CN113469940B - Fastener loosening detection method based on three-dimensional point cloud and two-dimensional image processing technology - Google Patents
Fastener loosening detection method based on three-dimensional point cloud and two-dimensional image processing technology Download PDFInfo
- Publication number
- CN113469940B CN113469940B CN202110579074.5A CN202110579074A CN113469940B CN 113469940 B CN113469940 B CN 113469940B CN 202110579074 A CN202110579074 A CN 202110579074A CN 113469940 B CN113469940 B CN 113469940B
- Authority
- CN
- China
- Prior art keywords
- point
- fastener
- point cloud
- dimensional
- dimensional image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T19/00—Manipulating 3D models or images for computer graphics
- G06T19/20—Editing of 3D images, e.g. changing shapes or colours, aligning objects or positioning parts
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10028—Range image; Depth image; 3D point clouds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30164—Workpiece; Machine component
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2219/00—Indexing scheme for manipulating 3D models or images for computer graphics
- G06T2219/20—Indexing scheme for editing of 3D models
- G06T2219/2004—Aligning objects, relative positioning of parts
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2219/00—Indexing scheme for manipulating 3D models or images for computer graphics
- G06T2219/20—Indexing scheme for editing of 3D models
- G06T2219/2016—Rotation, translation, scaling
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Computer Graphics (AREA)
- Architecture (AREA)
- Probability & Statistics with Applications (AREA)
- Quality & Reliability (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Hardware Design (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Length Measuring Devices By Optical Means (AREA)
- Image Analysis (AREA)
Abstract
The invention belongs to the technical field of rail transit work detection and detection, and particularly relates to a fastener loosening detection method based on a three-dimensional point cloud combined with a two-dimensional image processing technology, which comprises a steel rail positioning step, a fastener key measurement point positioning step, a fastener reference point positioning step and a fastener loosening judgment step.
Description
Technical Field
The invention belongs to the technical field of overhead line system rail transit service detection, and particularly relates to a fastener loosening detection method based on a three-dimensional point cloud combined with two-dimensional image processing technology.
Background
The middle connecting part is also called a rail fastener, which is a part of a fastening device of a rail with a rigid fastener and a sleeper, and is a 70-type buckle fastener for fastening 50, 43 (kg/m) and other rails of a concrete sleeper line in China, and the middle connecting part consists of parts such as a bolt spike, a nut, a flat washer, a spring washer, a rigid buckle, an iron seat, an insulating buffer washer, an under-rail rubber cushion, an insulating buffer washer, an insulating antirust paint, a sulfur anchor and the like. Rail fasteners fasten rail fittings of rails and machine lower parts, and are classified into two main categories, wood sleeper line fasteners and concrete sleeper line (other types of concrete machine lower part line) fasteners, according to sleeper types.
Currently, in the technical scheme of the track detection device, the detection method for the loosening of the fastener comprises an angle comparison method based on a two-dimensional image, a detection method based on infrared thermal imaging by acquiring the contact stress between the fastener and a steel rail, a fastener bolt floating detection method based on a three-dimensional point cloud, and the like. Compared with the detection of fastener loosening from a two-dimensional image, the fastener loosening identification precision is higher when the fastener loosening identification precision is finished from a three-dimensional point cloud, the three-dimensional point cloud-based fastener bolt floating detection method in the prior art is to convert the three-dimensional point cloud into a two-dimensional depth image, then extract the fastener by setting a gray threshold range, and then judge whether the fastener is loosened according to the height information of the bolt, but the method can only be applied to the point cloud acquisition equipment to keep a fixed distance from the fastener without interference data, the fastener can be directly extracted by setting a threshold value through coordinate position information, and the fastener loosening is often that the nut below the bolt is loosened and does not compress an elastic strip, so that the fastener loosening cannot be accurately reflected by only detecting the fastener bolt loosening, but the identification positioning and the filtering interference information of the fastener cannot be automatically finished in the three-dimensional point cloud, and the accurate positioning of the elastic strip key measurement point and the insulation block reference point can be realized.
Compared with the method, the method has the advantages that in the three-dimensional point cloud, the identification, positioning and interference information filtering of the fasteners are automatically completed, and the accurate positioning of the spring strip key measurement points and the insulation block reference points is realized, so that the accurate measurement of fastener looseness is achieved.
Disclosure of Invention
In order to overcome the problems and the defects in the prior art, the invention aims to provide a fastener loosening detection method for automatically completing the identification, positioning and filtering of interference signals of fasteners in three-dimensional point cloud, and realizing the accurate positioning of a spring strip key measurement point and an insulating block reference point so as to achieve the accurate measurement of fastener loosening.
A fastener loosening detection method based on a three-dimensional point cloud combined with a two-dimensional image processing technology comprises the following steps:
rail positioning: setting a threshold range (z) according to the height information in the track three-dimensional point cloud acquired by the three-dimensional acquisition system railmin ～z railmax ) Filtering three-dimensional point clouds of a track through a threshold range, setting a threshold range for height information, wherein the distance between a three-dimensional acquisition device and a track surface is a priori threshold, then carrying out density clustering on the filtered three-dimensional point clouds of the track to respectively acquire coordinates of external cubes of the left and right rail point clouds, specifically, carrying out interval calculation on all three-dimensional point clouds in the three-dimensional point clouds by setting a distance judgment threshold, dividing the interval into a cluster, applying a clustering method in three dimensions, determining that the position relationship between a fastener and a rail is necessarily in the external cube coordinates of the left and right rail point clouds, namely, carrying out preliminary division on the area where the fastener is positioned, and setting the diagonal two-point coordinates of the external cube as (x) min ，y min ，z min ) And (x) max ，y max ，z max )；
And a fastener positioning step: because the fasteners are located within T distance from the rail surface on both sides of the rail surface, x min-T ～x max-T Intercepting three-dimensional point clouds containing fasteners from the external cubic coordinates of the left and right rail point clouds obtained in the rail positioning step, namely the fastener point clouds, wherein T is the distance between two sides of a rail surface and the rail surface; according to the height range (h) min ～h max ) Filtering the fastener point cloud to obtain a fastener profileThree-dimensional point cloud, i.e. fastener gathering point cloud, height range (h min ～h max ) Depending on the distance between the installed three-dimensional acquisition equipment and the fastener, the threshold range set for the height information in the rail positioning step is just a larger and rough range, and finally the fastener template matching is performed in the fastener gathering point cloud through a standard fastener generation matching model, and all three-dimensional point coordinate point sets s of the fastener template are obtained 0 Performing translation rotation change treatment to obtain a three-dimensional point coordinate point set s 'of the corresponding matched fastener' 0 ；
The fastener is normally located within 150mm of the rail, and therefore, in the fastener locating step, the T distance may typically be set between 100 and 200mm, preferably 150mm.
In the fastener positioning step, a point cloud template matching algorithm based on a Fast Point Feature Histogram (FPFH) descriptor performs fastener matching in a fastener collecting point cloud, and specifically, the method comprises the following steps:
step 1, selecting n sampling points from a point cloud P to be registered, wherein the distance between the sampling points is more than a preset minimum distance threshold d, so that the sampled points can be ensured to have different FPFH characteristics;
step 2, searching one or more points with similar FPFH characteristics with sampling points in the point cloud P in the target point cloud Q, and randomly selecting one point from the similar points as a one-to-one corresponding point of the point cloud P in the target point cloud Q;
step 3, calculating a rigid body transformation matrix between the corresponding points, and then judging the performance of the current registration transformation by solving a distance error sum function after the transformation of the corresponding points;
the translation rotation change processing specifically includes the steps of matching a three-dimensional point coordinate point set s of a fastener template matched with a fastener in a three-dimensional point cloud 0 Using translational rotation of a matrix 0 (t x ,t y ,t z 1, alpha, beta, lambda) to obtain a three-dimensional point coordinate point set s 'of the fastener corresponding to the matched fastener template in the three-dimensional point cloud of the fastener region' 0 Wherein (t) x ，t y ，t z 1) is a translation parameter, (alpha, beta, lambda) bit is a rotation parameter, and the translation rotation change matrix is derived from the output of a template matching algorithm, so that the translation parameter and the rotation parameter are both values obtained through the algorithm.
More specifically, first, three-dimensional point coordinates (x 0 ，y 0 ，z 0 ) Translation processing to obtain translation position coordinatesThree-dimensional point coordinates (x) 0 ，y 0 ，z 0 ) The three-dimensional point coordinates of the fastener template can be transformed into the original coordinate system through a transformation matrix, and can be directly used in the three-dimensional point coordinate system.
Then the translational position coordinates are rotated by rotation parameters (alpha, beta, lambda) to obtain translational rotated position coordinates (x' 0 ，y′ 0 ，z′ 0 ) The point set of all three-dimensional point coordinates of the fastener template after translational rotation treatment is the position coordinate point set of the fastener in the three-dimensional point cloud, and is marked as s '' 0 ，
Positioning the spring strip key points: extracting the current fastener point cloud s from the fastener collecting point cloud according to the point cloud horizontal distribution range of each fastener 1 The method comprises the steps of carrying out a first treatment on the surface of the Calculating the current fastener point cloud s 1 And corresponding point cloud s' 0 Is to eliminate the point cloud s 1 The points with medium space distances not meeting the threshold value form a point cloud s 2 The method comprises the steps of carrying out a first treatment on the surface of the Point cloud s 0 Sum point cloud s 2 Generating a two-dimensional image template model 0 And a top-down two-dimensional image 0 Model of two-dimensional image template 0 And a top-down two-dimensional image 0 Image template matching is carried out, and a two-dimensional image template model is extracted 0 To a top-down two-dimensional image 0 And model a two-dimensional image template 0 The bullet strip key point areas in the matrix are mapped to the matched overlook two-dimensional image through the transformation matrix 0 Corresponding position in the plane view to obtain a plane view two-dimensional image 0 The elastic strip key point area in the image acquisition system is obtained by marking the key point area in the template according to a priori threshold value or experience by manpower, and the key point area and the template are simultaneously rotated and translated through a transformation matrix, so that key point coordinates in acquired data can be accurately found, secondary accurate registration is completed, and overlook two-dimensional image is obtained 0 The area of the spring strip key measuring point; from point cloud s 2 The height average value of the elastic strip key point area is extracted to obtain the relative height h of the elastic strip key point 1 。
In the spring bar key point positioning step, a KD-tree structure is adopted to quickly traverse and calculate a point set s 1 Center and point set s' 0 The distance dis between each point in the list, and recording the point set s 1 Is separated from the set of points s' 0 Minimum distance dis of each point in (a) min Reject minimum distance dis min Points greater than the distance threshold m, the set of points s 1 The remaining points in (a) constitute a point set s 2 And the distance threshold m is a judgment threshold.
In the spring bar key point positioning step, a point set s is formed 0 Generating a overlooking two-dimensional image and establishing a two-dimensional image template model 0 Then, the area a of the selected key measurement point is recorded 0 In a template model 0 Position (x) a0min ～x a0max ，y a0min ～y a0max )；
And then the point set s 2 Generating a top-down two-dimensional image 0 Image is taken of 0 And model 0 Performing image template matching, and obtaining a two-dimensional transformation matrix according to the template matching 1 For the region a of the key measurement point 0 Transforming to obtain a 0 ' region, a 0 ' region, i.e. the region of key measurement points, is in the point set s 2 Is a position in the middle;
direct calculation of the Point set s 2 In middle a 0 Height average of' regionThe relative height h of the key measurement point can be obtained 1 。
The area of the key measurement point is located in one part of the elastic strip component in the fastener, the elastic strip is one of the component parts of the fastener, and the key point is located in the elastic strip of the fastener.
And a fastener reference point positioning step: longitudinally expanding the elastic strip key point area from point cloud s 2 Intercepting the point cloud of the area, dividing the area into i parts by adopting a density clustering method, and respectively calculating the average height h of each part iave Taking the lowest average height h' iave Average height h of reference point of insulating block 2 ；
In the fastener reference point positioning step, the size of the longitudinal expansion key point area is a 1 [x a0min ～x a0max ，(y a0min -n)～(y a0max +n)]N is a fixed length selected from the template based on an a priori threshold.
Fastener loosening judging step, namely directly calculating overlooking two-dimensional image obtained in the fastener key measurement point positioning step 0 The height average value h of the key point area of the fastener 1 Will h 1 H in the step of positioning with the fastener reference point 2 Obtaining the difference to obtain the height difference delta h between the key point area and the reference point area Measurement of Comparison of Δh Measurement of A standard value delta h from the distance Standard of Whether the difference is larger than the threshold U, thereby judging whether the fastener is loose or not and the standard value delta h Standard of The distance between the key point area of the standard fastener and the insulating block is determined according to the standard design value of the fastener; the threshold U is an acceptable fastener loosening range, and is set depending on the operating environment.
Corresponding to the method, the invention also provides a computer device, which comprises: one or more processors, a memory storing instructions that, when executed by the one or more processors, cause the one or more processors to perform the method described above.
And a non-transitory machine-readable storage medium storing executable instructions that, when executed, cause the machine to perform the method described above.
The beneficial effects are that:
1. according to the technical scheme provided by the invention, the three-dimensional point cloud data is roughly divided and extracted to obtain the area where the targeted fastener is located in a mode of externally connecting with quadrilateral coordinate acquisition according to classical track and component data as priori thresholds, the point cloud containing the fastener is quickly and directly obtained from the acquired object data by adopting a single index, all invalid data are eliminated, the processing recognition operand is greatly reduced, and the subsequent matching recognition processing can be more efficient and quick.
2. Furthermore, the method of the invention uses a three-dimensional template matching method for the point cloud containing the fasteners, uses the fastener standard template as a reference, automatically finds the position of each fastener from the acquired whole track point cloud data, then uses the overlook two-dimensional diagram of the fastener standard template to carry out secondary accurate configuration on the detected fasteners, and finds the relative positions of key point areas such as elastic strips and the like and the fasteners, thus the height of the key point areas can be accurately obtained, thereby rapidly completing the detection of fastener looseness in a mode of determining the height threshold value, and replacing the complex three-dimensional data and two-dimensional data conversion calculation process in a mode of more rapidly and saving calculation force compared with the prior art, and rapidly and accurately realizing the looseness detection in a mode of template matching.
3. After the point cloud of the fastener is obtained through template matching, a point cloud template matching algorithm based on a Fast Point Feature Histogram (FPFH) descriptor is adopted to identify the profile of the fastener, an accurate fastener template is obtained, then the actual height of the elastic strip in the fastener is calculated by taking the elastic strip serving as a classical key point in the fastener as an object, the actual height of the elastic strip in the detected track is obtained, planar point cloud data serving as a reference point is separated from point cloud data of the elastic strip through two-dimensional processing, the actual height of the elastic strip is obtained, meanwhile, the actual height of the elastic strip in the fastener model serving as the template is correspondingly calculated, the height difference between the actual height of the elastic strip in the detected track and the relative height of the elastic strip in the fastener model serving as a reference standard is calculated, the height difference calculation result is used as data for judging whether the fastener is loose, and the method of combining the two-dimensional image data through three-dimensional template matching is utilized, and the fastener area, the elastic strip key point area and the insulation block reference point area are automatically extracted from the point cloud data of the whole track surface and the two sides, and the fastener is not influenced by the change of the position, the direction, the height and the like.
Drawings
The foregoing and the following detailed description of the invention will become more apparent when read in conjunction with the following drawings in which:
FIG. 1 is a schematic diagram of the logic of the present invention;
FIG. 2 is a schematic view of a three-dimensional point cloud of a track acquired by a three-dimensional acquisition system;
FIG. 3 is a schematic diagram of a three-dimensional point cloud of a filtered track through a threshold range of altitude information;
FIG. 4 is a schematic drawing of an external cube coordinate for acquiring left and right rail point clouds;
FIG. 5 is a schematic view of a three-dimensional point cloud including fasteners;
FIG. 6 is a three-dimensional point cloud schematic of a fastener area;
FIG. 7 is a set of three-dimensional point coordinates s for a fastener template 0 A schematic diagram;
FIG. 8 is a three-dimensional point coordinate point set s 'of a fastener correspondingly matching a three-dimensional point cloud of a track' 0 A schematic diagram;
FIG. 9 is a three-dimensional point coordinate point set s' 0 Extracting a range schematic diagram;
FIG. 10 shows a point set s in an embodiment of the invention 2 Schematic of (2);
FIG. 11 is a two-dimensional image template model 0 A schematic diagram of a key point area in the database;
FIG. 12 is a schematic view of a longitudinal expansion of the area of key measurement points;
FIG. 13 is a schematic illustration of standard region point and keypoint region data height differences;
Detailed Description
The technical solution for achieving the object of the present invention will be further described with reference to several specific examples, but it should be noted that the technical solution claimed in the present invention includes but is not limited to the following examples.
Example 1
As a specific implementation manner of the invention, this embodiment discloses a fastener loosening detection method based on a three-dimensional point cloud combined with a two-dimensional image processing technology, as shown in fig. 1, which comprises a steel rail positioning step, a fastener positioning step, a spring bar key point positioning step, a fastener reference point positioning step and a fastener loosening judgment step, wherein each fastener position is found in the whole track point cloud data automatically by using a three-dimensional template matching method by using a fastener standard template, then a two-dimensional view of a template fastener is utilized to carry out secondary accurate configuration on the detected fastener, the relative position of a spring bar key point area and the fastener is found, the height of the spring bar key point area is obtained, the relative height difference between an insulation block reference point and a key point on the spring bar is utilized, the plane point cloud data of the insulation block reference point is separated from the point cloud data, the relative height of the insulation block reference point is obtained, and finally whether the fastener is loosened or not is judged by using the relative height difference between the key point on the spring bar and the insulation block reference point, and whether the fastener is loosened or not can be judged accurately by selecting the spring bar key point and the insulation block reference point.
Specifically, in the rail positioning step, as shown in fig. 2, a three-dimensional point cloud of the rail is acquired through a three-dimensional acquisition system, and a threshold range (z) is set according to height information in the three-dimensional point cloud of the rail acquired by the three-dimensional acquisition system railmin ～z railmax ) And as shown in fig. 3, the three-dimensional point cloud of the track is filtered through a threshold range, and the height information setting threshold range is a priori threshold value depending on the distance between the three-dimensional acquisition equipment and the track surface; then as shown in fig. 4, density clustering is performed on the three-dimensional point clouds of the filtered track to obtain the external cubic coordinates of the point clouds of the left and right steel rails respectively, specifically, the density clustering is performed by setting a distance judgment threshold value, calculating the distance between all three-dimensional point clouds in the three-dimensional point clouds, dividing the distance into one cluster, wherein the distance falls into a threshold value range, the clustering method is applied to the mature technology in three dimensions, the position relationship between the fastener and the steel rails determines that the fastener is necessarily positioned in the external cubic coordinates of the point clouds of the left and right steel rails, namely, the purpose of obtaining the external cubic coordinates is to perform preliminary division on the area where the fastener is positioned, and the clustering method is externally arrangedThe diagonal two-point coordinates of the cube are (x min ，y min ，z min ) And (x) max ，y max ，z max ) It should be noted that, in the description of the technical solution of the present invention, the directions X, Y, Z of the front, rear, left, right, upper, lower and coordinate points are all relative position limitation descriptions, rather than absolute orientation limitations, for better understanding the technical solution.
Then, the fastener positioning step is carried out by setting T as the distance between the two sides of the rail surface and the rail surface, wherein the T distance is usually set to be between 100 and 200mm, preferably directly 150mm and then x is set because the fasteners are positioned within the T distance between the two sides of the rail surface and the fasteners are normally positioned within 150mm of the rail min-T ～x max-T And intercepting the three-dimensional point cloud comprising the fastener as shown in fig. 5 from the external cubic coordinates of the left and right rail point clouds obtained in the rail positioning step.
Then, as shown in fig. 6, according to the height range (h min ～h max ) Filtering the three-dimensional point cloud comprising fasteners to obtain a three-dimensional point cloud of fastener areas, the height range (h min ～h max ) Depending on the distance between the installed three-dimensional acquisition equipment and the fastener, here just a larger and rough range, as the threshold range set for the height information in the rail positioning step, the fastener template matching is finally performed in the three-dimensional point cloud of the fastener area by the standard fastener generation matching model, and all three-dimensional point coordinate point sets s of the fastener template as in fig. 7 are obtained 0 Performing translation rotation change treatment to obtain a three-dimensional point coordinate point set s 'of the fastener correspondingly matched with the track three-dimensional point cloud as shown in fig. 8' 0 。
The spring bar key point positioning step extracts the current fastener point cloud s from the fastener gathering point cloud according to the point cloud horizontal distribution range of each fastener 1 As shown in fig. 9, according to the range (X 'of the three-dimensional point cloud position coordinates of the fastener matched in the fastener positioning step on the X axis and the Y axis' 0min ～x′ 0max ) And (y' 0min ～y′ 0max ) Acquisition from a three-dimensional acquisition systemExtracting three-dimensional point clouds falling in a range from all track three-dimensional point clouds to form a point set s 1 And in the form of point set s' 0 For reference, according to a set decision threshold, from a set of points s 1 Eliminating points which do not meet the requirement of the decision threshold, and constructing a point set s as shown in FIG. 10 2 The method comprises the steps of carrying out a first treatment on the surface of the Set of points s 0 Sum point set s 2 Generating a two-dimensional image template model 0 And a top-down two-dimensional image 0 Model of two-dimensional image template 0 And a top-down two-dimensional image 0 Matching the image template, and modeling the two-dimensional image template 0 The key point areas in the image are transformed to a matched overlook two-dimensional image through a transformation matrix based on a two-dimensional image matching algorithm 0 The corresponding positions of the key point areas are shown in fig. 11, the key point areas are marked in the template according to prior threshold values or experience by manual work, the key point areas and the template are simultaneously rotated and translated through a transformation matrix, the key point coordinates in the acquired data can be accurately found, secondary accurate registration is completed, and overlook two-dimensional image is obtained 0 Calculating the height average value of the key measurement points in the region to obtain the relative height h of the key measurement points 1 。
In the fastener loosening determination step, as shown in fig. 12, a planar two-dimensional image is obtained in the fastener key measurement point positioning step 0 Longitudinally expanding the area of the spring strip key measurement points, and adopting density-based clustering (such as DBSCAN clustering method) to divide the longitudinally expanded key measurement point area into i parts, and respectively calculating the average height h of each part iave The part with the lowest average height is the fastener reference point area, and the average height h' iave Mean height h of the insulation block reference point region 2 Whether looseness exists or not can be judged through the height difference between the key measurement points and the insulating blocks.
Specifically, in the fastener loosening judging step, the overlooking two-dimensional image obtained in the fastener key measurement point positioning step is directly calculated 0 The height average value h of the key point area of the fastener 1 Will h 1 H in the step of positioning with the fastener reference point 2 The difference is obtained and the difference is calculated,obtaining the height difference delta h between the key point area and the reference point area Measurement of Comparison of Δh Measurement of A standard value delta h from the distance Standard of Whether the difference is larger than the threshold U, thereby judging whether the fastener is loose or not and the standard value delta h Standard of The distance between the key point area of the standard fastener and the insulating block is determined according to the standard design value of the fastener; the threshold U is an acceptable fastener loosening range, and is set depending on the operating environment.
Example 2
As a preferred embodiment of the present invention, this example discloses a fastener loosening detection method based on a three-dimensional point cloud combined with a two-dimensional image processing technology, as shown in fig. 1, including a rail positioning step, a fastener key measurement point positioning step, a fastener reference point positioning step, and a fastener loosening judgment step, specifically:
in the rail positioning step, as shown in fig. 2, a three-dimensional point cloud of the rail is acquired by a three-dimensional acquisition system, and a threshold range Z is set according to the height information in the three-dimensional point cloud of the rail acquired by the three-dimensional acquisition system railmin ～Z railmax And as shown in fig. 3, the three-dimensional point cloud of the track is filtered through a threshold range, and the height information setting threshold range is a priori threshold value depending on the distance between the three-dimensional acquisition equipment and the track surface; then as shown in fig. 4, density clustering is performed on the three-dimensional point clouds of the filtered track to obtain external cube coordinates of the left and right rail point clouds respectively, specifically, by setting a distance judgment threshold value, performing interval calculation on all three-dimensional point clouds in the three-dimensional point clouds, dividing the interval into one cluster, wherein the interval falls into a threshold value range, the clustering method is applied to the mature technology in three dimensions, the position relationship between the fastener and the rail determines that the fastener is necessarily positioned in the external cube coordinates of the left and right rail point clouds, namely, the purpose of obtaining the external cube coordinates is to perform preliminary division on the area where the fastener is positioned, and the diagonal two-point coordinates of the external cube are set as (x) min ，y min ，z min ) And (x) max ，y max ，z max ) It should be noted that the description of the technical proposal of the invention refers to the front, the back, the left, the right, the upper part,The directions X, Y, Z of the lower and coordinate points are relative position definition descriptions for easier understanding of the technical solution, and are not absolute orientation definitions.
Then, the fastener positioning step is carried out by setting T as the distance between the two sides of the rail surface and the rail surface, wherein the T distance is usually set to be between 100 and 200mm, preferably directly 150mm and then x is set because the fasteners are positioned within the T distance between the two sides of the rail surface and the fasteners are normally positioned within 150mm of the rail min-T ～x max-T And intercepting the three-dimensional point cloud comprising the fastener as shown in fig. 5 from the external cubic coordinates of the left and right rail point clouds obtained in the rail positioning step.
Then, as shown in fig. 6, according to the height range (h min ～h max ) Filtering the three-dimensional point cloud comprising fasteners to obtain a three-dimensional point cloud of fastener areas, the height range (h min ～h max ) Depending on the distance between the installed three-dimensional acquisition equipment and the fastener, here just a larger and rough range, as the threshold range set for the height information in the rail positioning step, the fastener template matching is finally performed in the three-dimensional point cloud of the fastener area by the standard fastener generation matching model, and all three-dimensional point coordinate point sets s of the fastener template as in fig. 7 are obtained 0 Performing translation rotation change treatment to obtain a three-dimensional point coordinate point set s 'of the fastener correspondingly matched with the track three-dimensional point cloud as shown in fig. 8' 0 More specifically, the point cloud template matching algorithm based on the FPFH descriptor performs fastener matching in a three-dimensional point cloud comprising fasteners:
further, in the step of positioning the fastener, a point cloud template matching algorithm based on a Fast Point Feature Histogram (FPFH) descriptor performs fastener matching in a fastener collecting point cloud, and specifically includes the following steps:
step 1, selecting n sampling points from a point cloud P to be registered, wherein the distance between the sampling points is more than a preset minimum distance threshold d, so that the sampled points can be ensured to have different FPFH characteristics;
step 2, searching one or more points with similar FPFH characteristics with sampling points in the point cloud P in the target point cloud Q, and randomly selecting one point from the similar points as a one-to-one corresponding point of the point cloud P in the target point cloud Q;
step 3, calculating a rigid body transformation matrix between the corresponding points, and then judging the performance of the current registration transformation by solving a distance error sum function after the transformation of the corresponding points;
the translation rotation change processing specifically includes the steps of matching a three-dimensional point coordinate point set s of a fastener template matched with a fastener in a three-dimensional point cloud 0 Using translational rotation of a matrix 0 (t x ,t y ,t z 1, alpha, beta, lambda) to obtain a three-dimensional point coordinate point set s 'of the fastener corresponding to the matched fastener template in the three-dimensional point cloud of the fastener region' 0 Wherein (t) x ，t y ，t z 1) is a translation parameter, (alpha, beta, lambda) bit is a rotation parameter, and the translation rotation change matrix is derived from the output of a template matching algorithm, so that the translation parameter and the rotation parameter are both values obtained through the algorithm.
More specifically, first, three-dimensional point coordinates (x 0 ，y 0 ，z 0 ) Translation processing to obtain translation position coordinatesThree-dimensional point coordinates (x) 0 ，y 0 ，z 0 ) The three-dimensional point coordinates of the fastener template can be transformed into the original coordinate system through a transformation matrix, and can be directly used in the three-dimensional point coordinate system.
Then the translational position coordinates are rotated by rotation parameters (alpha, beta, lambda) to obtain translational rotated position coordinates (x' 0 ，y′ 0 ，z′ 0 ) The point set of all three-dimensional point coordinates of the fastener template after translational rotation treatment is the position coordinate point set of the fastener in the three-dimensional point cloud, and is marked as s '' 0 ，
The key measurement point positioning step of the fastener is shown in fig. 9, and the range (X ' of the three-dimensional point cloud position coordinates of the fastener matched in the fastener positioning step on the X axis and the Y axis is determined according to the range (X ' of the three-dimensional point cloud position coordinates ' 0min ～x′ 0max ) And (y' 0min ～y′ 0max ) Extracting three-dimensional point clouds falling into a range from all track three-dimensional point clouds acquired by a three-dimensional acquisition system to form a point set s 1 And in the form of point set s' 0 For reference, according to a set decision threshold, from a set of points s 1 Eliminating points which do not meet the requirement of the decision threshold, and constructing a point set s as shown in FIG. 10 2 Specifically, the KD-tree structure is adopted to quickly traverse and calculate the point set s 1 Center and point set s' 0 The distance dis between each point in the list, and recording the point set s 1 Is separated from the set of points s' 0 Minimum distance dis of each point in (a) min Reject minimum distance dis min Points greater than the distance threshold m, the set of points s 1 The remaining points in (a) constitute a point set s 2 And the distance threshold m is a judgment threshold.
Then the point set s 0 Sum point set s 2 Generating a two-dimensional image template model 0 And a top-down two-dimensional image 0 The point set s 0 Generating a overlooking two-dimensional image and establishing a two-dimensional image template model 0 Then, the area a of the selected key measurement point is recorded 0 In a template model 0 Position (x) a0min ～x a0max ，y a0min ～y a0max )；
And then the point set s 2 Generating a top-down two-dimensional image 0 Image is taken of 0 And model 0 Performing image template matching, and obtaining a two-dimensional transformation matrix according to the template matching 1 For the region a of the key measurement point 0 Transforming to obtain a 0 ' region, a 0 ' region, i.e. the region of key measurement points, is in the point set s 2 Is a position in the middle; that is, a two-dimensional image template model 0 In a top viewTwo-dimensional image 0 Matching the image template, and modeling the two-dimensional image template 0 The key point areas in the image are transformed to a matched overlook two-dimensional image through a transformation matrix based on a two-dimensional image matching algorithm 0 The corresponding positions of the key point areas are shown in fig. 11, the key point areas are marked in the template according to prior threshold values or experience by manual work, the key point areas and the template are simultaneously rotated and translated through a transformation matrix, the key point coordinates in the acquired data can be accurately found, secondary accurate registration is completed, and overlook two-dimensional image is obtained 0 In the elastic strip key measuring point area, directly calculating point set s 2 In middle a 0 The relative height h of the key measurement point can be obtained by the height average value of the' area 1 。
And preferably the area of the key measurement point is located in one of the spring elements of the fastener, the spring being one of the component parts of the fastener, the key point being located in the spring of the fastener.
In the fastener reference point positioning step, as shown in fig. 12, a planar two-dimensional image is obtained in the fastener key measurement point positioning step 0 Longitudinally expanding the area of the elastic strip key measurement point, wherein the size of the longitudinally expanded key point area is a 1 [x a0min ～x a0max ，(y a0min -n)～(y a0max +n)]N is a fixed length selected from the template based on an a priori threshold. Dividing the key measurement point region after longitudinal expansion into i parts by adopting clustering (DBSCAN) based on a density method, and respectively solving the average height h of each part iave The part with the lowest average height is the fastener reference point area, and the average height h' iave Mean height h of the insulation block reference point region 2 Whether looseness exists or not can be judged through the height difference between the key measurement points and the insulating blocks.
Specifically, in the fastener loosening judging step, the overlooking two-dimensional image obtained in the fastener key measurement point positioning step is directly calculated 0 The height average value h of the key point area of the fastener 1 Will h 1 H in the step of positioning with the fastener reference point 2 The difference is obtained and the difference is calculated,obtaining the height difference delta h between the key point area and the reference point area Measurement of As shown in FIG. 13, Δh is compared Measurement of A standard value delta h from the distance Standard of Whether the difference is larger than the threshold U, thereby judging whether the fastener is loose or not and the standard value delta h Standard of The distance between the key point area of the standard fastener and the insulating block is determined according to the standard design value of the fastener; the threshold U is an acceptable fastener loosening range, and is set depending on the operating environment.
Claims (10)
1. The fastener loosening detection method based on three-dimensional point cloud and two-dimensional image processing is characterized by comprising the following steps of:
rail positioning: acquiring a three-dimensional point cloud from a three-dimensional acquisition system, and determining a three-dimensional point cloud according to a point cloud height threshold range (z railmin ～z railmax ) Screening the steel rail point cloud and carrying out density clustering on the steel rail point cloud to obtain an external cube of the steel rail point cloud;
and a fastener positioning step: in x min-T ～x max-T Intercepting a point cloud containing a fastener from the external cube, wherein T is the transverse distance from the outer side of the fastener to the rail surface; according to the height range (h) min ～h max ) Screening the point cloud containing the fasteners to obtain a point cloud containing the fasteners; and then integrating the fastener point cloud with a template point cloud s generated by a standard fastener 0 Performing template matching and obtaining the template point cloud s 0 Translational rotation transformation matrix to each fastener, for the template point cloud s 0 Carrying out translation rotation, and recording the point cloud after the translation rotation as s' 0 ；
Positioning the spring strip key points: extracting the current fastener point cloud s from the fastener collecting point cloud according to the point cloud horizontal distribution range of each fastener 1 The method comprises the steps of carrying out a first treatment on the surface of the Calculating the current fastener point cloud s 1 And corresponding point cloud s' 0 Is to eliminate the point cloud s 1 The points with medium space distances not meeting the threshold value form a point cloud s 2 The method comprises the steps of carrying out a first treatment on the surface of the Point cloud s 0 Sum point cloud s 2 Generating a two-dimensional image template model 0 And a top-down two-dimensional image 0 Modeling a two-dimensional imagePlate model 0 And a top-down two-dimensional image 0 Image template matching is carried out, and a two-dimensional image template model is extracted 0 To a top-down two-dimensional image 0 And model a two-dimensional image template 0 The bullet strip key point areas in the matrix are mapped to the matched overlook two-dimensional image through the transformation matrix 0 Corresponding position in the plane view to obtain a plane view two-dimensional image 0 A spring strip key point area in the middle; from point cloud s 2 The height average value of the elastic strip key point area is extracted to obtain the relative height h of the elastic strip key point 1 ；
And a fastener reference point positioning step: longitudinally expanding the elastic strip key point area from point cloud s 2 Intercepting the point cloud of the area, dividing the area into i parts by adopting a density clustering method, and respectively calculating the average height h of each part iave Taking the lowest average height h' iave Average height h of reference point of insulating block 2 ；
Fastener loosening judging step: spring bar key point relative height h 1 Average height h from insulation block reference point 2 Performing difference to obtain the height difference delta h between the spring strip key point and the insulating block reference point Measurement of According to Deltah Measurement of Whether the threshold is exceeded or not, and whether the fastener is loosened or not is judged.
2. The fastener loosening detection method based on three-dimensional point cloud and two-dimensional image processing as claimed in claim 1, wherein: in the fastener positioning step, the T distance is set to be between 100 and 200 mm.
3. The fastener loosening detection method based on three-dimensional point cloud and two-dimensional image processing as claimed in claim 1, wherein in the fastener positioning step, a Fast Point Feature Histogram (FPFH) descriptor-based point cloud template matching algorithm performs fastener matching in a fastener collecting point cloud, specifically comprising the following steps:
step 1, selecting n sampling points from a point cloud P to be registered, wherein the distance between the sampling points is more than a preset minimum distance threshold d;
step 2, searching one or more points with similar FPFH characteristics with sampling points in the point cloud P in the target point cloud Q, and randomly selecting one point from the similar points as a one-to-one corresponding point of the point cloud P in the target point cloud Q;
step 3, calculating a rigid body transformation matrix between the corresponding points, and then judging the performance of the current registration transformation by solving a distance error sum function after the transformation of the corresponding points;
the translation rotation transformation processing specifically includes the steps of matching a three-dimensional point coordinate point set s of a fastener template matched with a fastener in a three-dimensional point cloud 0 Transforming matrix with translational rotation 0 (t x ,t y ,t z 1, alpha, beta, lambda) to obtain a three-dimensional point coordinate point set s 'of the fastener corresponding to the matched fastener template in the three-dimensional point cloud of the fastener region' 0 Wherein (t) x ，t y ，t z 1) is a translation parameter and (α, β, λ) is a rotation parameter.
4. The fastener loosening detection method based on three-dimensional point cloud and two-dimensional image processing as claimed in claim 3, wherein:
first, three-dimensional point coordinates (x 0 ，y 0 ，z 0 ) Translation processing to obtain translation position coordinates
Then the translational position coordinates are rotated by rotation parameters (alpha, beta, lambda) to obtain translational rotated position coordinates (x' 0 ，y′ 0 ，z′ 0 ) The point set of all three-dimensional point coordinates of the fastener template after translational rotation treatment is the position coordinate point set of the fastener in the three-dimensional point cloud, and is marked as s '' 0 ，
5. The fastener loosening detection method based on three-dimensional point cloud and two-dimensional image processing as claimed in claim 1, wherein: in the spring bar key point positioning step, a KD-tree structure is adopted to quickly traverse and calculate a point set s 1 Center and point set s' 0 The distance dis between each point in the list, and recording the point set s 1 Is separated from the set of points s' 0 Minimum distance dis of each point in (a) min Reject minimum distance dis min Points greater than the distance threshold m, the set of points s 1 The remaining points in (a) constitute a point set s 2 And the distance threshold m is a judgment threshold.
6. The fastener loosening detection method based on three-dimensional point cloud and two-dimensional image processing as claimed in claim 5, wherein: in the spring bar key point positioning step, a point set s is formed 0 Generating a overlooking two-dimensional image and establishing a two-dimensional image template model 0 Then, the area a of the selected key measurement point is recorded 0 In a template model 0 Position (x) a0min ～x a0max ，y a0min ～y a0max )；
And then the point set s 2 Generating a top-down two-dimensional image 0 Image is taken of 0 And model 0 Performing image template matching, and obtaining a two-dimensional transformation matrix according to the template matching 1 For the region a of the key measurement point 0 Transforming to obtain a 0 ' region, a 0 ' region, i.e. the region of key measurement points, is in the point set s 2 Is a position in the middle;
direct calculation of the Point set s 2 In a 0 The relative height h of the key measurement point can be obtained by the height average value of the' area 1 。
7. The fastener loosening detection method based on three-dimensional point cloud and two-dimensional image processing as claimed in claim 6, wherein: the area of the key measurement point is located in one part of the elastic strip component in the fastener, the elastic strip is one of the component parts of the fastener, and the key point is located in the elastic strip of the fastener.
8. The fastener loosening detection method based on three-dimensional point cloud and two-dimensional image processing as claimed in claim 1, wherein: in the fastener reference point positioning step, the size of the longitudinal expansion key point area is a 1 [x a0min ～x a0max ，(y a0min -n)～(y a0max +n)]N is a fixed length selected from the template based on an a priori threshold.
9. A computer device, characterized by: comprising one or more processors, a memory storing instructions that, when executed by the one or more processors, cause the one or more processors to perform the method of any of the preceding claims 1-8.
10. A non-transitory machine-readable storage medium, characterized by: which stores executable instructions that, when executed, cause the machine to perform the method of any of the preceding claims 1-8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110579074.5A CN113469940B (en) | 2021-05-26 | 2021-05-26 | Fastener loosening detection method based on three-dimensional point cloud and two-dimensional image processing technology |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110579074.5A CN113469940B (en) | 2021-05-26 | 2021-05-26 | Fastener loosening detection method based on three-dimensional point cloud and two-dimensional image processing technology |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113469940A CN113469940A (en) | 2021-10-01 |
| CN113469940B true CN113469940B (en) | 2023-08-04 |
Family
ID=77871348
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110579074.5A Active CN113469940B (en) | 2021-05-26 | 2021-05-26 | Fastener loosening detection method based on three-dimensional point cloud and two-dimensional image processing technology |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113469940B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114354618A (en) * | 2021-12-16 | 2022-04-15 | 浙江大华技术股份有限公司 | Method and device for detecting welding seam |
| CN115100109B (en) * | 2022-05-19 | 2023-04-18 | 南昌工程学院 | Method for detecting tightness state of rail elastic strip fastener |
| CN115830019B (en) * | 2023-02-14 | 2023-04-28 | 南京慧然科技有限公司 | Three-dimensional point cloud calibration processing method and device for steel rail detection |
| CN117635520A (en) * | 2023-03-20 | 2024-03-01 | 南昌工程学院 | Elastic strip fastener tightness detection method by fusing two-dimensional image and three-dimensional point cloud |
| CN117830199A (en) * | 2023-04-14 | 2024-04-05 | 南昌工程学院 | Comprehensive detection method for defects of track elastic strip fastener by fusing image and point cloud |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2009210276A (en) * | 2008-02-29 | 2009-09-17 | Mitsubishi Heavy Ind Ltd | System and method for detecting loosening of fastening implement |
| CN106192634A (en) * | 2016-08-31 | 2016-12-07 | 武汉汉宁轨道交通技术有限公司 | A kind of railroad track elastic bar fastener condition automatic detection device and method |
| CN108797241A (en) * | 2018-06-05 | 2018-11-13 | 成都精工华耀科技有限公司 | A kind of rail clip nut loosening detection method compared based on height |
| CN108830891A (en) * | 2018-06-05 | 2018-11-16 | 成都精工华耀科技有限公司 | A kind of rail splice fastener loosening detection method |
| CN110634122A (en) * | 2018-06-05 | 2019-12-31 | 成都精工华耀科技有限公司 | Fastener elastic strip displacement detection method |
| CN112365461A (en) * | 2020-11-06 | 2021-02-12 | 北京格灵深瞳信息技术有限公司 | Fastener loosening identification method, system, terminal and storage medium |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10410043B2 (en) * | 2016-06-24 | 2019-09-10 | Skusub LLC | System and method for part identification using 3D imaging |
| WO2019183475A1 (en) * | 2018-03-23 | 2019-09-26 | University Of Kansas | Vision-based fastener loosening detection |
-
2021
- 2021-05-26 CN CN202110579074.5A patent/CN113469940B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2009210276A (en) * | 2008-02-29 | 2009-09-17 | Mitsubishi Heavy Ind Ltd | System and method for detecting loosening of fastening implement |
| CN106192634A (en) * | 2016-08-31 | 2016-12-07 | 武汉汉宁轨道交通技术有限公司 | A kind of railroad track elastic bar fastener condition automatic detection device and method |
| CN108797241A (en) * | 2018-06-05 | 2018-11-13 | 成都精工华耀科技有限公司 | A kind of rail clip nut loosening detection method compared based on height |
| CN108830891A (en) * | 2018-06-05 | 2018-11-16 | 成都精工华耀科技有限公司 | A kind of rail splice fastener loosening detection method |
| CN110634122A (en) * | 2018-06-05 | 2019-12-31 | 成都精工华耀科技有限公司 | Fastener elastic strip displacement detection method |
| CN112365461A (en) * | 2020-11-06 | 2021-02-12 | 北京格灵深瞳信息技术有限公司 | Fastener loosening identification method, system, terminal and storage medium |
Non-Patent Citations (1)
| Title |
|---|
| 高速铁路扣件缺陷三维检测***集成与识别算法;代先星;《中国博士学位论文全文数据库 工程科技Ⅱ辑》(第（2020）03期);C033-3 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113469940A (en) | 2021-10-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113469940B (en) | Fastener loosening detection method based on three-dimensional point cloud and two-dimensional image processing technology | |
| CN110443836B (en) | Point cloud data automatic registration method and device based on plane features | |
| CN109544612B (en) | Point cloud registration method based on feature point geometric surface description | |
| CN108107444B (en) | Transformer substation foreign matter identification method based on laser data | |
| CN100485710C (en) | Method for recognizing vehicle type by digital picture processing technology | |
| CN106683076B (en) | The method of train wheel tread damage detection based on textural characteristics cluster | |
| CN103295232B (en) | Based on the SAR image registration method in straight line and region | |
| CN107358617B (en) | Method for detecting target vertex in ground penetrating radar recording profile | |
| CN112465948A (en) | Vehicle-mounted laser pavement point cloud rarefying method capable of retaining spatial features | |
| CN109409385A (en) | A kind of pointer instrument automatic identifying method | |
| CN101604450A (en) | The method of integrated images and LiDAR data extract contour of building | |
| CN111179321B (en) | Point cloud registration method based on template matching | |
| CN103727930A (en) | Edge-matching-based relative pose calibration method of laser range finder and camera | |
| CN102750449B (en) | Point cloud linear feature extraction method based on substep three-dimensional space and feature dimension mapping | |
| CN110807781A (en) | Point cloud simplification method capable of retaining details and boundary features | |
| CN111508073A (en) | Method for extracting roof contour line of three-dimensional building model | |
| CN113487722A (en) | Automatic concrete member detection method based on three-dimensional laser scanning method | |
| CN107610174B (en) | Robust depth information-based plane detection method and system | |
| CN108931206A (en) | Method for distinguishing is known for rail profile outlier detection and effective profile | |
| CN106886988B (en) | Linear target detection method and system based on unmanned aerial vehicle remote sensing | |
| CN108205645A (en) | A kind of reference map quality evaluating method of heterologous image matching system | |
| CN112232248A (en) | Method and device for extracting plane features of multi-line LiDAR point cloud data | |
| CN116681912A (en) | Rail gauge detection method and device for railway turnout | |
| CN101635028A (en) | Image detecting method and image detecting device | |
| CN116469092A (en) | Self-adaptive Alpha Shapes contour extraction method based on DBSCAN algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |