CN110473223B - Two-dimensional image auxiliary segmentation method based on three-dimensional point cloud of catenary cantilever system - Google Patents
Two-dimensional image auxiliary segmentation method based on three-dimensional point cloud of catenary cantilever system Download PDFInfo
- Publication number
- CN110473223B CN110473223B CN201910753299.0A CN201910753299A CN110473223B CN 110473223 B CN110473223 B CN 110473223B CN 201910753299 A CN201910753299 A CN 201910753299A CN 110473223 B CN110473223 B CN 110473223B
- Authority
- CN
- China
- Prior art keywords
- point cloud
- dimensional
- dimensional point
- dimensional image
- cantilever
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a two-dimensional image auxiliary segmentation method based on a three-dimensional point cloud of a catenary cantilever system, which comprises the following steps: step 1: acquiring three-dimensional point cloud data of a cantilever system of the overhead line system; step 2: uniformly resampling the three-dimensional point cloud data by adopting voxel filtering; step 3: converting the uniformly resampled three-dimensional point cloud data of the wrist system into a two-dimensional image; step 4: dividing the two-dimensional image obtained in the step 3, then sequentially performing image closing operation and median filtering treatment, and returning to the three-dimensional point cloud to obtain the dividing result of each linear part of the three-dimensional point cloud of the wrist system; the method adopts the two-dimensional image to carry out auxiliary segmentation, and finally returns the high-efficiency segmentation result of each linear part of the catenary cantilever system, thereby having the characteristics of good noise immunity, strong robustness and higher precision and improving the segmentation effect of the catenary cantilever system; the consumption of manpower and material resources is reduced, and the influence of weather and experience judgment of operators is avoided.
Description
Technical Field
The invention relates to the field of maintenance and detection of high-speed railway overhead lines, in particular to a two-dimensional image auxiliary segmentation method based on a three-dimensional point cloud of an overhead line cantilever system.
Background
With the great development of electrified railways, train operation elements are also continuously lifted. In order to ensure the high efficiency of train operation, high requirements are put on the stability and reliability of current receiving of the pantograph. Therefore, rigidity invariance of the wrist system must be ensured. The connecting parts of the cantilever system are loosened and shifted, so that on one hand, the carrier rope and the contact line deviate from the inherent position to generate the phenomena of bowing and the like, and on the other hand, the internal stress structure of the cantilever system can be changed to cause the phenomenon of local or regional loosening of the contact net system to influence the normal operation of a train. Therefore, it becomes important to acquire three-dimensional point cloud data of the catenary cantilever system by using a 3D visual technology, and combine the converted two-dimensional images to perform auxiliary segmentation, so that the segmentation result is improved, and a better basis is provided for the calculation of the related parameters of the cantilever system.
At present, the contact network cantilever system is maintained and detected mainly by manpower, a large amount of manpower and material resources are consumed, driving is interfered, and the influence is judged by weather and experience of operators.
Disclosure of Invention
The two-dimensional image auxiliary segmentation method based on the three-dimensional point cloud of the contact net cantilever system, provided by the invention, has the advantages of good noise immunity, strong robustness and higher precision, and is used for segmenting the contact net cantilever system.
The technical scheme adopted by the invention is as follows: a two-dimensional image auxiliary segmentation method based on a three-dimensional point cloud of a catenary cantilever system comprises the following steps:
step 1: acquiring three-dimensional point cloud data of a cantilever system of the overhead line system;
step 2: uniformly resampling the three-dimensional point cloud data by adopting voxel filtering;
step 3: converting the uniformly resampled three-dimensional point cloud data of the wrist system into a two-dimensional image;
step 4: and (3) dividing the two-dimensional image obtained in the step (3), then sequentially performing image closing operation and median filtering treatment, and returning to the three-dimensional point cloud to obtain the dividing result of each linear part of the three-dimensional point cloud of the wrist system.
Further, the specific process of the step 3 is as follows:
s11: determining a plane coordinate system x-o-z, x-o-y and y-o-z;
s12: determining the projection range of a two-dimensional image plane, calculating the maximum value and the minimum value in the X, Y, Z direction in the three-dimensional point cloud, and setting the unit scale of a plane coordinate system as the side length of a voxel;
s13: determining the position of the three-dimensional point cloud in the plane coordinates;
s14: determining a gray value of the two-dimensional image:
wherein: color (i, j) is the gray value of the two-dimensional image corresponding to the calculated current point, Y (i, j) is the coordinate needed to be obtained by the current point, Y min Is the minimum value of the three-dimensional point cloud on the Y axis, Y max The maximum value of the three-dimensional point cloud on the Y axis;
s15: and restoring the three-dimensional coordinates.
Further, the specific process of the step 4 is as follows:
s21: dividing the two-dimensional image obtained in the step 3 by adopting an SC-LCCP algorithm, respectively and independently extracting each linear part, and converting the rest part into a two-dimensional image;
s22: subtracting the two-dimensional image of the residual part corresponding to each linear part in the S11 from the two-dimensional image of the three-dimensional point cloud in the step 3, and then sequentially performing image closing operation and median filtering treatment;
s23: returning the segmentation result of the two-dimensional image obtained in the step S22 to the segmentation result of each linear part in the three-dimensional point cloud data of the catenary cantilever system according to the following steps;
wherein: color (i, j) is the gray value of the two-dimensional image corresponding to the calculated current point, Y (i, j) is the coordinate needed to be obtained by the current point, Y min Is the minimum value of the three-dimensional point cloud on the Y axis, Y max Is the maximum value of the three-dimensional point cloud on the Y axis.
Further, the voxel filtering in the step 2 is adopted to uniformly resample the three-dimensional point cloud data, which is as follows:
adopting a straight-through filter, respectively setting threshold ranges of x, y and z axes according to the imaging position of a coordinate system of a camera adopted in the imaging process of the wrist system, and filtering the background environment of the wrist system; and setting the size of a voxel body by utilizing voxel filtering, and uniformly resampling the three-dimensional point cloud of the wrist system.
Further, the linear portion in step S21 includes a flat arm, a cantilever support, an oblique arm, a positioning tube support, a positioning tube and a positioner.
The beneficial effects of the invention are as follows:
(1) According to the method, the three-dimensional point cloud data of the catenary cantilever system are obtained through a 3D visual image technology, so that the load of the catenary system is not increased;
(2) The method adopts the two-dimensional image to carry out auxiliary segmentation, and finally returns the high-efficiency segmentation result of each linear part of the catenary cantilever system, thereby having the characteristics of good noise immunity, strong robustness and higher precision and improving the segmentation effect of the catenary cantilever system;
(3) The invention adopts a non-contact 3D visual image technology, reduces the consumption of manpower and material resources, and is not influenced by weather and experience judgment of operators.
Drawings
FIG. 1 is a schematic diagram of a segmentation process according to the present invention.
FIG. 2 is a schematic diagram of a detection apparatus used in the present invention.
Fig. 3 is a schematic diagram of acquiring point cloud data of a frame of wrist system in the present invention.
Fig. 4 is a view of point cloud data of a field one-frame wrist system obtained in the present invention.
Fig. 5 is a diagram of the wrist system point cloud data after the straight-through filtering in the present invention.
Fig. 6 is a view of the point cloud data of the cantilever system after voxel filtering in the present invention.
Fig. 7 is a two-dimensional planar coordinate system constructed in the present invention.
FIG. 8 is a position of a point cloud in the x-o-z plane of the present invention.
Fig. 9 is a two-dimensional image converted in the present invention.
Fig. 10 is a two-dimensional image of the SC-LCCP segmentation result in the present invention.
FIG. 11 is a schematic diagram of a two-dimensional image segmentation result according to the present invention.
Fig. 12 is a three-dimensional point cloud segmentation result of the mapped cantilever system in the present invention.
Fig. 13 is a three-dimensional point cloud dataset of an original catenary cantilever system of the present invention.
Fig. 14 is a three-dimensional point cloud segmentation result of the cantilever system in the present invention.
In the figure: 1-detecting vehicle, 2-depth camera, 3-carrier cable, 4-hanger wire, 5-track, 6-contact line, 7-cantilever system.
Detailed Description
The invention will be further described with reference to the drawings and specific examples.
As shown in fig. 1, a two-dimensional image auxiliary segmentation method based on a three-dimensional point cloud of a catenary cantilever system comprises the following steps:
step 1: acquiring three-dimensional point cloud data of a cantilever system of the overhead line system;
the depth camera is placed directly above the inspection vehicle, with the camera horizontal and tilted at an angle, as shown in fig. 2. The detection vehicle is moved, so that the whole wrist system can acquire multi-frame wrist system point cloud data in real time in the visible range of the depth camera, as shown in fig. 3, and fig. 4 is one frame of acquired wrist point cloud data.
Step 2: uniformly resampling the three-dimensional point cloud data by adopting voxel filtering;
the process is as follows:
adopting a straight-through filter, respectively setting threshold ranges of x, y and z axes according to the imaging position of a coordinate system of a camera adopted in the imaging process of the wrist system, and filtering the background environment of the wrist system; the pass filter parameter reference settings are shown in table 1 and the filtering results are shown in fig. 5. And (3) uniformly resampling the three-dimensional point cloud of the wrist system by utilizing voxel filtering, wherein the size of the voxel is set to be 0.02m, and the filtering result is shown in fig. 6.
TABLE 1 pass filter parameter settings
Step 3: converting the uniformly resampled three-dimensional point cloud data of the wrist system into a two-dimensional image;
the specific process is as follows:
s11: determining a plane coordinate system x-o-z, x-o-y and y-o-z (any coordinate axis can be positioned on a plane), wherein the coordinate system is shown in fig. 7;
s12: determining the projection range of the two-dimensional image plane, calculating the maximum value and the minimum value in the X, Y, Z direction in the three-dimensional point cloud, and respectively marking as X max 、Y max 、Z max And X min 、Y min 、Z min The method comprises the steps of carrying out a first treatment on the surface of the Setting upThe unit scale of the plane coordinate system is the side length of the voxel; thereby ensuring that the area corresponding to the unit scale of the plane coordinate system is the surface size of the voxel.
S13: determining the position of the three-dimensional point cloud in plane coordinates (such as x-o-z, wherein any coordinate axis is located on a plane) as shown in fig. 8;
s14: determining a gray value of the two-dimensional image:
wherein: color (i, j) is the gray value of the two-dimensional image corresponding to the calculated current point, Y (i, j) is the coordinate needed to be obtained by the current point, Y min Is the minimum value of the three-dimensional point cloud on the Y axis, Y max The maximum value of the three-dimensional point cloud on the Y axis;
s15: the three-dimensional coordinates are restored and the two-dimensional image is shown in fig. 9.
Step 4: and (3) dividing the two-dimensional image obtained in the step (3), then sequentially performing image closing operation and median filtering treatment, and returning to the three-dimensional point cloud to obtain the dividing result of each linear part of the three-dimensional point cloud of the wrist system.
The specific process is as follows:
s21: and (3) dividing the two-dimensional image obtained in the step (3) by adopting a convex connection packing (Slope Constrained Locally Convex Connected Patches: SC-LCCP) SC-LCCP algorithm based on slope constraint, and respectively and independently extracting each linear part, wherein each linear part comprises a flat wrist arm, a wrist arm support, an inclined wrist arm, a positioning pipe support, a positioning pipe and a positioner. And converts the remaining portion into a two-dimensional image as shown in fig. 10.
S22: subtracting the two-dimensional image of the residual part corresponding to each linear part in the S11 from the two-dimensional image of the three-dimensional point cloud in the step 3, and then sequentially performing image closing operation and median filtering treatment, as shown in fig. 11;
s23: the segmentation result of the two-dimensional image obtained in the step S22 is returned to the segmentation result of each linear part in the three-dimensional point cloud data of the catenary cantilever system according to the following, as shown in fig. 12;
wherein: color (i, j) is the gray value of the two-dimensional image corresponding to the calculated current point, Y (i, j) is the coordinate needed to be obtained by the current point, Y min Is the minimum value of the three-dimensional point cloud on the Y axis, Y max Is the maximum value of the three-dimensional point cloud on the Y axis.
By adopting the method, the point cloud data of 500 groups of wrist systems acquired by the depth camera are subjected to efficient segmentation of each linear part, and partial original data sets and 6 groups of segmentation results are displayed. The original data is shown in fig. 13, and the division result data is shown in fig. 14.
According to the invention, through a 3D visual image technology, the contact net cantilever system is subjected to efficient segmentation of each linear part. The non-contact type cantilever system segmentation method does not add extra load to the contact network system and does not interfere driving; the consumption of manpower and material resources is reduced. Is not influenced by weather constraint and experience judgment of operators. The convex connection packing (Slope Constrained Locally Convex Connected Patches:SC-LCCP) algorithm based on the slope constraint is adopted, so that the segmentation result of each linear part is improved, a better foundation is provided for subsequent parameter detection, and the method has a better use prospect.
Claims (2)
1. The two-dimensional image auxiliary segmentation method based on the three-dimensional point cloud of the catenary cantilever system is characterized by comprising the following steps of:
step 1: acquiring three-dimensional point cloud data of a cantilever system of the overhead line system;
step 2: uniformly resampling the three-dimensional point cloud data by adopting voxel filtering;
adopting a straight-through filter, respectively setting threshold ranges of x, y and z axes according to the imaging position of a coordinate system of a camera adopted in the imaging process of the wrist system, and filtering the background environment of the wrist system; setting the size of a voxel body by utilizing voxel filtering, and uniformly resampling the three-dimensional point cloud of the wrist system;
step 3: converting the uniformly resampled three-dimensional point cloud data of the wrist system into a two-dimensional image;
s31: determining a plane coordinate system x-o-z, x-o-y and y-o-z;
s32: determining the projection range of a two-dimensional image plane, calculating the maximum value and the minimum value in the X, Y, Z direction in the three-dimensional point cloud, and setting the unit scale of a plane coordinate system as the side length of a voxel;
s33: determining the position of the three-dimensional point cloud in the plane coordinates;
s34: determining a gray value of the two-dimensional image:
wherein: color (i, j) is the gray value of the two-dimensional image corresponding to the calculated current point, Y (i, j) is the coordinate needed to be obtained by the current point, Y min Is the minimum value of the three-dimensional point cloud on the Y axis, Y max The maximum value of the three-dimensional point cloud on the Y axis;
s35: restoring the three-dimensional coordinates;
step 4: dividing the two-dimensional image obtained in the step 3, then sequentially performing image closing operation and median filtering treatment, and returning to the three-dimensional point cloud to obtain the dividing result of each linear part of the three-dimensional point cloud of the wrist system;
s41: dividing the two-dimensional image obtained in the step 3 by adopting an SC-LCCP algorithm, respectively and independently extracting each linear part, and converting the rest part into a two-dimensional image;
s42: subtracting the two-dimensional image of the residual part corresponding to each linear part in the S31 from the two-dimensional image of the three-dimensional point cloud in the step 3, and then sequentially performing image closing operation and median filtering treatment;
s43: returning the segmentation result of the two-dimensional image obtained in the step S42 to the segmentation result of each linear part in the three-dimensional point cloud data of the catenary cantilever system according to the following steps;
wherein: color (i, j) is the gray value of the two-dimensional image corresponding to the calculated current point, Y (i, j) is the coordinate needed to be obtained by the current point, Y min Is the minimum value of the three-dimensional point cloud on the Y axis, Y max Is the maximum value of the three-dimensional point cloud on the Y axis.
2. The two-dimensional image aided segmentation method based on the three-dimensional point cloud of the catenary cantilever system according to claim 1, wherein the linear part in the step S41 comprises a flat cantilever, a cantilever support, an inclined cantilever, a positioning pipe support, a positioning pipe and a positioner.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910753299.0A CN110473223B (en) | 2019-08-15 | 2019-08-15 | Two-dimensional image auxiliary segmentation method based on three-dimensional point cloud of catenary cantilever system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910753299.0A CN110473223B (en) | 2019-08-15 | 2019-08-15 | Two-dimensional image auxiliary segmentation method based on three-dimensional point cloud of catenary cantilever system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110473223A CN110473223A (en) | 2019-11-19 |
| CN110473223B true CN110473223B (en) | 2023-05-05 |
Family
ID=68510090
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910753299.0A Active CN110473223B (en) | 2019-08-15 | 2019-08-15 | Two-dimensional image auxiliary segmentation method based on three-dimensional point cloud of catenary cantilever system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110473223B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110335263A (en) * | 2019-06-28 | 2019-10-15 | 珠海博明软件有限公司 | It is a kind of to identify the measurement scheme for improving 3D difference in height computational accuracy by brightness |
| CN111814679B (en) * | 2020-07-08 | 2023-10-13 | 上海雪湖科技有限公司 | FPGA-based implementation algorithm for voxel-encoder and VFE of voxel-encoder 3D network |
| CN112948954B (en) * | 2021-04-16 | 2023-06-02 | 中铁第四勘察设计院集团有限公司 | Three-dimensional modeling method and system for catenary cantilever positioning device with driving constraint |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2011070927A1 (en) * | 2009-12-11 | 2011-06-16 | 株式会社トプコン | Point group data processing device, point group data processing method, and point group data processing program |
| CN104299260A (en) * | 2014-09-10 | 2015-01-21 | 西南交通大学 | Contact network three-dimensional reconstruction method based on SIFT and LBP point cloud registration |
| CN107123161A (en) * | 2017-06-14 | 2017-09-01 | 西南交通大学 | A kind of the whole network three-dimensional rebuilding method of contact net zero based on NARF and FPFH |
| CN108564575A (en) * | 2018-04-11 | 2018-09-21 | 西南交通大学 | A kind of contactless catenary's parameters detection method based on three dimensional point cloud |
| CN109118500A (en) * | 2018-07-16 | 2019-01-01 | 重庆大学产业技术研究院 | A kind of dividing method of the Point Cloud Data from Three Dimension Laser Scanning based on image |
| CN109816682A (en) * | 2019-01-22 | 2019-05-28 | 西南交通大学 | A kind of bracket System Partition and parameter detection method based on concavity and convexity |
-
2019
- 2019-08-15 CN CN201910753299.0A patent/CN110473223B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2011070927A1 (en) * | 2009-12-11 | 2011-06-16 | 株式会社トプコン | Point group data processing device, point group data processing method, and point group data processing program |
| CN104299260A (en) * | 2014-09-10 | 2015-01-21 | 西南交通大学 | Contact network three-dimensional reconstruction method based on SIFT and LBP point cloud registration |
| CN107123161A (en) * | 2017-06-14 | 2017-09-01 | 西南交通大学 | A kind of the whole network three-dimensional rebuilding method of contact net zero based on NARF and FPFH |
| CN108564575A (en) * | 2018-04-11 | 2018-09-21 | 西南交通大学 | A kind of contactless catenary's parameters detection method based on three dimensional point cloud |
| CN109118500A (en) * | 2018-07-16 | 2019-01-01 | 重庆大学产业技术研究院 | A kind of dividing method of the Point Cloud Data from Three Dimension Laser Scanning based on image |
| CN109816682A (en) * | 2019-01-22 | 2019-05-28 | 西南交通大学 | A kind of bracket System Partition and parameter detection method based on concavity and convexity |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110473223A (en) | 2019-11-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110473223B (en) | Two-dimensional image auxiliary segmentation method based on three-dimensional point cloud of catenary cantilever system | |
| CN109903327B (en) | Target size measurement method of sparse point cloud | |
| CN107833270B (en) | Real-time object three-dimensional reconstruction method based on depth camera | |
| CN103814306B (en) | Depth survey quality strengthens | |
| CN106875437B (en) | RGBD three-dimensional reconstruction-oriented key frame extraction method | |
| CN110148217A (en) | A kind of real-time three-dimensional method for reconstructing, device and equipment | |
| CN102881047B (en) | Automatic non-closed implicit curved surface reconstruction method | |
| CN105741276B (en) | A kind of ship waterline extracting method | |
| CN104268519B (en) | Image recognition terminal and its recognition methods based on pattern match | |
| CN110910443A (en) | Contact net geometric parameter real-time measuring method and device based on single monitoring camera | |
| CN111859805B (en) | Method for detecting topological relation of electric power drawing based on artificial intelligence | |
| CN108734715B (en) | Contact net rigid wire identification method and device for adaptive learning | |
| CN105303616A (en) | Embossment modeling method based on single photograph | |
| CN104574432A (en) | Three-dimensional face reconstruction method and three-dimensional face reconstruction system for automatic multi-view-angle face auto-shooting image | |
| CN107705256B (en) | Forward vehicle-mounted video image enhancement method based on contact network semantics | |
| CN110363769B (en) | Image segmentation method for cantilever system of high-speed rail contact net supporting device | |
| CN108416803A (en) | A kind of scene depth restoration methods of the Multi-information acquisition based on deep neural network | |
| JPWO2019021569A1 (en) | Information processing apparatus, information processing method, and program | |
| CN110660094A (en) | Subway tunnel mobile scanning point cloud fine division method based on image recognition | |
| CN109727255B (en) | Building three-dimensional model segmentation method | |
| CN109816682B (en) | Wrist-arm system segmentation and parameter detection method based on concavity and convexity | |
| CN103514587B (en) | Ship-based image-stabilizing method based on sea-sky boundary detecting | |
| CN113240790B (en) | Rail defect image generation method based on 3D model and point cloud processing | |
| CN103106663B (en) | Realize the method for SIM card defects detection based on image procossing in computer system | |
| CN108759671B (en) | Dynamic detection method for contact line abrasion |
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 |