature points, using known reference lengths, or calculating the dimensions of external calibration components, in order to achieve true scale restoration of the drill bit blade curve model.

[0146] In this embodiment, target three-dimensional feature points corresponding to the feature points at the top of all cutting teeth are extracted to form a representative point set of the tooth row. .

[0147] During curve fitting, it is necessary to obtain a representative set of points for the tooth row. Establish a sequence consistent with the actual tooth row orientation of the cutting teeth. Since the output sequence of the cutting teeth detection is unstable and its spatial distribution may be curved, a sorting method based on endpoint determination and diameter-preserving tooth constraints is adopted to determine the true tooth row endpoints and sorting direction, including:

[0148] In the set of representative points of the tooth row In the calculation, the Euclidean distance between representative points of each tooth row is calculated, and the two representative points with the largest distance are selected as the starting point and ending point of the tooth row, denoted as q. a With q b .

[0149] Based on the preset number of diameter-maintaining teeth, respectively q a With q b Starting from the point, select the points that are spatially closest to the current tooth row representative point and whose directions are continuous, forming a candidate point set R containing n points. a and R b .

[0150] For the candidate point set R respectively a and R b Perform a straight line fit and calculate the line fit with the candidate point set R. a and R b The corresponding fitting error; where the fitting error can be represented by the average distance or mean square distance from the representative points of the tooth row in the candidate point set to the fitted line.

[0151] Since the diameter-maintaining teeth are usually approximately collinearly distributed in space, the set of candidate points with the smallest fitting error is determined as the true diameter-maintaining tooth set, and the representative point of the tooth row corresponding to the diameter-maintaining tooth set is determined as the diameter-maintaining tooth endpoint of the tooth row.

[0152] After determining the endpoint of the diameter-maintaining tooth, this endpoint is used as the end point for tooth row sorting. The remaining points are then sorted sequentially along the tooth row direction to obtain an ordered sequence of points consistent with the physical tooth row direction. .

[0153] For an ordered point sequence Q ord Curve fitting is performed to characterize the spatial distribution of the tooth rows, including:

[0154] For the diameter-preserving tooth region, a straight line segment is fitted; for the non-diameter-preserving tooth region, a smooth curve segment is fitted.

[0155] Among them, fitting straight line segments to the diameter-preserving tooth region refers to fitting straight lines to the diameter-preserving tooth set to generate corresponding straight line segment point sequences; fitting smooth curve segments to the non-diameter-preserving tooth region refers to fitting the point sequence of the non-diameter-preserving tooth part using spline interpolation. The straight line segments and smooth curve segments are connected at the boundary to ensure the continuity and visualization consistency of the overall tooth row curve.

[0156] In some embodiments, to improve the stability of three-dimensional curve fitting during the fitting of smooth curve segments, cumulative arc length parameterization is used for fitting, as shown below:

[0157]

[0158] In the formula, s is the arc length parameter, s m This is the corresponding spatial distribution curve of the tooth row, where s1 is the initial arc length parameter with a value of 0. For an ordered point sequence Q ord The coordinate vector of the point represented by the t-th tooth row; For an ordered point sequence Q ord The coordinate vector of the point represented by the (t-1)th tooth row; m is the ordered point number, and M is the number of cutting teeth.

[0159] Based on the scaling factor and the tooth row spatial distribution curve, construct as follows: Figure 4 The drill bit blade curve model shown; where, Figure 4 The red dots in the image represent the tooth rows of the smooth curve segment. Figure 4 The green squares represent the diameter-preserving tooth points, the blue line segments represent the fitted smooth curve segments, and the green line segments represent the fitted straight line segments. X, Y, and Z represent the world coordinate system, and the unit is mm. Figure 4 The numbers in the table are ordered point numbers.

[0160] S207. Extract the parameters of the drill bit blade to be detected based on the drill bit blade curve model.

[0161] In this embodiment, the three-dimensional design parameters of the PDC drill bit blade are extracted based on the drill bit blade curve model to meet the engineering requirements of drill bit design, manufacturing, and service analysis.

[0162] The drill bit blade parameter extraction method provided in this application, through continuous video acquisition of PDC drill bit blades and construction of multi-view visual data from the video sequence, achieves 3D structural reconstruction and design parameter extraction of PDC drill bit blade cutting teeth without the need for dedicated 3D scanning equipment. This significantly reduces reliance on high-cost 3D scanners and improves the system's engineering applicability and deployment flexibility. Using cutting teeth as the basic analysis unit, spatial geometric relationships are constructed through multi-view image data. By fusing 2D feature points with 3D reconstruction results, the method solves the problems of poor stability in multi-view data fusion and insufficient accuracy due to occlusion in existing technologies, thus avoiding... The problem of difficulty in engineering parameterization by relying solely on the overall surface point cloud is addressed. By using the known structural dimensions of the cutting teeth as a scale constraint, the 3D reconstruction results are normalized, realizing the conversion from scale-free 3D reconstruction results to the actual physical dimensions. This can be directly used for the calculation and comparative analysis of blade design parameters. In the process of tooth row spatial modeling, by combining the prior structural information of the number of diameter-preserving teeth, and through endpoint determination and sorting direction selection, tooth row point sorting and curve fitting consistent with the actual blade structure are achieved. This can accurately describe the spatial distribution characteristics of the cutting teeth along the blade direction, avoiding the problem of order errors that are prone to occur under curved structures or local noise conditions in traditional point cloud sorting methods.

[0163] In some embodiments, the method for extracting drill bit blade parameters provided in this application is applicable to the extraction of parameters of a single drill bit blade, and can also be extended to PDC drill bit blades of different specifications and structural forms. It has good versatility and promotional value in drill bit design, structural evaluation and manufacturing optimization scenarios.

[0164] Figure 5 Flowchart of the method for extracting drill bit blade parameters provided in this application Figure 2 ,like Figure 5 As shown, in this embodiment... Figure 2 Based on the embodiments, the method of performing three-dimensional reconstruction processing on multi-view image data to obtain three-dimensional reconstruction results in step S203 above is described in detail. The method includes:

[0165] S501. Based on the multi-view image data, perform statistical processing on the number of cutting teeth to determine the baseline detection quantity.

[0166] In this embodiment, the number of cutting teeth corresponding to each image frame in the multi-view image data is counted to obtain the distribution of the number of cutting teeth in each image frame, and the number of cutting teeth that appears most frequently is selected as the benchmark detection number.

[0167] S502. Based on the number of baseline detections, filter multi-view image data and determine the target view image frame; wherein, the target view image frame refers to multi-view image data with the same number of cutting teeth as the number of baseline detections.

[0168] In this embodiment, if the number of cutting teeth in an image frame of the multi-view image data is equal to the number of baseline detections, then the image frame is used as the target view image data; if the number of cutting teeth in an image frame of the multi-view image data is not equal to the number of baseline detections, then the image frame is used as an abnormal frame and is discarded.

[0169] By using a benchmark detection count, multi-view image data is filtered to determine the target view image frame, reducing interference caused by abnormal viewpoints or unstable detection, and improving the reliability of subsequent relative pose estimation and 3D reconstruction results.

[0170] S503. For each target view image frame, perform cutting tooth recognition processing to obtain the corresponding cutting tooth feature points.

[0171] In this embodiment, for example, by using the SIFT (Scale-Invariant Feature Transform) algorithm, cutting tooth recognition processing is performed on each target viewpoint image frame to obtain the corresponding cutting tooth feature points.

[0172] S504. Calculate the similarity between target view image frames based on the corresponding cutting tooth feature points.

[0173] In this embodiment, for example, for target view image frame I a and target viewpoint image frame I b Based on the feature points of the cutting teeth, calculate the target view image frame I. a and target viewpoint image frame I b The similarity between them.

[0174] S505. If the similarity between target view image frames meets the preset similarity threshold requirement, then geometric consistency screening is performed based on the corresponding cutting tooth feature points to generate a target matching set.

[0175] In this embodiment, if the similarity between target view image frames meets the preset similarity threshold requirement, the cutting tooth feature points corresponding to the target view image frames are stored as matching point pairs in the initial matching set; if the similarity between target view image frames does not meet the preset similarity threshold requirement, the matching point pair is removed.

[0176] Geometric consistency screening is performed on the matching point pairs in the initial matching set to remove matching point pairs that do not meet the multi-view geometric constraints, thereby obtaining the target matching point pairs and forming a matching relationship graph covering multiple views.

[0177] S506. Based on the matching point pairs in the target matching set and the preset polar geometric constraint relationship, perform polar geometric constraint processing to obtain the target matching point pairs.

[0178] In this embodiment, the preset polar geometric constraint relationship is shown in the following formula:

[0179]

[0180] In the formula, x a x b Target view image frame I a and target viewpoint image frame I b Matching point pairs (x a x b ); E a，b For target view image frame I a and target viewpoint image frame I b The essential matrix between them.

[0181] If the matching point pairs in the target matching set satisfy the preset polar geometric constraint relationship, then the matching point pairs in the target matching set will be used as the target matching point pairs.

[0182] S507. Calculate the relative pose of the corresponding target viewpoint image frame based on the target matching point pair.

[0183] In this embodiment, based on the target matching point pairs that satisfy the polar geometric constraints, the rotation matrix and translation vector between the corresponding target view image frames are calculated based on the essential matrix decomposition to obtain the relative pose of the corresponding target view image frames.

[0184] S508. Based on the relative pose, triangulate the target matching point pairs to generate a sparse three-dimensional point set.

[0185] In some embodiments, the target matching point pairs are triangulated according to their relative poses to generate a sparse 3D point set, including:

[0186] S5081. Based on the relative pose, determine the reference target view image frame and the incremental target view image frame.

[0187] In this embodiment, based on the relative pose, the two target view image frames with the largest overlapping area and the most matching points are determined as the reference target view image frame and the first incremental target view image frame. For target view image frames other than the reference target view image frame and the first incremental target view image frame, they are used as subsequent incremental target view image frames in the order of view progression.

[0188] S5082. Based on the reference target view image frame, triangulate the target matching point pairs corresponding to the reference target view image frame to obtain an initial sparse three-dimensional point set.

[0189] In this embodiment, using the reference target view image frame as a reference, target matching point pairs between the reference target view image frame and the first incremental target view image frame are extracted. Through the triangulation algorithm, the target matching point pairs between the reference target view image frame and the first incremental target view image frame are triangulated to obtain the initial sparse three-dimensional point set.

[0190] S5083. Calculate the relative pose of the incremental target viewpoint image frame based on the initial sparse 3D point set and the incremental target viewpoint image frame.

[0191] In this embodiment, the initial sparse 3D point set is taken as the existing 3D structure. By using the constraint relationship between the incremental target view image frame and the existing 3D structure, the PnP (Perspective-n-Point) algorithm is used to calculate the projection constraint between the initial sparse 3D point set and the target matching point in the next incremental target view image frame, and solve the relative pose of the next incremental target view image frame.

[0192] S5084. Based on the relative pose of the incremental target view image frame, the target matching point pairs corresponding to the incremental target view image frame are triangulated to update the sparse three-dimensional point set.

[0193] In this embodiment, a triangulation algorithm is used to perform triangulation calculations on the target matching point pairs between the incremental target view image frame and the existing three-dimensional structure in order to update the sparse three-dimensional point set.

[0194] In some embodiments, after triangulating the target matching point pairs corresponding to the incremental target view image frames according to their relative poses to update the sparse 3D point set, the method further includes:

[0195] S5085, Obtain camera internal parameters.

[0196] S5086. Based on the camera intrinsic parameters, relative pose, and sparse 3D point set, perform overall reprojection error minimization processing to obtain optimized camera intrinsic parameters, optimized relative pose, and optimized sparse 3D point set.

[0197] In this embodiment, the overall reprojection error minimization process is as follows:

[0198]

[0199] In the formula, k is the frame number of the target view image, K k For camera intrinsic parameters, R kt k For relative pose, i is the 3D point index, Xi is a 3D point in the sparse 3D point set, and X... k，i Let Xi be the observed image point of the three-dimensional point in the k-th frame of the target view image. Let be the set of observable 3D point indices in the k-th target viewpoint image frame. This is the projection function.

[0200] In this embodiment, steps S5083 to S5086 are repeated until all incremental target view image frames are processed, and a multi-view spatial relationship model is established based on the optimized camera intrinsic parameters, optimized relative pose, and optimized sparse 3D point set.

[0201] In some embodiments, performing three-dimensional reconstruction processing based on multi-view image data to obtain a three-dimensional reconstruction result further includes:

[0202] Obtain a preset set of candidate depth maps.

[0203] In this embodiment, the preset candidate depth map set includes a set of discrete depth values ​​sampled at a fixed step size, or an initial depth map generated based on a sparse 3D point set.

[0204] Iterate through all target viewpoint image frames, and for each target viewpoint image frame, perform the following operations:

[0205] Based on a pre-defined set of candidate depth maps, photometric consistency calculations are performed on the current target viewpoint image frame to obtain the corresponding cost function.

[0206] In this embodiment, the set of auxiliary viewpoint image frames corresponding to the target viewpoint image frames is determined according to the progressive order of the viewpoints of the target viewpoint image frames or according to the order of descending order of the number of target matching point pairs.

[0207] Based on the camera intrinsic parameters and relative pose of the target view image frame, the pixel u in the target view image frame is back-projected into three-dimensional space at depth d to obtain the three-dimensional point X; the three-dimensional point X is projected into the set of all auxiliary view image frames to obtain the corresponding pixel coordinates; the photometric difference between the target view image frame and each auxiliary view image frame in the set of auxiliary view image frames is calculated, and the photometric differences of all auxiliary view image frames are summed to obtain the cost function C(u, d) under the depth d assumption.

[0208] Based on the cost function, determine the depth map corresponding to the current target viewpoint image frame.

[0209] In this embodiment, the depth map corresponding to the current target viewpoint image frame is determined according to the cost function as follows:

[0210]

[0211] In the formula, r is the target's viewpoint, and d r (u) is the depth map corresponding to the pixel with pixel coordinate u under the target viewpoint r; u is the pixel coordinate on the image frame of the target viewpoint; mind∈D(u) is the function that finds the depth value d in the preset candidate depth map set D(u) that makes the cost function C(u,d) reach the minimum value; D(u) is the candidate depth map set corresponding to pixel u.

[0212] Furthermore, in the process of determining the depth map corresponding to the current target view image frame, neighborhood smoothing or consistency constraints are introduced, and cross-view consistency checks are performed on the depth map to eliminate unstable or inconsistent depth maps.

[0213] For example, the depth map corresponding to the current target viewpoint image frame is as follows: Figure 6 As shown; where the colors corresponding to different depths are as follows: Figure 6 The horizontal axis represents the relative depth, which is dimensionless.

[0214] Once the traversal is complete, a depth map set is constructed based on the depth map corresponding to the target view image frame; the depth map set includes the target view image frame and the depth map corresponding to the target view image frame.

[0215] Furthermore, after constructing a depth map set based on the depth maps corresponding to the target viewpoint image frames, the depth map set is 3D fused. The depth pixels from each viewpoint are back-projected into spatial points and then merged and deduplicated to obtain the result shown below. Figure 7 The output shows the dense point cloud and its corresponding multi-view spatial relationships.

[0216] The method for extracting drill bit blade parameters provided in this application reconstructs the cutting teeth of PDC drill bit blades in 3D using multi-view image data. First, the number of cutting teeth is counted to determine the baseline detection quantity, and abnormal image frames are filtered and removed to reduce interference from viewing angles and detection, thus improving data reliability. The SIFT algorithm is used to identify and extract feature points of the cutting teeth. Similarity filtering, geometric consistency, and polar geometric constraints are used to optimize the matching relationship, improving feature matching accuracy. The relative pose of the camera is solved based on essential matrix factorization, and a sparse 3D point set is generated through incremental triangulation. The PnP algorithm and reprojection error minimization are combined to optimize camera parameters and 3D structure, ensuring the accuracy of pose and spatial structure. Then, a depth map is generated through photometric consistency calculation, and a dense point cloud is obtained through cross-view consistency verification and 3D fusion. This effectively improves the robustness, accuracy, and completeness of the 3D reconstruction of the cutting teeth, providing stable and reliable 3D data support for the extraction of PDC drill bit blade parameters.

[0217] Figure 8 Flowchart of the method for extracting drill bit blade parameters provided in this application Figure 3 ,like Figure 8 As shown, in this embodiment... Figure 5 Based on the examples, the method for extracting drill bit blade parameters is described in detail. This method includes:

[0218] S801. Determine the depth values ​​of two-dimensional feature points based on the depth map corresponding to the target view image frame.

[0219] In this embodiment, the pixel coordinates of two-dimensional feature points in the target view image frame are obtained, and the depth values ​​corresponding to the two-dimensional feature points are obtained on the depth map corresponding to the target view image frame by bilinear interpolation or neighborhood search method.

[0220] Optionally, if the depth of the interpolation location is invalid, missing, or unstable, the search radius is gradually expanded in the neighborhood to find a valid depth value until the nearest valid depth value is found, which is then used as the depth value of the two-dimensional feature point.

[0221] S802. Obtain the pixel coordinates of two-dimensional feature points in the target view image frame.

[0222] In this embodiment, for each target view image frame, the pixel coordinates u of the two-dimensional feature points of the cutting teeth detected in the target view image frame are obtained.

[0223] S803. Based on the pixel coordinates, depth values, and camera intrinsic parameters of the two-dimensional feature points, perform back projection processing on the two-dimensional feature points to obtain the corresponding set of three-dimensional feature point coordinates.

[0224] In this embodiment, based on the camera intrinsic parameters, the pixel coordinates and depth values ​​of the two-dimensional feature points are back-projected onto the camera coordinate system to obtain the corresponding three-dimensional point coordinates; wherein, the back-projection process is shown in the following formula:

[0225]

[0226] In the formula, X cam Y cam Z cam Here, u and v are the pixel coordinates of a 3D point, z is the depth value, and c is the depth value. x c y f represents the coordinates of the principal point in the camera's intrinsic parameters. x f y This refers to the focal length parameter in the camera's intrinsic parameters.

[0227] Based on the relative pose, the 3D points in the camera coordinate system are transformed to the world coordinate system to obtain the corresponding 3D feature point coordinates; the transformation of the 3D feature point coordinates is shown below:

[0228]

[0229]

[0230]

[0231] In the formula, X word Y word Z word These are the coordinates of the three-dimensional feature point; other parameters are the same as above.

[0232] Step S803 is performed on all target viewpoint image frames to obtain the set of three-dimensional feature point coordinates of the two-dimensional feature key points of each cutting tooth under different viewpoints.

[0233] S804. Based on the set of three-dimensional feature point coordinates, perform target matching processing on the cutting teeth in image frames from different target viewpoints to establish the cutting tooth matching relationship.

[0234] In this embodiment, the cutting teeth detected in the first target view image frame are used as reference targets to establish a reference cutting tooth list. , among which, T ref The reference cutting tooth is represented by N, which is the number of cutting teeth; for each reference cutting tooth in the reference cutting tooth list There are 5 two-dimensional feature points. ,in, For reference cutting teeth The coordinates of the corresponding two-dimensional feature points in three dimensions.

[0235] The cutting teeth detected in all target viewpoint image frames other than the first target viewpoint image frame are used as candidate targets to establish a candidate cutting tooth list and obtain the candidate cutting teeth. The corresponding 25-dimensional feature points .

[0236] Calculate reference cutting teeth and candidate cutting teeth The average Euclidean distance between them is given by the formula shown below:

[0237]

[0238] In the formula, D(a, b) is the average Euclidean distance; l is the index of the three-dimensional feature point, corresponding to the l-th two-dimensional feature point on the cutting tooth, and l takes a value from 1 to 5; V a，b For a valid set of keypoint indices, The coordinates of the three-dimensional key point of the l-th two-dimensional feature point of the reference cutting tooth; The coordinates of the three-dimensional key point of the l-th two-dimensional feature point of the candidate cutting tooth are given.

[0239] By calculating the average Euclidean distance between cutting teeth in image frames from different target perspectives, the reference cutting tooth and candidate cutting tooth with the smallest average Euclidean distance are selected as target pairs. If the average Euclidean distance of the target pair is greater than a preset distance threshold, no cutting tooth matching relationship is established. If the average Euclidean distance of the target pair is not greater than the preset distance threshold, the corresponding cutting tooth matching relationship is established.

[0240] For example, the matching relationship of cutting teeth is as follows: Figure 9 As shown, where, Figure 9 The red dots in the image represent two-dimensional feature points of the reference cutting tooth. Figure 9 The triangular points in the diagram represent the two-dimensional feature points of the candidate cutting teeth. X, Y, and Z represent the world coordinate system, with units of mm.

[0241] S805. Construct an observation set based on the cutting tooth matching relationship and the set of three-dimensional feature point coordinates.

[0242] In this embodiment, based on the cutting tooth matching relationship, the three-dimensional feature point coordinates of the same two-dimensional feature point of the same cutting tooth under different target view frames are obtained from the three-dimensional feature point coordinate set to construct an observation set of the same two-dimensional feature point; wherein, the observation set is as follows:

[0243]

[0244] In the formula, O a，l Let be the observation set of the l-th two-dimensional feature point of cutting tooth a, and let represent the set of three-dimensional feature point coordinates observed in all target view image frames; each element Represents a single three-dimensional observation R 3 .

[0245] S806. Based on the observation set, outlier removal is performed to obtain the target observation point set.

[0246] In this embodiment, a random sampling method is used to select observation point samples from the observation set and calculate the mean of the observation point samples as the center position of the observation point samples. The distance from all observation points in the observation set to the center position is calculated, and observation points whose distance to the center position is less than an adaptive threshold are determined as interior points. The steps of random sampling, distance calculation from observation points to the center position and interior point determination are repeated iteratively. The observation set with the most interior points is selected as the target observation set.

[0247] Furthermore, the adaptive threshold can be adaptively determined based on the distribution of distances between observation points; for example, it can be set proportionally to the median distance between observation points.

[0248] S807. Based on the target observation point set, perform mean fusion processing to obtain the target's three-dimensional feature points.

[0249] In this embodiment, the mean fusion process is as follows:

[0250]

[0251] In the formula, Let l be the final three-dimensional position of the l-th two-dimensional feature point of cutting tooth a. This represents the number of interior points in the target observation set; other parameters are the same as above.

[0252] For example, such as Figure 10 As shown, where, Figure 10 The circles in the diagram represent interior points, the crosses represent outliers, and the star-shaped points represent the average value of the interior points (i.e., the final 3D position). X, Y, and Z represent the world coordinate system, and the unit is mm.

[0253] By taking the arithmetic mean of the coordinates of all three-dimensional feature points in the target observation set, the final three-dimensional position of the two-dimensional feature point is calculated, and the two-dimensional feature point corresponding to the final three-dimensional position is used as the target three-dimensional feature point; through multi-view fusion, the observation error of a single viewpoint is effectively offset.

[0254] For example, the target observation set is as follows Figure 11 As shown, X, Y, and Z represent the world coordinate system, with units in mm.

[0255] The drill bit blade parameter extraction method provided in this application obtains the pixel coordinates and depth values ​​of two-dimensional feature points, combines them with camera intrinsic parameters for backprojection to obtain the coordinates of three-dimensional feature points, and transforms them to the world coordinate system, laying the foundation for cutting tooth matching. By calculating the average Euclidean distance between the reference cutting tooth and the candidate cutting tooth, a precise cutting tooth matching relationship is established, ensuring the correspondence of cutting teeth from multiple perspectives. An observation set is constructed based on the matching relationship, and outliers are iteratively removed through random sampling to obtain the target observation set. Then, the target three-dimensional feature points are obtained through mean fusion. This effectively offsets the observation error from a single perspective, improves the accuracy of three-dimensional feature point positioning, ensures the accuracy and stability of the three-dimensional feature information of the cutting teeth, and provides reliable support for subsequent three-dimensional reconstruction and parameter detection of the cutting teeth.

[0256] Figure 12 A schematic diagram of the structure of the drill bit blade parameter extraction device provided in this application is shown below. Figure 12 As shown, the drill bit blade parameter extraction device provided in this embodiment includes:

[0257] The acquisition module 1201 is used to acquire multi-view image data of the drill bit blade to be detected.

[0258] The detection module 1202 is used to perform cutting tooth detection processing based on multi-view image data to obtain two-dimensional feature points.

[0259] The reconstruction module 1203 is used to perform three-dimensional reconstruction processing based on multi-view image data to obtain three-dimensional reconstruction results; wherein, the three-dimensional reconstruction results include a sparse three-dimensional point set and a depth map set.

[0260] The matching module 1204 is used to perform feature point matching processing based on the two-dimensional feature points and the three-dimensional reconstruction results to determine the set of three-dimensional feature point coordinates.

[0261] The fusion module 1205 is used to perform fusion processing based on the set of three-dimensional feature point coordinates to obtain the target three-dimensional feature points.

[0262] Module 1206 is used to construct the drill bit blade curve model based on the target's three-dimensional feature points.

[0263] The extraction module 1207 is used to extract the parameters of the drill bit blade to be detected based on the drill bit blade curve model.

[0264] In one possible implementation, the reconstruction module 1203 can also be used for:

[0265] Based on multi-view image data, the number of cutting teeth is statistically processed to determine the baseline detection quantity;

[0266] Based on the baseline detection quantity, multi-view image data is filtered to determine the target view image frame; where the target view image frame refers to multi-view image data with the same number of cutting teeth as the baseline detection quantity.

[0267] For each target view image frame, cutting tooth identification processing is performed to obtain the corresponding cutting tooth feature points;

[0268] Calculate the similarity between target view image frames based on the corresponding cutting tooth feature points;

[0269] If the similarity between target view image frames meets the preset similarity threshold requirement, then geometric consistency screening is performed based on the corresponding cutting tooth feature points to generate a target matching set.

[0270] Based on the matching point pairs in the target matching set and the preset polar geometric constraint relationship, polar geometric constraint processing is performed to obtain the target matching point pairs;

[0271] Calculate the relative pose of the corresponding target view image frame based on the target matching point pair;

[0272] Based on the relative pose, the target matching point pairs are triangulated to generate a sparse 3D point set.

[0273] In one possible implementation, the reconstruction module 1203 can also be used for:

[0274] Based on the relative pose, determine the reference target view image frame and the incremental target view image frame;

[0275] Based on the reference target view image frame, the target matching point pairs corresponding to the reference target view image frame are triangulated to obtain the initial sparse three-dimensional point set.

[0276] Calculate the relative pose of the incremental target viewpoint image frame based on the initial sparse 3D point set and the incremental target viewpoint image frame;

[0277] Based on the relative pose of the incremental target view image frames, the target matching point pairs corresponding to the incremental target view image frames are triangulated to update the sparse 3D point set.

[0278] In one possible implementation, the reconstruction module 1203 can also be used for:

[0279] Obtain camera intrinsic parameters;

[0280] Based on the camera intrinsic parameters, relative pose, and sparse 3D point set, the overall reprojection error is minimized to obtain optimized camera intrinsic parameters, optimized relative pose, and optimized sparse 3D point set.

[0281] In one possible implementation, the reconstruction module 1203 can also be used for:

[0282] Obtain a preset set of candidate depth maps;

[0283] Iterate through all target viewpoint image frames, and for each target viewpoint image frame, perform the following operations:

[0284] Based on a pre-defined set of candidate depth maps, photometric consistency calculations are performed on the target viewpoint image frames to obtain the corresponding cost function.

[0285] Based on the cost function, determine the depth map corresponding to the current target viewpoint image frame;

[0286] Once the traversal is complete, a depth map set is constructed based on the depth map corresponding to the target view image frame; the depth map set includes the target view image frame and the depth map corresponding to the target view image frame.

[0287] In one possible implementation, the two-dimensional feature points include the top feature point of the cutting tooth, the bottom feature point of the cutting tooth, the center feature point of the cutting tooth, the front boundary feature point of the cutting tooth, and the rear boundary feature point of the cutting tooth.

[0288] Accordingly, the matching module 1204 can also be used for:

[0289] Determine the depth values ​​of two-dimensional feature points based on the depth map corresponding to the target view image frame;

[0290] Obtain the pixel coordinates of two-dimensional feature points in the target view image frame;

[0291] Based on the pixel coordinates, depth values, and camera intrinsic parameters of the two-dimensional feature points, back projection processing is performed on the two-dimensional feature points to obtain the corresponding set of three-dimensional feature point coordinates.

[0292] In one possible implementation, the fusion module 1205 can also be used for:

[0293] Based on the set of three-dimensional feature point coordinates, target matching processing is performed on the cutting teeth in image frames from different target viewpoints to establish the cutting tooth matching relationship;

[0294] Based on the cutting tooth matching relationship and the set of three-dimensional feature point coordinates, an observation set is constructed;

[0295] Based on the observation set, outlier removal is performed to obtain the target observation point set;

[0296] Based on the target observation point set, mean fusion processing is performed to obtain the target's three-dimensional feature points.

[0297] The drill bit blade parameter extraction device provided in this embodiment can perform the method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0298] Figure 13 A schematic diagram of the structure of the electronic device provided in this application. Figure 13 As shown, the electronic device provided in this embodiment includes at least one processor 1301 and a memory 1302. Optionally, the electronic device further includes a communication component 1303. The processor 1301, the memory 1302, and the communication component 1303 are connected via a bus 1304.

[0299] In a specific implementation, at least one processor 1301 executes computer execution instructions stored in memory 1302, causing at least one processor 1301 to perform the above-described method.

[0300] The specific implementation process of processor 1301 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0301] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0302] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0303] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0304] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0305] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0306] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0307] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0308] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0309] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0310] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0311] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0312] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0313] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A method for extracting drill bit blade parameters, characterized in that, include: Acquire multi-view image data of the drill bit blade to be inspected; Based on the multi-view image data, cutting tooth detection processing is performed to obtain two-dimensional feature points; the two-dimensional feature points include the cutting tooth top feature point, the cutting tooth bottom feature point, the cutting tooth center feature point, the cutting tooth front boundary feature point, and the cutting tooth rear boundary feature point. Based on the multi-view image data, a three-dimensional reconstruction process is performed to obtain a three-dimensional reconstruction result; wherein, the three-dimensional reconstruction result includes a sparse three-dimensional point set and a depth map set; The depth value of the two-dimensional feature point is determined based on the depth map corresponding to the target view image frame; wherein, the target view image frame refers to multi-view image data in which the number of cutting teeth is the same as the number of reference detections; the number of reference detections refers to the number of cutting teeth that appears most frequently in all image frames of the multi-view image data. Obtain the pixel coordinates of the two-dimensional feature points in the target view image frame; Based on the pixel coordinates of the two-dimensional feature points, the depth value, and the camera intrinsic parameters, the two-dimensional feature points are back-projected to obtain the corresponding set of three-dimensional feature point coordinates. Based on the set of three-dimensional feature point coordinates, target matching processing is performed on the cutting teeth in image frames from different target viewpoints to establish cutting tooth matching relationships; Based on the cutting tooth matching relationship and the set of three-dimensional feature point coordinates, an observation set is constructed; Based on the observation set, outlier removal is performed to obtain the target observation point set; Based on the target observation point set, mean fusion processing is performed to obtain the target's three-dimensional feature points; Based on the target three-dimensional feature points, construct the drill bit blade curve model; Based on the drill bit blade curve model, extract the parameters of the drill bit blade to be detected; The step of constructing a drill bit blade curve model based on the target three-dimensional feature points includes: Extract the target three-dimensional feature points corresponding to the feature points at the top of all cutting teeth to form a set of representative points of the tooth row; A sorting method based on endpoint determination and diameter-preserving tooth constraints is adopted to sort the representative point set of the tooth row, resulting in an ordered point sequence consistent with the physical tooth row direction. The sorting method includes: calculating the Euclidean distance between the representative points of each tooth row, selecting the two points with the largest distance as the starting point and ending point of the tooth row; using the starting point and ending point as benchmarks, and combining with the preset number of diameter-preserving teeth, selecting points with continuous direction to form a candidate point set; performing linear fitting on the candidate point set, selecting the set with the smallest fitting error as the real diameter-preserving tooth set, and determining its endpoint as the endpoint of the tooth row sorting, and sorting the remaining points sequentially along the tooth row direction. Curve fitting is performed on the ordered point sequence, which includes: fitting straight line segments to the points corresponding to the diameter-preserving tooth set, and using spline interpolation to fit smooth curve segments for the non-diameter-preserving tooth parts using cumulative arc length parameterization; connecting the straight line segments and the smooth curve segments to construct the drill bit blade curve model.

2. The method according to claim 1, characterized in that, The step of performing 3D reconstruction processing based on the multi-view image data to obtain the 3D reconstruction result includes: Based on the multi-view image data, the number of cutting teeth is statistically processed to determine the baseline detection quantity; Based on the benchmark detection count, the multi-view image data is filtered to determine the target view image frame; For each target view image frame, cutting tooth identification processing is performed to obtain the corresponding cutting tooth feature points; Calculate the similarity between target view image frames based on the corresponding cutting tooth feature points; If the similarity between the target view image frames meets the preset similarity threshold requirement, then geometric consistency screening is performed based on the corresponding cutting tooth feature points to generate a target matching set. Based on the matching point pairs in the target matching set and the preset polar geometric constraint relationship, polar geometric constraint processing is performed to obtain the target matching point pairs; Based on the target matching point pair, calculate the relative pose of the corresponding target view image frame; Based on the relative pose, the target matching point pairs are triangulated to generate a sparse three-dimensional point set.

3. The method of claim 2, wherein, The step of triangulating the target matching point pairs based on the relative pose to generate a sparse 3D point set includes: Based on the relative pose, determine the reference target view image frame and the incremental target view image frame; Based on the reference target view image frame, the target matching point pairs corresponding to the reference target view image frame are triangulated to obtain an initial sparse three-dimensional point set. The relative pose of the incremental target view image frame is calculated based on the initial sparse 3D point set and the incremental target view image frame. Based on the relative pose of the incremental target view image frame, the target matching point pairs corresponding to the incremental target view image frame are triangulated to update the sparse 3D point set.

4. The method of claim 3, wherein, After performing triangulation on the target matching point pairs corresponding to the incremental target view image frames based on their relative poses to update the sparse 3D point set, the method further includes: Obtain camera intrinsic parameters; Based on the camera intrinsic parameters, the relative pose, and the sparse 3D point set, the overall reprojection error is minimized to obtain optimized camera intrinsic parameters, optimized relative pose, and optimized sparse 3D point set.

5. The method of claim 4, wherein, The step of performing 3D reconstruction processing based on the multi-view image data to obtain the 3D reconstruction result includes: Obtain a preset set of candidate depth maps; Iterate through all target viewpoint image frames, and for each target viewpoint image frame, perform the following operations: Based on the preset set of candidate depth maps, the photometric consistency calculation is performed on the target viewpoint image frame to obtain the corresponding cost function; Based on the cost function, determine the depth map corresponding to the current target viewpoint image frame; After the traversal is completed, a depth map set is constructed based on the depth map corresponding to the target view image frame; wherein, the depth map set includes the target view image frame and the depth map corresponding to the target view image frame.

6. A device for extracting parameters of a drill bit blade, characterized in that include: The acquisition module is used to acquire multi-view image data of the drill bit blades to be detected; The detection module is used to perform cutting tooth detection processing based on the multi-view image data to obtain two-dimensional feature points; the two-dimensional feature points include the cutting tooth top feature point, the cutting tooth bottom feature point, the cutting tooth center feature point, the cutting tooth front boundary feature point, and the cutting tooth rear boundary feature point. The reconstruction module is used to perform three-dimensional reconstruction processing based on the multi-view image data to obtain a three-dimensional reconstruction result; wherein, the three-dimensional reconstruction result includes a sparse three-dimensional point set and a depth map set; The matching module is used to determine the depth value of the two-dimensional feature point based on the depth map corresponding to the target view image frame; wherein, the target view image frame refers to multi-view image data in which the number of cutting teeth is the same as the number of reference detections; the number of reference detections refers to the number of cutting teeth that appears most frequently in all image frames of the multi-view image data; the module obtains the pixel coordinates of the two-dimensional feature point in the target view image frame; and performs back projection processing on the two-dimensional feature point based on the pixel coordinates of the two-dimensional feature point, the depth value, and camera intrinsic parameters to obtain the corresponding three-dimensional feature point coordinate set; The fusion module is used to perform target matching processing on the cutting teeth in image frames from different target perspectives based on the set of three-dimensional feature point coordinates, so as to establish a cutting tooth matching relationship; construct an observation set based on the cutting tooth matching relationship and the set of three-dimensional feature point coordinates; perform outlier removal processing based on the observation set to obtain a target observation point set; and perform mean fusion processing based on the target observation point set to obtain the target three-dimensional feature points. A construction module is used to construct a drill bit blade curve model based on the target three-dimensional feature points; The extraction module is used to extract the parameters of the drill bit blade to be detected based on the drill bit blade curve model; The building module is specifically used for: Extract the target three-dimensional feature points corresponding to the feature points at the top of all cutting teeth to form a set of representative points of the tooth row; A sorting method based on endpoint determination and diameter-preserving tooth constraints is adopted to sort the representative point set of the tooth row, resulting in an ordered point sequence consistent with the physical tooth row direction. The sorting method includes: calculating the Euclidean distance between the representative points of each tooth row, selecting the two points with the largest distance as the starting point and ending point of the tooth row; using the starting point and ending point as benchmarks, and combining with the preset number of diameter-preserving teeth, selecting points with continuous direction to form a candidate point set; performing linear fitting on the candidate point set, selecting the set with the smallest fitting error as the real diameter-preserving tooth set, and determining its endpoint as the endpoint of the tooth row sorting, and sorting the remaining points sequentially along the tooth row direction. Curve fitting is performed on the ordered point sequence, which includes: fitting straight line segments to the points corresponding to the diameter-preserving tooth set, and using spline interpolation to fit smooth curve segments for the non-diameter-preserving tooth parts using cumulative arc length parameterization; connecting the straight line segments and the smooth curve segments to construct the drill bit blade curve model.

7. An electronic device, comprising: include: Memory, processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory, causing the processor to perform the method as described in any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1-5.

9. A computer program product comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the method as described in any one of claims 1-5.

Images