CN116958089A - Petroleum pipeline crack detection method based on dual-attention mechanism - Google Patents
Petroleum pipeline crack detection method based on dual-attention mechanism Download PDFInfo
- Publication number
- CN116958089A CN116958089A CN202310913981.8A CN202310913981A CN116958089A CN 116958089 A CN116958089 A CN 116958089A CN 202310913981 A CN202310913981 A CN 202310913981A CN 116958089 A CN116958089 A CN 116958089A
- Authority
- CN
- China
- Prior art keywords
- crack
- image
- network
- petroleum pipeline
- attention mechanism
- 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.)
- Pending
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 117
- 239000003208 petroleum Substances 0.000 title claims abstract description 92
- 230000007246 mechanism Effects 0.000 title claims abstract description 62
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 48
- 238000012545 processing Methods 0.000 claims description 44
- 238000000034 method Methods 0.000 claims description 34
- 238000012549 training Methods 0.000 claims description 23
- 238000001914 filtration Methods 0.000 claims description 12
- 230000009977 dual effect Effects 0.000 claims description 7
- 230000011218 segmentation Effects 0.000 claims description 6
- 238000012360 testing method Methods 0.000 claims description 6
- 238000013519 translation Methods 0.000 claims description 6
- 230000002159 abnormal effect Effects 0.000 claims description 5
- 238000004590 computer program Methods 0.000 claims description 5
- 238000010200 validation analysis Methods 0.000 claims description 3
- 238000013527 convolutional neural network Methods 0.000 abstract description 4
- 238000000354 decomposition reaction Methods 0.000 abstract description 4
- 238000010586 diagram Methods 0.000 description 8
- 238000005070 sampling Methods 0.000 description 8
- 230000006870 function Effects 0.000 description 7
- 238000003708 edge detection Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 230000007547 defect Effects 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 239000004744 fabric Substances 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- KLDZYURQCUYZBL-UHFFFAOYSA-N 2-[3-[(2-hydroxyphenyl)methylideneamino]propyliminomethyl]phenol Chemical compound OC1=CC=CC=C1C=NCCCN=CC1=CC=CC=C1O KLDZYURQCUYZBL-UHFFFAOYSA-N 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 201000001098 delayed sleep phase syndrome Diseases 0.000 description 1
- 208000033921 delayed sleep phase type circadian rhythm sleep disease Diseases 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000011049 filling Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000003707 image sharpening Methods 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 239000013598 vector Substances 0.000 description 1
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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4038—Image mosaicing, e.g. composing plane images from plane sub-images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/762—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using clustering, e.g. of similar faces in social networks
- G06V10/763—Non-hierarchical techniques, e.g. based on statistics of modelling distributions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2200/00—Indexing scheme for image data processing or generation, in general
- G06T2200/32—Indexing scheme for image data processing or generation, in general involving image mosaicing
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Medical Informatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Probability & Statistics with Applications (AREA)
- Quality & Reliability (AREA)
- Computer Graphics (AREA)
- Geometry (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a petroleum pipeline crack detection method based on a dual-attention mechanism, which comprises the following steps: acquiring an internal image of a petroleum pipeline to be detected; constructing a crack detection model based on a dual-attention mechanism and a yolov5-seg network; performing crack detection on the internal image by using a crack detection model to determine a crack region; clustering and segmenting the crack image of the crack region by using a density clustering algorithm, and fitting the geometric form of the crack; extracting crack contours and skeletons, splicing the crack images by using an image splicing algorithm, and reconstructing a crack network by using a corner detection algorithm; and 3D modeling is carried out on the cracks in the pipeline according to the position information corresponding to the crack images in the crack network. Therefore, the invention can realize the accurate detection of the pipeline cracks through the convolutional neural network with the double-attention mechanism, and perform cluster decomposition on complex cracks, accurately extract crack parameters, reduce the workload and effectively improve the accuracy of the crack identification of the petroleum pipeline.
Description
Technical Field
The invention relates to the technical field of pipeline defect detection, in particular to a petroleum pipeline crack detection method based on a dual-attention mechanism.
Background
Currently, the petrochemical industry is rapidly developing, and petroleum long-distance pipelines are widely used with the advantages of high efficiency and low cost. However, when the pipeline fails, the oil gas leakage can bring huge losses to the country. Defects in petroleum pipelines seriously jeopardize pipeline safety, so that the detection of defects in petroleum pipelines becomes an important research topic, wherein crack detection is an important ring in pipeline maintenance.
Early-stage petroleum pipeline crack detection mainly adopts human eyes to observe pipeline crack images, and manually records and counts crack conditions, so that the problems of low efficiency, great influence of acceptance and the like exist. With the development of image processing technology, methods for manually counting pipeline cracks have been eliminated. The existing method for realizing pipeline crack detection based on the traditional image processing technology is to process a crack image according to different gray values of a crack region and a picture background region through image processing technologies such as image binarization, image filtering, image enhancement and the like to obtain a crack detection result; however, the method has complicated processing flow, low accuracy rate for detecting tiny cracks, frequent missed detection and poor crack detection effect for small gray value change of the crack existence area and the background area; and only shallow detection is carried out on the crack, no crack morphology is carried out, the crack parameters are deeply analyzed, the crack detection accuracy is relatively low, and the time consumption is long.
Disclosure of Invention
In order to solve the problems, the invention provides a petroleum pipeline crack detection method, a petroleum pipeline crack detection device, a petroleum pipeline crack detection terminal and a petroleum pipeline crack detection storage medium based on a dual-attention mechanism.
The technical scheme of the invention is realized as follows:
a petroleum pipeline crack detection method based on a dual-attention mechanism, the method comprising:
s101: acquiring an internal image of a petroleum pipeline to be detected, and recording position information corresponding to the internal image;
s102: constructing a crack detection model of the petroleum pipeline based on a dual-attention mechanism and a yolov5-seg network;
s103: performing crack detection on an internal image of the petroleum pipeline to be detected by using the crack detection model, and determining a crack area;
s104: clustering and segmenting crack images included in the crack region by using a density clustering algorithm, and fitting the geometric form of the crack;
s105: extracting crack contours and crack skeletons in the crack images, splicing the crack images by using an image splicing algorithm, and reconstructing a crack network by using an angular point detection algorithm;
s106: and 3D modeling is carried out on the pipeline according to the position information corresponding to the crack image in the crack network, and the crack network morphology is visually displayed.
In some embodiments, the yolov5-seg network in step S102 includes a backbone network, a neck network, and a head network, the dual attention mechanism includes a channel attention mechanism and a spatial attention mechanism, and step S102 includes:
adding a channel attention mechanism at the connection part of the backbone network and the neck network, and adding a space attention mechanism behind the backbone network to improve the yolov5-seg network;
and constructing the crack detection model by adopting the improved yolov5-seg network as a main network.
In some embodiments, the step S103 further includes:
acquiring crack images in the petroleum pipeline to form an original image set;
expanding the original image concentrated crack image through cutting, rotation and translation;
dividing the expanded original image set into the training set and the testing set according to a certain proportion;
marking the crack areas of the crack images in the training set by using image marking software;
training the crack detection model by using the marked training set;
validating the trained crack detection model using the validation set;
and carrying out crack detection on the internal image of the petroleum pipeline to be detected by using the verified crack detection model, and determining the crack area.
In some embodiments, the step S104 further includes:
extracting pixel point information of a crack image included in the crack region, and selecting a proper gray threshold to perform binarization processing on the crack image;
and clustering and segmenting the binarized crack image by using the density clustering algorithm, and fitting the geometric form of the crack.
In some embodiments, the step S105 further includes:
filtering abnormal pixel points in the crack area through Gaussian filtering;
extracting a crack contour and a crack skeleton of the crack image included in the filtered crack region;
based on the crack contour and the crack skeleton, splicing the same crack on different crack images included in the crack region by using an image splicing algorithm, and reconstructing the crack network by using a corner detection algorithm.
In some embodiments, the extracting the crack contours and crack skeletons of the crack images included in the filtered crack region comprises:
extracting a crack contour of the crack region according to the central pixel gray value and other surrounding pixel gray values of the crack region;
extracting the crack skeleton of the crack region by using a Zhang-Suen refinement algorithm.
In some embodiments, the step S106 includes:
and constructing a three-dimensional space coordinate system by taking the tangent point of the outer wall of the petroleum pipeline and the plane as a reference coordinate point according to the position information corresponding to the crack image, modeling the whole petroleum pipeline in 3D, and simultaneously carrying out three-dimensional display on the distribution of the crack network on the petroleum pipeline in the space coordinate system.
The embodiment of the invention also provides an intelligent detection device for the crack of the petroleum pipeline based on the dual-attention mechanism, which comprises the following components:
the acquisition module is used for acquiring an internal image of the petroleum pipeline to be detected and recording position information corresponding to the internal image;
the processing module is used for constructing a crack detection model of the petroleum pipeline based on the dual attention mechanism and the yolov5-seg network;
the processing model is also used for carrying out crack detection on the internal image of the petroleum pipeline to be detected by utilizing the crack detection model, and determining a crack area;
the processing module is further used for carrying out clustering segmentation on the crack images included in the crack region by utilizing a density clustering algorithm, and fitting the geometric form of the crack;
the processing module is also used for extracting crack contours and crack frameworks in the crack images, splicing the crack images by using an image splicing algorithm and reconstructing a crack network by using an angular point detection algorithm;
And the output module is used for carrying out 3D modeling on the pipeline according to the position information corresponding to the crack image in the crack network, and visually displaying the crack network morphology.
The embodiment of the invention also provides a terminal, which comprises a processor and a memory for storing a computer program capable of running on the processor; the processor is used for realizing the petroleum pipeline crack detection method based on the dual-attention mechanism according to any embodiment of the invention when running a computer program.
The embodiment of the invention also provides a storage medium, wherein the storage medium is provided with computer executable instructions, and the method is characterized in that the computer executable instructions are executed by a processor to realize the petroleum pipeline crack detection method based on the dual-attention mechanism.
The embodiment of the invention provides a petroleum pipeline crack detection method based on a dual-attention mechanism, which is used for collecting an internal image of a petroleum pipeline to be detected; constructing a crack detection model of the petroleum pipeline based on a dual-attention mechanism and a yolov5-seg network; performing crack detection on the internal image by using a crack detection model to determine a crack region; clustering and segmenting the crack image of the crack region by using a density clustering algorithm, and fitting the geometric form of the crack; extracting outline features of cracks, splicing the crack images by using an image splicing algorithm, and reconstructing a crack network by using a corner detection algorithm; and 3D modeling is carried out on the spatial distribution of the cracks in the pipeline according to the position information corresponding to the crack images in the crack network. Therefore, the invention can realize the rapid and accurate detection of the pipeline cracks through the convolutional neural network with the double-attention mechanism, and perform cluster decomposition on complex cracks, accurately extract crack parameters, reduce the workload and effectively improve the accuracy of crack identification of petroleum pipelines.
Drawings
The invention will be further described with reference to the accompanying drawings and examples, in which:
FIG. 1 is a schematic flow chart of a petroleum pipeline crack detection method based on a dual-attention mechanism in an embodiment of the invention;
FIG. 2 is a schematic diagram of a yolov5-seg network of a petroleum pipeline crack detection method based on a dual-attention mechanism in an embodiment of the invention;
FIG. 3 is a schematic diagram of deformation convolution of an intelligent detection method for cracks of petroleum pipelines based on a dual-attention mechanism in an embodiment of the invention;
FIG. 4 is a schematic diagram of a network structure of a dual-attention mechanism based intelligent detection method for cracks of petroleum pipelines based on the dual-attention mechanism in an embodiment of the invention;
FIG. 5 is a schematic diagram of a petroleum pipeline crack clustering process of an intelligent petroleum pipeline crack detection method based on a dual-attention mechanism in an embodiment of the invention;
FIG. 6 illustrates a pixel region of an image of an intelligent detection method for cracks of a petroleum pipeline based on a dual-attention mechanism in an embodiment of the invention;
FIG. 7 is a schematic diagram of an intelligent detection device for cracks of petroleum pipelines based on a dual-attention mechanism in an embodiment of the invention;
fig. 8 is a schematic diagram of a terminal hardware structure of a petroleum pipeline crack detection method based on a dual-attention mechanism in an embodiment of the invention.
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to provide a clearer understanding of the technical features, objects and effects of the present invention. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
In the following description, suffixes such as "module", "component", or "unit" for representing elements are used only for facilitating the description of the present invention, and have no specific meaning per se. Thus, "module," "component," or "unit" may be used in combination.
Referring to fig. 1, the embodiment of the invention provides a petroleum pipeline crack detection method based on a dual-attention mechanism, which specifically comprises the following steps:
s101: acquiring an internal image of a petroleum transportation pipeline to be detected, and recording position information corresponding to the internal image;
s102: constructing a crack detection model of the petroleum pipeline based on a dual-attention mechanism and a yolov5-seg network;
s103: performing crack detection on an internal image of the petroleum pipeline to be detected by using the crack detection model, and determining a crack area;
s104: clustering and segmenting crack images included in the crack region by using a density clustering algorithm, and fitting the geometric form of the crack;
S105: extracting crack contours and crack skeletons in the crack images, splicing the crack images by using an image splicing algorithm, and reconstructing a crack network by using an angular point detection algorithm;
s106: and 3D modeling is carried out on the pipeline according to the position information corresponding to the crack image in the crack network, and the crack network morphology is visually displayed.
The method of the embodiment of the invention is executed by the terminal. The terminals may be various types of terminals; for example, the terminal may be, but is not limited to being, at least one of: a server, computer, tablet, or other electronic device.
In some embodiments, the step S101 includes: and operating the periscope to collect internal images of the pipeline at a constant speed in the petroleum pipeline to be detected, and simultaneously recording position information corresponding to each internal image.
The periscope comprises a terminal control system, a camera system, video-to-image software and an image storage system; the running speed of the periscope is controlled by the terminal control system, the internal image of the petroleum pipeline is obtained by the camera system, the internal image of the pipeline is converted into a picture by using video-to-image software, and the image storage system stores video image files.
Here, the internal image includes a crack image and a normal image of the inside of the petroleum transportation pipeline.
Here, the position information includes depth information and coordinate information; the depth information is the depth of the position of the internal image in the pipeline, and the coordinate information is the specific position of the internal image on the pipeline wall of the pipeline at the depth.
It can be understood that the periscope can be used for integrally collecting and classifying the internal images of the whole petroleum pipeline leaving the factory, so that the corresponding position information of each internal image in the pipeline can be extracted for subsequent quality grading and modeling.
In some embodiments, the yolov5-seg network in step S102 includes a backbone network, a neck network, and a head network, the dual attention mechanism includes a channel attention mechanism and a spatial attention mechanism, and step S102 includes:
s102a: adding a channel attention mechanism at the connection part of the backbone network and the neck network, and adding a space attention mechanism behind the backbone network to improve the yolov5-seg network;
s102b: and constructing the crack detection model by adopting the improved yolov5-seg network as a main network.
In one embodiment, the step S102a includes:
s102a1: adding an example segmentation function on the basis of a convolutional neural network yolov5 network to obtain a yolov5-seg network; the convolutional neural network yolov5 network comprises a backbone network, a neck network and a head network;
s102a2: adopting a Darknet53 light-weight network as a backbone network of the yolov5-seg network, and introducing an Fcous module into the backbone network for slicing the internal image;
s102a3: introducing a cavity convolution module SPP after the yolov5-seg network Fplus module obtained in the step 102a2 is used for extracting the characteristics of the internal image after slicing;
s102a4: and adding a channel attention mechanism at a connection channel of the backbone network and the neck network of the yolov5-seg network obtained in the step 102a3, and adding a space attention mechanism after the backbone network to obtain an improved yolov5-seg network.
It should be noted that the yolov5 network may divide the whole image into several grids, and each grid predicts the type and position information of the object in the grid. Thus, the yolov5 network with added instance segmentation functionality can sort pixels on an image one by one.
It should be noted that, the yolov5 network may perform screening of the target frame according to the IOU value between the prediction frame and the real frame, and finally output the category and the position information of the prediction frame.
In one implementation, the cavity convolution module SPP of the yolov5-seg network is used for extracting features of images with different scales, and the frame regression loss function is a GIOU function expressed as:
wherein A is a real frame, B is a prediction frame, A c For the minimum closure region, U is the union of the prediction and real frames.
Exemplary, yolov5-seg network structures, as shown in FIG. 2, include three parts: backbone network dark 53, neck fabric, and head fabric. The backbone network dark 53 is responsible for feature extraction, and comprises an Fplus module and an SPP module, wherein the Fplus module takes a plurality of image slices once every other pixel, and features of the image slices extracted by the SPP module obtain 4 feature layer structures; the neck network is responsible for feature fusion and comprises an FPN module from top to bottom and a PAN module from bottom to top, wherein the FPN module transmits deep semantic features to a shallow layer, and the PAN module transmits shallow layer strong positioning features to the deep layer; the head network is responsible for outputting detection results, comprises three detection heads, respectively outputs detection information of cracks with different scales, and finally obtains a detection result.
Therefore, the embodiment of the invention can obtain a plurality of smaller image slices by slicing the images based on the introduced Fplus module, and greatly optimize the parameters, thereby enhancing the accuracy of crack detection of the yolov5-seg network; meanwhile, based on the added cavity convolution structure SPP, the characteristics of the obtained image slice can be extracted to obtain a plurality of characteristic layer structures, and then the receptive field is amplified.
In some embodiments, the step S103 further includes:
s103a: acquiring crack images in the petroleum pipeline to form an original image set;
s103b: expanding the original image concentrated crack image through cutting, rotation and translation;
s103c: dividing the expanded original image set into the training set and the testing set according to a certain proportion;
s103d: manually marking the crack areas of the crack images in the training set by using image marking software;
s103e: training the crack detection model by using the marked training set;
s103f: validating the trained crack detection model using the validation set;
s103g: and carrying out crack detection on the internal image of the petroleum pipeline to be detected by using the verified crack detection model, and determining the crack area.
It will be appreciated that the image of the crack acquired in step S103a is not a normal image without a crack. Step S103a is specifically configured to acquire a crack image of a petroleum pipeline and position information corresponding to each crack image in the pipeline, so as to form an original image set; the pipe position information includes depth information and coordinate information including an abscissa and an ordinate.
In one embodiment, the steps S103b to S103d specifically include: performing data expansion on the original image concentrated crack image according to the uniform size through cutting, rotation and translation; dividing the original image set after data expansion and feature enhancement into the training set and the testing set according to a certain proportion; and manually marking the crack region of the images in the training set by using image marking software.
Here, the image labeling software may be Labelme.
For example, the obtained original image set includes 400 images, the sizes of the images are unified to 608 x 608, and the original image set is expanded by 5 times through equal step cutting, rotation and translation to obtain 2000 images; dividing an original image set comprising 2000 crack images into a training set and a testing set according to the proportion of 8:2; finally, labelme image labeling software is used for manually labeling the crack areas of the images in the training set.
In one embodiment, the step S103e further includes: a morph convolution is employed in the final stage of the backbone network, i.e., the dark 53 network.
For example, as shown in the schematic diagram of the centroid convolution structure in fig. 3, the gray dots are original sampling points, the arrows are the offsets, and are vectors, and the black dots are offset sampling points after the offset change, where the sampling points correspond to a pixel point on the internal image. The training process is understood to mean that an offset is set to the original sampling point, by which the original sampling point is changed to the offset sampling point. Therefore, according to the embodiment of the invention, based on the deformation convolution, an offset is set for the sampling points on each feature map, and according to the learning features of the offset pertinence, the information of cracks with different regular shapes is fully extracted, so that the loss of details is prevented.
In one implementation, the step S103f further includes: and verifying the trained crack detection model by using the verification set to judge whether the crack detection model is converged, and if so, taking the verified crack recognition model as a final crack recognition model for detecting an image to be detected.
The determining whether the crack detection model converges may specifically be determining that
Therefore, the embodiment of the invention can be based on training the crack detection model and verifying the effectiveness of the crack detection model, and optimize the tuning to achieve the convergence of the crack detection model.
In another embodiment, the step S103f further includes: if not, returning to step S103e to continue model training.
In some embodiments, the step S103g further includes: inputting an internal image of the petroleum pipeline to be detected into the crack detection model; the internal image determines a crack image with a crack in the internal image through a dual-attention mechanism of the crack detection model; and determining the position of the crack in the crack image as the crack area.
It is understood that the internal image corresponding to the crack region is a crack image, and the internal images except for the crack image are normal images, and the normal images have no crack.
It will be appreciated that the crack is typically a net-like structure and that one image or picture may not fully reveal a complete crack, and that the crack region detected by the crack detection model is an image that includes all of the cracks.
Illustratively, in the network structure shown in fig. 2, a channel attention mechanism is added at the connection part of the backbone network and the neck network, a space attention mechanism is added after the backbone network, and the feature extraction network of yolov5-seg is improved, so that the network structure of the dual attention mechanism shown in fig. 4 is formed. In fig. 4, the input image is the internal image to be detected, and the pixel points of the crack region can be emphasized through a spatial attention mechanism and a channel attention mechanism, and the pixels of the background region except the crack region are restrained, so that the feature extraction of the internal image is completed, and the detected information is transmitted backwards.
Therefore, the embodiment of the invention can emphasize the pixel points of the crack region in the crack image and inhibit the pixels of the background region based on the crack detection model added with the dual-attention mechanism, so that the detected information can be transmitted backwards while the characteristics of the crack image are extracted, the important information is reserved, the invalid information is inhibited, the generalization capability of a detection network can be increased, and the detection precision is improved.
When the pipeline crack detection is carried out, the crack area is the object of important attention, the calculation amount is increased, the time consumption is increased when the whole feature map is analyzed, the crack detection model which is integrated with the attention mechanism detects the internal image of the petroleum pipeline to be detected, and the performance of the crack detection model is greatly improved.
In some embodiments, the step S104 includes:
s104a: extracting pixel point information of a crack image included in the crack region, and selecting a proper gray threshold to perform binarization processing on the crack image;
s104b: and clustering and segmenting the binarized crack image by using the density clustering algorithm, and fitting the geometric form of the crack.
Here, the geometry of the crack includes, but is not limited to: geometric parameter information, length and width.
Here, the pixel point information is pixel point information.
Here, the crack image after the binarization processing is a binarized image.
Illustratively, based on pixel point information of the crack region, selecting a proper gray threshold to perform binarization processing on the crack region, separating the crack region from a background region, and determining a binary image corresponding to the crack image; defining core points and a minimum radius Eps based on the separated distribution density of the core points in the crack region, and setting a proper core point number threshold MinPts; and separating different cracks in the crack area into different clusters by using the density clustering algorithm, carrying out cluster decomposition, and fitting the geometric form of each crack.
In one embodiment, the density clustering algorithm adopts a DBSCAN clustering algorithm, and the step S104b further includes: and constructing a crack expansion clustering model by using a DBSCAN clustering algorithm based on the coordinate information corresponding to the pixel points and the binarized image after binarization processing.
Exemplary, as shown in a schematic diagram of a petroleum pipeline crack region expansion clustering process in fig. 5, a crack geometry obtained after expansion clustering of a petroleum pipeline crack region on the right side is obtained from a binary image on the left side through the constructed crack expansion clustering model, a white region and a gray region are crack regions, the gray region is a cluster of pixels formed by clustering and splitting, and the black region is a background region.
Therefore, the embodiment of the invention can combine the clustering algorithm with the random sampling consistency algorithm to perform clustering decomposition on complex bonding cracks, thereby improving the accuracy of crack identification. Meanwhile, the image stitching method is adopted, and the contour detection algorithm is combined, so that the situation that the same crack of a plurality of images is judged to be a plurality of cracks in the prior art is optimized in a targeted mode, and the problems are successfully solved.
In some embodiments, the step S105 includes:
S105a: filtering abnormal pixel points in the crack area through Gaussian filtering;
s105b: extracting a crack contour and a crack skeleton of the crack image included in the filtered crack region;
s105c: based on the crack contour and the crack skeleton, splicing the same crack on different crack images included in the crack region by using an image splicing algorithm, and reconstructing the crack network by using a corner detection algorithm.
Here, the gaussian filtering uses a two-dimensional gaussian function, which can be expressed as:
wherein x is the abscissa corresponding to the pixel point, y is the ordinate corresponding to the pixel point, and sigma is the index difference.
In some embodiments, the step S105b further includes:
s105b1: extracting a crack contour of the crack region according to the central pixel gray value and other surrounding pixel gray values of the crack region;
s105b2: extracting the crack skeleton of the crack region by using a Zhang-Suen refinement algorithm.
In one embodiment, the step S105b1 includes: determining difference information of the crack region and the background region according to the pixel gray value of the crack region and the pixel gray value of other surrounding regions, namely the background region; completing edge detection of a crack image according to the differential information, and determining a crack contour of a crack area; the method comprises the following steps:
Judging the gray value of the central pixel point of the crack image and the gray values of other surrounding pixel points, and if the gray value of the central pixel point is higher, improving the gray value of the central pixel point; otherwise, the gray value of the central pixel point is reduced, and the image sharpening operation is realized;
calculating a four-direction gradient or an eight-direction gradient in the field of the central pixel point, and adding the gradients to perform gradient operation so as to judge the relation between the gray value of the central pixel point and the gray values of other pixel points in the field;
and adjusting the gray value of the pixel point according to the result of the gradient operation, and finishing the edge detection of the crack image.
For example, in the crack geometry obtained after the clustering analysis on the right side in fig. 5, the crack area is a white part, the background area is a black part, the gray value corresponding to white is 255, and the gray value corresponding to black is 0, so that the difference between the pixel gray value of the crack area and the pixel gray value of the background area is very large, and a differential information 255 is formed, and the edge of the crack area is found according to the differential information.
The step S105b2 is specifically exemplified by: assuming a 3*3 area in the petroleum pipeline fracture image, defining a central pixel point as P1, determining the positions of eight adjacent areas P2, P3, … and P9 (refer to FIG. 6), and determining whether to delete the P1 point according to the actual situation of the P1 point adjacent area in the petroleum pipeline fracture binarization image by adopting a Zhang-Suen refinement algorithm, wherein the specific steps are as follows:
In the first step, the background is black, the value is 0, and the pixel value of the foreground object to be thinned is 1. All foreground pixel points are circulated, and the pixel points meeting the first condition are marked as deleted; the first condition includes: 2< = N (P1) < = 6, s (P1) = 1, p2×p4×p6=0, p4×p6×p8=0; n (P1) represents the number of 8 pixels adjacent to P1, the pixel value being 1, S (P1) represents the cumulative number of occurrences from 0 to 1 among the pixels P2 to P9, wherein 0 represents the background and 1 represents the foreground;
second, marking the pixel P1 meeting the second condition as deleted; the second condition includes: 2< = N (P1) < = 6, s (P1) = 1, p2×p4×p8=0, p2×p6×p8=0;
and thirdly, the two steps are circulated until no pixel is marked to be deleted in the two steps, and the output result is the skeleton after the binary image is thinned.
In one embodiment, the step S105c includes:
after the edge detection of the crack image is completed, determining the corner points of the crack outline according to the curvature of the crack outline;
judging whether the distance from the corner point of the crack in the crack image to the image edge is smaller than 25 pixel points, if so, reserving the corner point, and if so, deleting the corner point;
And comparing the horizontal distances between the corner points of the cracks in pairs, if the horizontal distances are smaller than 25 pixel points, reserving the corner points, and connecting the corner points with the corner points to realize the splicing and reconstruction of a crack network.
Illustratively, the extracting the corner points of the crack contour according to the curvature of the crack contour specifically includes: filling fine gaps in the binarized edge contour according to the obtained edge detection result, and marking T-shaped corner points appearing in the edge contour; calculating the curvature of a pixel point on the edge, and if the curvature value of a certain point exceeds a given curvature threshold value and the absolute value of the curvature is extremely large in a certain local adjacent range, judging the point as a candidate angular point; tracking each pixel point in the candidate corner point set under a small scale, accurately positioning the position of the corner point, and improving the positioning accuracy of the corner point; searching whether the corner points positioned in the third step exist near the T-shaped corner points marked in the first step, deleting the T-shaped corner points if the corner points exist, and judging that the rest corner points are the finally selected corner point sets.
Illustratively, filtering some abnormal pixel points in the crack area, namely noise points, through Gaussian filtering, and utilizing differential information generated by pixel gray values in a specific area to finish edge detection of an image; and extracting corner points of the contour based on the curvature of the contour, determining crack corner points by utilizing the curvature of the crack contour, and realizing the splicing and reconstruction of a crack network.
In one embodiment, the step S105c further includes: according to the reconstructed crack contour, solving a maximum inscribed circle for each pixel point of the crack region, and taking the diameter of the maximum inscribed circle as the crack width;
determining the ratio of the number of pixels in the crack area to the number of pixels in the image inside the whole pipeline to be detected as crack coverage rate;
and the crack width and the crack coverage rate form parameter information of the crack, and are used for grading the petroleum pipeline and performing 3D modeling.
Here, the calculation formula of the crack coverage can be expressed as:
wherein S is C The number of the pixel points in the whole pipeline crack area is indicated, and the number of the pixel points in the whole pipeline image is indicated by S.
Here, the petroleum pipeline may be classified into three grades, i.e., good grade and inferior grade, and the grade classification standard table may be referred to as table 1, as follows:
TABLE 1 pipeline quality rating criteria
Therefore, the embodiment of the invention can deeply extract crack parameters such as the maximum width and the coverage rate of the crack, and evaluate the quality of the petroleum pipeline delivered from the factory according to the coverage rate of the crack, so that the quality rating is given, and compared with the prior art, the shallow identification of the crack only can be realized, thereby greatly enriching the crack information.
In some embodiments, the step S106 includes: and constructing a three-dimensional space coordinate system by taking the tangent point of the outer wall of the petroleum pipeline and the plane as a reference coordinate point according to the position information corresponding to the crack image, modeling the whole petroleum pipeline in 3D, and simultaneously carrying out three-dimensional display on the distribution of the crack network on the petroleum pipeline in the space coordinate system.
Illustratively, modeling the spatial distribution of cracks within a pipeline by 3D modeling techniques includes: and 3D modeling is carried out on the cracks in the pipeline according to the reconstructed crack network, the crack images in the crack network and the corresponding geometric form and position information of the crack images. Therefore, the embodiment of the invention can adopt a 3D modeling technology to perform 3D modeling on the whole petroleum pipeline and the crack region, and more intuitively shows the distribution form of the crack region on the space of the whole petroleum pipeline.
Referring to fig. 7, the embodiment of the invention further provides a petroleum pipeline crack detection device based on a dual-attention mechanism, which comprises: the device comprises an acquisition module 201, a processing module 202 and an output module 203; wherein,,
the acquiring module 201 is configured to acquire an internal image of a petroleum pipeline to be detected, and record position information corresponding to the internal image;
The processing module 202 is configured to construct a crack detection model of the petroleum pipeline based on a dual-attention mechanism and a yolov5-seg network;
the processing module 202 is also configured to perform crack detection on an internal image of the petroleum pipeline to be detected by using the crack detection model, so as to determine a crack area;
the processing module 202 is further configured to perform cluster segmentation on a crack image included in the crack region by using a density clustering algorithm, and fit a geometric form of the crack;
the processing module 202 is further configured to extract a crack contour and a crack skeleton in the crack image, splice the crack image by using an image stitching algorithm, and reconstruct a crack network by using a corner detection algorithm;
the output module 203 is configured to perform 3D modeling on the pipeline according to the position information corresponding to the crack image in the crack network, and visually display the crack network morphology.
In some embodiments, the method further comprises:
the processing module is used for adding a channel attention mechanism at the joint of the backbone network and the neck network and adding a space attention mechanism after the backbone network to improve the yolov5-seg network;
the processing module is used for constructing the crack detection model by adopting the improved yolov5-seg network as a main network.
In some embodiments, the method further comprises:
the processing module is used for acquiring crack images in the petroleum pipeline to form an original image set;
the processing module is used for expanding the concentrated crack image of the original image through cutting, rotation and translation;
the processing module is used for dividing the expanded original image set into the training set and the testing set according to a certain proportion;
the processing module is used for marking the crack areas of the crack images in the training set by using image marking software;
the processing module is used for training the crack detection model by using the marked training set;
the processing module is used for verifying the trained crack detection model by using the verification set;
and the processing module is used for carrying out crack detection on the internal image of the petroleum pipeline to be detected by using the verified crack detection model, and determining the crack area.
In some embodiments, the method further comprises:
the processing module is used for extracting pixel point information of the crack image included in the crack region, selecting a proper gray threshold value and carrying out binarization processing on the crack image;
The processing module is used for carrying out clustering segmentation on the crack images after binarization processing by utilizing the density clustering algorithm, and fitting the geometric form of the crack.
In some embodiments, the method further comprises:
the processing module is used for filtering abnormal pixel points in the crack area through Gaussian filtering;
the processing module is used for extracting a crack contour and a crack skeleton of the crack image included in the filtered crack region;
the processing module is used for splicing the same crack on different crack images included in the crack area by utilizing an image splicing algorithm based on the crack outline and the crack skeleton, and reconstructing the crack network by utilizing an angular point detection algorithm.
In some embodiments, the method further comprises:
the processing module is used for extracting a crack contour of the crack region according to the central pixel gray value and other surrounding pixel gray values of the crack region;
and the processing module is used for extracting the crack skeleton of the crack region by adopting a Zhang-Suen refinement algorithm.
In some embodiments, the method further comprises:
and the output module is used for constructing a three-dimensional space coordinate system by taking the tangent point of the outer wall of the petroleum pipeline and the plane as a reference coordinate point according to the position information corresponding to the crack image, modeling the whole petroleum pipeline in 3D, and simultaneously carrying out three-dimensional display on the distribution of the crack network on the petroleum pipeline in the space coordinate system.
Referring to fig. 8, the embodiment of the present invention further provides a terminal, where the terminal includes a processor 301 and a memory 302; the processor 301 is configured to implement the dual-attention mechanism-based petroleum pipeline crack detection method according to any one of the embodiments of the present invention when running a computer program, and the memory 302 stores instructions and data.
In some embodiments, the memory 302 in embodiments of the present invention may be either volatile memory or nonvolatile memory, or may include both volatile and nonvolatile memory. The nonvolatile Memory may be a Read-Only Memory (ROM), a Programmable ROM (PROM), an Erasable PROM (EPROM), an Electrically Erasable EPROM (EEPROM), or a flash Memory. The volatile memory may be random access memory (Random access Memory, RAM) which acts as an external cache. By way of example, and not limitation, many forms of RAM are available, such as Static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (Double Data Rate SDRAM), enhanced SDRAM (ESDRAM), synchronous DRAM (SLDRAM), and Direct RAM (DRRAM). The memory 302 of the systems and methods described herein is intended to comprise, without being limited to, these and any other suitable types of memory.
While processor 301 may be an integrated circuit chip with signal processing capabilities. In implementation, the steps of the above method may be performed by integrated logic circuitry of hardware in the processor 301 or instructions in the form of software. The processor 301 may be a general purpose processor, a digital signal processor (Digital Signal Processor, DSP), an application specific integrated circuit (application Specific Integrated Circuit, ASIC), an off-the-shelf programmable gate array (Field Programmable Gate array, FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware components. The disclosed methods, steps, and logic blocks in the embodiments of the present invention may be implemented or performed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of the method disclosed in connection with the embodiments of the present invention may be embodied directly in the execution of a hardware decoding processor, or in the execution of a combination of hardware and software modules in a decoding processor. The software modules may be located in a random access memory, flash memory, read only memory, programmable read only memory, or electrically erasable programmable memory, registers, etc. as well known in the art. The storage medium is located in a memory 302 and the processor 301 reads the information in the memory 302 and in combination with its hardware performs the steps of the above method.
In some embodiments, the embodiments described herein may be implemented in hardware, software, firmware, middleware, microcode, or a combination thereof. For a hardware implementation, the processing units may be implemented within one or more application specific integrated circuits (application Specific Integrated Circuits, ASIC), digital signal processors (Digital SignalProcessing, DSP), digital signal processing devices (DSP devices, DSPD), programmable logic devices (ProgrammableLogic Device, PLD), field programmable gate arrays (Field-Programmable Gate array, FPGA), general purpose processors, controllers, micro-controllers, microprocessors, other electronic units designed to perform the functions described herein, or a combination thereof.
For a software implementation, the techniques described herein may be implemented with modules (e.g., procedures, functions, and so on) that perform the functions described herein. The software codes may be stored in a memory and executed by a processor. The memory may be implemented within the processor or external to the processor.
Yet another embodiment of the present invention provides a computer storage medium storing an executable program which, when executed by the processor 301, can implement steps of an information processing method applied to the terminal. Such as one or more of the methods shown in fig. 1-6.
In some embodiments, the computer storage medium may include: a U-disk, a removable hard disk, a Read Only Memory (ROM), a random access Memory (RAM, random access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
It should be noted that: the technical schemes described in the embodiments of the present invention may be arbitrarily combined without any collision.
The foregoing is merely illustrative of the present invention, and the present invention is not limited thereto, and any person skilled in the art will readily recognize that variations or substitutions are within the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (10)
1. A petroleum pipeline crack detection method based on a dual-attention mechanism, the method comprising:
s101: acquiring an internal image of a petroleum pipeline to be detected, and recording position information corresponding to the internal image;
s102: constructing a crack detection model of the petroleum pipeline based on a dual-attention mechanism and a yolov5-seg network;
S103: performing crack detection on an internal image of the petroleum pipeline to be detected by using the crack detection model, and determining a crack area;
s104: clustering and segmenting crack images included in the crack region by using a density clustering algorithm, and fitting the geometric form of the crack;
s105: extracting crack contours and crack skeletons in the crack images, splicing the crack images by using an image splicing algorithm, and reconstructing a crack network by using an angular point detection algorithm;
s106: and 3D modeling is carried out on the pipeline according to the position information corresponding to the crack image in the crack network, and the crack network morphology is visually displayed.
2. The method according to claim 1, wherein the yolov5-seg network in step S102 includes a backbone network, a neck network, and a head network, the dual attention mechanism includes a channel attention mechanism and a spatial attention mechanism, and the step S102 includes:
adding a channel attention mechanism at the connection part of the backbone network and the neck network, and adding a space attention mechanism behind the backbone network to improve the yolov5-seg network;
and constructing the crack detection model by adopting the improved yolov5-seg network as a main network.
3. The method according to claim 1, wherein the step S103 further comprises:
acquiring crack images in the petroleum pipeline to form an original image set;
expanding the original image concentrated crack image through cutting, rotation and translation;
dividing the expanded original image set into the training set and the testing set according to a certain proportion;
marking the crack areas of the crack images in the training set by using image marking software;
training the crack detection model by using the marked training set;
validating the trained crack detection model using the validation set;
and carrying out crack detection on the internal image of the petroleum pipeline to be detected by using the verified crack detection model, and determining the crack area.
4. The method according to claim 1, wherein the step S104 includes:
extracting pixel point information of a crack image included in the crack region, and selecting a proper gray threshold to perform binarization processing on the crack image;
and clustering and segmenting the binarized crack image by using the density clustering algorithm, and fitting the geometric form of the crack.
5. The method according to claim 1, wherein the step S105 includes:
filtering abnormal pixel points in the crack area through Gaussian filtering;
extracting a crack contour and a crack skeleton of the crack image included in the filtered crack region;
based on the crack contour and the crack skeleton, splicing the same crack on different crack images included in the crack region by using an image splicing algorithm, and reconstructing the crack network by using a corner detection algorithm.
6. The method of claim 5, wherein the extracting the filtered crack region comprises a crack profile and a crack skeleton of a crack image, comprising:
extracting a crack contour of the crack region according to the central pixel gray value and other surrounding pixel gray values of the crack region;
extracting the crack skeleton of the crack region by using a Zhang-Suen refinement algorithm.
7. The method according to claim 1, wherein the step S106 includes:
and constructing a three-dimensional space coordinate system by taking the tangent point of the outer wall of the petroleum pipeline and the plane as a reference coordinate point according to the position information corresponding to the crack image, modeling the whole petroleum pipeline in 3D, and simultaneously carrying out three-dimensional display on the distribution of the crack network on the petroleum pipeline in the space coordinate system.
8. Petroleum pipeline crack intelligent detection device based on dual attention mechanism, characterized in that, the device includes:
the acquisition module is used for acquiring an internal image of the petroleum pipeline to be detected and recording position information corresponding to the internal image;
the processing module is used for constructing a crack detection model of the petroleum pipeline based on the dual attention mechanism and the yolov5-seg network;
the processing module is also used for carrying out crack detection on the internal image of the petroleum pipeline to be detected by utilizing the crack detection model, and determining a crack area;
the processing module is further used for carrying out clustering segmentation on the crack images included in the crack region by utilizing a density clustering algorithm, and fitting the geometric form of the crack;
the processing module is also used for extracting crack contours and crack frameworks in the crack images, splicing the crack images by using an image splicing algorithm and reconstructing a crack network by using a corner detection algorithm; and the output module is used for carrying out 3D modeling on the pipeline according to the position information corresponding to the crack image in the crack network, and visually displaying the crack network morphology.
9. A terminal comprising a processor and a memory for storing a computer program capable of running on the processor; wherein the processor is configured to implement the dual-attention mechanism-based petroleum pipeline crack detection method of any one of claims 1-7 when the computer program is run.
10. A computer readable storage medium having computer executable instructions embodied therein, wherein the computer executable instructions are executed by a processor to implement the dual-attention-mechanism-based petroleum pipeline crack detection method of any one of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310913981.8A CN116958089A (en) | 2023-07-24 | 2023-07-24 | Petroleum pipeline crack detection method based on dual-attention mechanism |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310913981.8A CN116958089A (en) | 2023-07-24 | 2023-07-24 | Petroleum pipeline crack detection method based on dual-attention mechanism |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116958089A true CN116958089A (en) | 2023-10-27 |
Family
ID=88457820
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310913981.8A Pending CN116958089A (en) | 2023-07-24 | 2023-07-24 | Petroleum pipeline crack detection method based on dual-attention mechanism |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116958089A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117557460A (en) * | 2024-01-12 | 2024-02-13 | 济南科汛智能科技有限公司 | Angiography image enhancement method |
-
2023
- 2023-07-24 CN CN202310913981.8A patent/CN116958089A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117557460A (en) * | 2024-01-12 | 2024-02-13 | 济南科汛智能科技有限公司 | Angiography image enhancement method |
| CN117557460B (en) * | 2024-01-12 | 2024-03-29 | 济南科汛智能科技有限公司 | Angiography image enhancement method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112017189B (en) | Image segmentation method and device, computer equipment and storage medium | |
| CN110428432B (en) | Deep neural network algorithm for automatically segmenting colon gland image | |
| CN111696094B (en) | Immunohistochemical PD-L1 membrane staining pathological section image processing method, device and equipment | |
| CN111461114A (en) | Multi-scale feature pyramid text detection method based on segmentation | |
| CN109146832B (en) | Video image splicing method and device, terminal equipment and storage medium | |
| CN114332473B (en) | Object detection method, device, computer apparatus, storage medium, and program product | |
| CN116539619B (en) | Product defect detection method, system, device and storage medium | |
| CN116958089A (en) | Petroleum pipeline crack detection method based on dual-attention mechanism | |
| CN112215217B (en) | Digital image recognition method and device for simulating doctor to read film | |
| CN113065551A (en) | Method for performing image segmentation using a deep neural network model | |
| CN114821350B (en) | Building change detection method for high-resolution remote sensing image through multi-stage information fusion | |
| CN115082909B (en) | Method and system for identifying lung lesions | |
| CN106469293A (en) | The method and system of quick detection target | |
| Adu-Gyamfi et al. | Functional evaluation of pavement condition using a complete vision system | |
| Yu | YOLO V5s-based deep learning approach for concrete cracks detection | |
| CN116977334B (en) | Optical cable surface flaw detection method and device | |
| CN110889418A (en) | Gas contour identification method | |
| CN116735723A (en) | Rail damage ultrasonic positioning and identifying system | |
| CN116958031A (en) | Defect detection method and related device | |
| CN115294489B (en) | Semantic segmentation method and system for disaster video data | |
| CN117649415B (en) | Cell balance analysis method based on optical flow diagram detection | |
| CN117274817B (en) | Automatic crack identification method and device, terminal equipment and storage medium | |
| CN114372994B (en) | Method for generating background image in video concentration | |
| CN111553925B (en) | FCN-based end-to-end crop image segmentation method and system | |
| CN117474917B (en) | Defect detection method and device, electronic equipment and storage medium |
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 |