CN114267039B - Recognition result fine processing method based on shield tunnel transverse seam priori rules - Google Patents
Recognition result fine processing method based on shield tunnel transverse seam priori rules Download PDFInfo
- Publication number
- CN114267039B CN114267039B CN202111532466.2A CN202111532466A CN114267039B CN 114267039 B CN114267039 B CN 114267039B CN 202111532466 A CN202111532466 A CN 202111532466A CN 114267039 B CN114267039 B CN 114267039B
- Authority
- CN
- China
- Prior art keywords
- transverse seam
- ring
- shield tunnel
- section
- segment
- 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
- 238000003672 processing method Methods 0.000 title claims abstract description 8
- 238000000034 method Methods 0.000 claims description 19
- 239000011159 matrix material Substances 0.000 claims description 12
- 238000001914 filtration Methods 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 238000011426 transformation method Methods 0.000 claims description 2
- 238000001514 detection method Methods 0.000 abstract description 11
- 238000013136 deep learning model Methods 0.000 abstract description 5
- 230000007547 defect Effects 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000002372 labelling Methods 0.000 description 2
- 241000282414 Homo sapiens Species 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Landscapes
- Image Analysis (AREA)
- Lining And Supports For Tunnels (AREA)
Abstract
The invention discloses a precise processing method of a recognition result based on a shield tunnel transverse seam priori rule. The invention has the advantages that: the accuracy of the recognition of the segment transverse seam is improved, and the defects of missed detection and false detection of the conventional deep learning model transverse seam recognition result are overcome.
Description
Technical Field
The invention belongs to the technical field of image recognition, and particularly relates to a recognition result fine processing method based on a shield tunnel transverse seam priori rule.
Background
The accurate identification of the transverse seam of the shield tunnel is an important ring in the three-dimensional laser scanning data processing of the tunnel.
The automatic segment joint position identification method in shield tunnel images adopts a model based on image segmentation and deep learning to identify a transverse joint binary image, and transverse joint line segments are detected from the binary image through Hough transformation and used as a final vector result of a circular joint; a linear detection method of tunnel image based on convolutional neural network is to directly generate vector result of transverse seam of image by linear detection method based on deep learning.
Both of the above methods are supervised learning methods in labeling datasets. On one hand, a single transverse seam of a shield tunnel is short, the range of the single transverse seam is limited in characteristics, and partial transverse seams can be blocked by a tunnel cable, so that a deep learning model based on data is difficult to learn the characteristics; on the other hand, the deep learning model is hard to learn the knowledge about the transverse seams which are easy to understand by human beings in the data supervision training, for example, the number of the transverse seams of each ring is the same, the positions of a plurality of transverse seams in one interval are the same, and the like.
Therefore, the deep learning model trained on the labeling data set always has the conditions of detection omission and false detection to a certain extent, and the number of transverse slits in one shield tunnel section is usually thousands of, so that the data processing efficiency is greatly affected even if 2% of false detection and false detection are omitted.
Disclosure of Invention
According to the defects of the prior art, the invention provides a precise processing method for the recognition result based on the prior rule of the transverse seam of the shield tunnel.
The invention is realized by the following technical scheme:
A recognition result fine processing method based on a shield tunnel transverse seam priori rule is characterized by comprising the following steps:
(1) Collecting a three-dimensional laser scanning image of a shield tunnel section, wherein the shield tunnel section is provided with l-ring segments, and each ring segment is provided with m transverse slits; recognizing a binary image of the three-dimensional laser scanning image of the shield tunnel section, and carrying out ring-by-ring slicing according to a ring seam recognition result to obtain a binary image of each ring segment; wherein the X-axis coordinate of the starting point of the binary image of the jth ring segment is
(2) Detecting all possible line segments L i by adopting a Hough transformation method on the binary image of the jth ring segment, wherein each line segment L i is represented by using starting and starting point coordinates and a weight vector:
Li＝(xi1 yi1 xi2 yi2 wi)
wherein:
w i＝(xi2-xi1)/widthj,widthj is the width of the slice of the j-th loop;
(3) Combining all detected and identified line segments L i on the jth ring segment by using a DBSCAN clustering method, wherein the line segments possibly belonging to the same transverse seam, and the clustering distance between any two line segments L i is defined as follows:
Fitting the clustered line segment sets into a new line segment by adopting a least square method, summing the weights to be used as new line segment weights, and finally obtaining a transverse seam vector matrix HS j of the j-th ring of the line segments as follows:
(4) Processing the binary images of the rest segments in the shield tunnel Section according to the methods of the step (2) and the step (3) to obtain a vector result Section for identifying the transverse seam segments of the binary images in the whole shield tunnel Section:
Section＝(HS1 … HSj … HSl)
Wherein: j=1, 2, …, l;
(5) A DBSCAN clustering method is adopted in the vector result Section of the transverse seam line segment to obtain a template SectionFM of the Section transverse seam;
(6) For the transverse seam vector matrix HS j (j=1, 2, …, l) of the j-th ring segment, the correction recognition result is matched ring by ring according to the previously acquired section transverse seam template SectionFM.
The step (5) specifically comprises the following steps:
(5.1) converting the transverse seam vector matrix HS j identified ring by ring into the following form, filtering out segments with the number of transverse seams not being m, and converting into a distance representation form HM j according to the following formula;
wherein i=1, 2 … m-1;
SectionM＝(HM1 … HMj … HMl0)
(5.2) for SectionM, obtaining a template SectionFM of the interval transverse seam by adopting a DBSCAN clustering method, wherein the clustering distance between any two HMs is defined as follows:
the threshold value of the clustering distance is set to 0.1, and the minimum sampling number is set to 10; the centroid of each cluster after clustering serves as a template SectionFM for the final interval transverse seam:
SectionFM＝(HM1 … HMi … HMk)。
The step (6) specifically comprises the following steps:
(6.1) introducing an unknown quantity Δy to convert the template SectionFM of the section transverse seam from the distance representation to the coordinate representation HMY i:
(6.2) converting the transverse seam vector matrix HS j of the jth ring segment into an expression form HSY j without X-axis coordinates:
(6.3) define HSY j distance from HMY i:
wherein:
(6.4) solving the optimally matched delta y min and interval transverse seam template number i min by taking the following formula as an objective function, and calculating a transverse seam modification_HS j after the j-th loop sheet is corrected;
and (6.5) repeating the steps (6.1) - (6.4), and finishing the transverse seam matching of all the rings in the shield tunnel section from the j-th ring segment to the n-th ring segment.
The invention has the advantages that: the accuracy of the recognition of the segment transverse seam is improved, and the defects of missed detection and false detection of the conventional deep learning model transverse seam recognition result are overcome.
Drawings
FIG. 1 is a three-dimensional laser scanned image of a shield tunnel section in accordance with the present invention;
FIG. 2 is a binary diagram of segment ring identification in the present invention;
FIG. 3 is a schematic diagram of a segment detection result on a segment ring according to the present invention;
fig. 4 is a schematic diagram of a clustering result of transverse seam segments on a segment ring in the present invention.
Detailed Description
The features of the present invention and other related features are described in further detail below by way of example in conjunction with the following drawings, to facilitate understanding by those skilled in the art:
Examples: as shown in fig. 1-4, the embodiment specifically relates to a recognition result fine processing method based on a priori rule of a transverse seam of a shield tunnel, which specifically comprises the following steps:
(1) As shown in fig. 1, a three-dimensional laser scanning image of a shield tunnel section is acquired, wherein the shield tunnel section is provided with l-ring segments, and each ring segment is provided with m transverse slits; recognizing a binary image of a three-dimensional laser scanning image of a shield tunnel section, and carrying out ring-by-ring slicing according to a ring seam recognition result to obtain a binary image of each ring segment, such as a binary image of a single ring segment shown in fig. 2; wherein the X-axis coordinate of the starting point of the binary image of the jth ring segment is
(2) As shown in fig. 1,2 and 3, the binary image of the jth ring segment is subjected to hough transform to detect all possible line segments L i, and each line segment L i is represented by using the start and start coordinates and a weight vector thereon:
Li＝(xi1 yi1 xi2 yi2 wi)
wherein:
w i＝(xi2-xi1)/widthj,widthj is the width of the slice of the j-th loop;
With respect to the start and start point coordinates (x i1,yi1),(xi2,yi2) on the line segment L i, reference can be made to the example coordinates on fig. 1, where one segment ring and the transverse seam thereon are identified on fig. 1, showing the relevant line segment start point coordinates.
(3) As shown in fig. 3 and 4, for all detected and identified line segments L i on the jth ring segment, merging line segments possibly belonging to the same transverse seam by using a DBSCAN clustering method, wherein the clustering distance between any two line segments L i is defined as follows:
Fitting the clustered line segment sets into a new line segment by adopting a least square method, summing the weights to be used as new line segment weights, and finally obtaining a transverse seam vector matrix HS j of the j-th ring of the line segments as follows:
(4) Processing the binary images of the rest segments in the shield tunnel Section according to the methods of the step (2) and the step (3) to obtain a vector result Section for identifying the transverse seam segments of the binary images in the whole shield tunnel Section:
Section＝(HS1 … HSj … HSl)
Wherein: j=1, 2, …, l.
(5) The template SectionFM of the Section transverse seam is obtained by adopting a DBSCAN clustering method in the vector result Section of the transverse seam segment, and the specific steps are as follows:
(5.1) converting the transverse seam vector matrix HS j identified ring by ring into the following form, filtering out segments with the number of transverse seams not being m, and converting into a distance representation form HM j according to the following formula;
wherein i=1, 2 … m-1;
SectionM＝(HM1 … HMj … HMl0)
(5.2) for SectionM, obtaining a template SectionFM of the interval transverse seam by adopting a DBSCAN clustering method, wherein the clustering distance between any two HMs is defined as follows:
the threshold value of the clustering distance is set to 0.1, and the minimum sampling number is set to 10; the centroid of each cluster after clustering serves as a template SectionFM for the final interval transverse seam:
SectionFM＝(HM1 … HMi … HMk)。
(6) For the transverse seam vector matrix HS j (j=1, 2, …, l) of the jth ring segment, the correction recognition result is matched ring by ring according to the previously acquired section transverse seam template SectionFM, which comprises the following steps:
(6.1) introducing an unknown quantity Δy to convert the template SectionFM of the section transverse seam from the distance representation to the coordinate representation HMY i:
(6.2) converting the transverse seam vector matrix HS j of the jth ring segment into an expression form HSY j without X-axis coordinates:
(6.3) define HSY j distance from HMY i:
wherein:
(6.4) solving the optimally matched delta y min and interval transverse seam template number i min by taking the following formula as an objective function, and calculating a transverse seam modification_HS j after the j-th loop sheet is corrected;
and (6.5) repeating the steps (6.1) - (6.4), and finishing the transverse seam matching of all the rings in the shield tunnel section from the j-th ring segment to the n-th ring segment.
Claims (2)
1. A recognition result fine processing method based on a shield tunnel transverse seam priori rule is characterized by comprising the following steps:
(1) Collecting a three-dimensional laser scanning image of a shield tunnel section, wherein the shield tunnel section is provided with l-ring segments, and each ring segment is provided with m transverse slits; recognizing a binary image of the three-dimensional laser scanning image of the shield tunnel section, and carrying out ring-by-ring slicing according to a ring seam recognition result to obtain a binary image of each ring segment; wherein the X-axis coordinate of the starting point of the binary image of the jth ring segment is
(2) Detecting all possible line segments L i by adopting a Hough transformation method on the binary image of the jth ring segment, wherein each line segment L i is represented by using starting and starting point coordinates and a weight vector:
Li＝(xi1 yi1 xi2 yi2 wi)
wherein:
wi= (xi 2-xi1)/widthj,widthj is the width of the j-th loop slice;
(3) Combining all detected and identified line segments L i on the jth ring segment by using a DBSCAN clustering method, wherein the line segments possibly belonging to the same transverse seam, and the clustering distance between any two line segments L i is defined as follows:
Fitting the clustered line segment sets into a new line segment by adopting a least square method, summing the weights to be used as new line segment weights, and finally obtaining a transverse seam vector matrix HS j of the j-th ring of the line segments as follows:
(4) Processing the binary images of the rest segments in the shield tunnel Section according to the methods of the step (2) and the step (3) to obtain a vector result Section for identifying the transverse seam segments of the binary images in the whole shield tunnel Section:
Section＝(HS1…HSj…HSl)
Wherein: j=1, 2, …, l;
(5) A DBSCAN clustering method is adopted in the vector result Section of the transverse seam line segment to obtain a template SectionFM of the Section transverse seam;
(6) For a transverse seam vector matrix HS j (j=1, 2, …, l) of the jth ring segment, the correction recognition result is matched ring by ring according to the previously acquired section transverse seam template SectionFM;
(6.1) introducing an unknown quantity Δy to convert the template SectionFM of the section transverse seam from the distance representation to the coordinate representation HMY i:
(6.2) converting the transverse seam vector matrix HS j of the jth ring segment into an expression form HSY j without X-axis coordinates:
(6.3) define HSY j distance from HMY i:
wherein:
(6.4) solving the optimally matched delta y min and interval transverse seam template number i min by taking the following formula as an objective function, and calculating a transverse seam modification_HS j after the j-th loop sheet is corrected;
and (6.5) repeating the steps (6.1) - (6.4), and finishing the transverse seam matching of all the rings in the shield tunnel section from the j-th ring segment to the n-th ring segment.
2. The method for accurately processing the recognition result based on the shield tunnel transverse seam priori rule according to claim 1, wherein the step (5) specifically comprises the following steps:
(5.1) converting the transverse seam vector matrix HS j identified ring by ring into the following form, filtering out segments with the number of transverse seams not being m, and converting into a distance representation form HM j according to the following formula;
wherein i=1, 2 … m-1;
SectionM＝(HM1…HMj…HMl0)
(5.2) for SectionM, obtaining a template SectionFM of the interval transverse seam by adopting a DBSCAN clustering method, wherein the clustering distance between any two HMs is defined as follows:
the threshold value of the clustering distance is set to 0.1, and the minimum sampling number is set to 10; the centroid of each cluster after clustering serves as a template SectionFM for the final interval transverse seam:
SectionFM＝(HM1…HMi…HMk)。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111532466.2A CN114267039B (en) | 2021-12-15 | 2021-12-15 | Recognition result fine processing method based on shield tunnel transverse seam priori rules |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111532466.2A CN114267039B (en) | 2021-12-15 | 2021-12-15 | Recognition result fine processing method based on shield tunnel transverse seam priori rules |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114267039A CN114267039A (en) | 2022-04-01 |
| CN114267039B true CN114267039B (en) | 2024-06-11 |
Family
ID=80827221
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111532466.2A Active CN114267039B (en) | 2021-12-15 | 2021-12-15 | Recognition result fine processing method based on shield tunnel transverse seam priori rules |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114267039B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116620360B (en) * | 2023-05-17 | 2024-04-23 | 中建三局信息科技有限公司 | Rail car positioning system and method |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2005194837A (en) * | 2004-01-09 | 2005-07-21 | Shimizu Corp | Widening method for tunnel |
| CN104457572A (en) * | 2014-12-04 | 2015-03-25 | 上海岩土工程勘察设计研究院有限公司 | Method for extracting shield tunnel staggered joint and dislocation quantity on basis of three-dimensional scanning technology |
| CN111486797A (en) * | 2019-01-29 | 2020-08-04 | 上海京海工程技术有限公司 | Automatic extraction method for transverse diameter of subway circular shield tunnel |
| CN112036508A (en) * | 2020-09-27 | 2020-12-04 | 上海京海工程技术有限公司 | Automatic circular seam identification method based on shield tunnel lining structure |
| AU2020103105A4 (en) * | 2020-10-29 | 2021-01-07 | Nanjing Forestry University | A Method for Segmenting Point Cloud Data on Shield Tunnel Surface |
-
2021
- 2021-12-15 CN CN202111532466.2A patent/CN114267039B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2005194837A (en) * | 2004-01-09 | 2005-07-21 | Shimizu Corp | Widening method for tunnel |
| CN104457572A (en) * | 2014-12-04 | 2015-03-25 | 上海岩土工程勘察设计研究院有限公司 | Method for extracting shield tunnel staggered joint and dislocation quantity on basis of three-dimensional scanning technology |
| CN111486797A (en) * | 2019-01-29 | 2020-08-04 | 上海京海工程技术有限公司 | Automatic extraction method for transverse diameter of subway circular shield tunnel |
| CN112036508A (en) * | 2020-09-27 | 2020-12-04 | 上海京海工程技术有限公司 | Automatic circular seam identification method based on shield tunnel lining structure |
| AU2020103105A4 (en) * | 2020-10-29 | 2021-01-07 | Nanjing Forestry University | A Method for Segmenting Point Cloud Data on Shield Tunnel Surface |
Non-Patent Citations (1)
| Title |
|---|
| 基于盾构隧道环缝点云的中轴线提取方法;陈林;《河海大学学报(自然科学版)》;20200131;第48卷(第01期);第7-78页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114267039A (en) | 2022-04-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110543878B (en) | Pointer instrument reading identification method based on neural network | |
| CN109800648B (en) | Face detection and recognition method and device based on face key point correction | |
| CN111611643B (en) | Household vectorization data acquisition method and device, electronic equipment and storage medium | |
| CN101561866B (en) | Character recognition method based on SIFT feature and gray scale difference value histogram feature | |
| CN107392141B (en) | Airport extraction method based on significance detection and LSD (least squares distortion) line detection | |
| CN109410238B (en) | Wolfberry identification and counting method based on PointNet + + network | |
| CN111274915B (en) | Deep local aggregation descriptor extraction method and system for finger vein image | |
| CN111539330B (en) | Transformer substation digital display instrument identification method based on double-SVM multi-classifier | |
| CN110929713B (en) | Steel seal character recognition method based on BP neural network | |
| CN108509950B (en) | Railway contact net support number plate detection and identification method based on probability feature weighted fusion | |
| CN103473545B (en) | A kind of text image method for measuring similarity based on multiple features | |
| CN111310622A (en) | Fish swarm target identification method for intelligent operation of underwater robot | |
| CN109242047A (en) | Bank card number detection and recognition methods based on K-means++ cluster and residual error network class | |
| CN114267039B (en) | Recognition result fine processing method based on shield tunnel transverse seam priori rules | |
| CN111950556A (en) | License plate printing quality detection method based on deep learning | |
| US10223587B2 (en) | Pairing of images of postal articles with descriptors of singularities of the gradient field | |
| CN107729863B (en) | Human finger vein recognition method | |
| CN114332172A (en) | Improved laser point cloud registration method based on covariance matrix | |
| CN116704270A (en) | Intelligent equipment positioning marking method based on image processing | |
| CN111145314A (en) | Method for extracting place name symbol of scanning electronic map by combining place name labeling | |
| CN114926635B (en) | Target segmentation method in multi-focus image combined with deep learning method | |
| CN114898347A (en) | Machine vision identification method for pointer instrument | |
| CN104766085B (en) | A kind of multiple dimensioned pattern recognition method | |
| CN115100733A (en) | RFID gesture recognition method, computer device, product and storage medium | |
| CN114863163A (en) | Method and system for cell classification based on cell image |
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 | ||
| CB02 | Change of applicant information |
Country or region after: China Address after: 200093 No. 38 Shui Feng Road, Yangpu District, Shanghai. Applicant after: Shanghai Survey, Design and Research Institute (Group) Co.,Ltd. Address before: 200093 No. 38 Shui Feng Road, Yangpu District, Shanghai. Applicant before: SGIDI ENGINEERING CONSULTING (Group) Co.,Ltd. Country or region before: China |
|
| CB02 | Change of applicant information | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |