CN110910350A - Nut loosening detection method for wind power tower cylinder - Google Patents
Nut loosening detection method for wind power tower cylinder Download PDFInfo
- Publication number
- CN110910350A CN110910350A CN201911046957.9A CN201911046957A CN110910350A CN 110910350 A CN110910350 A CN 110910350A CN 201911046957 A CN201911046957 A CN 201911046957A CN 110910350 A CN110910350 A CN 110910350A
- Authority
- CN
- China
- Prior art keywords
- nut
- image
- model
- dimensional image
- max
- 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.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/187—Segmentation; Edge detection involving region growing; involving region merging; involving connected component labelling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a nut loosening detection method for a wind power tower, which comprises the following steps: collecting two-dimensional image data; detecting the position of the nut in the two-dimensional image through a random forest classification model; acquiring a primary translation vector of each nut in the two-dimensional image relative to a camera coordinate system through a regression model; selecting a preset nut posture template which is closest to the preset nut posture template; optimizing an objective function to obtain a current rotation vector R 'and a current translation vector T' of each nut; the rotation vector R 'and the current translation vector T' are compared with the preset original rotation vector R0And the original translation vector T0And comparing to judge that the nut is loosened. Compared with the prior art, the method has the advantages that the computer vision and machine learning technology are used, the relatively accurate initial pose can be obtained by a small number of templates, the algorithm implementation cost is reduced, the precision is high, and the labor cost is effectively reduced.
Description
Technical Field
The invention relates to the field of fan inspection, in particular to a nut loosening detection method for a wind power tower.
Background
The wind power plant has the advantages that the distribution points of the lines and the fan areas are multiple and wide, the landform is complex, the natural environment is severe, the wind power plant is often damaged by the natural environment in various degrees, and the routing inspection of the fan and the power transmission line is indispensable work. The current fan inspection method mainly adopts manual inspection, and inspects the conditions of blades, a tower barrel, electric wires and the like by inspection personnel through a telescope under the fan. And what is most troublesome in the fan inspection work is the loosening identification of the nuts at the specific positions of the fan units. The traditional manual inspection mode is time-consuming and labor-consuming, and the efficiency is not high. The conventional nut loosening recognition method is to draw a straight line mark from the top to the bottom of the bolt and the nut as a whole in an initial state of the bolt being installed. The state of the line is observed after a period of time: if the mark is still a straight line when viewed from the front, the bolt is not loosened; if the condition that the upper part and the lower part are separated from each other in the middle of the mark and do not form a whole straight line is observed, the nut is loosened. The observation mark is generally judged by manual direct on-site judgment and remote shooting through a camera device, so that the time and the energy of inspection personnel are consumed, and misjudgment can occur.
Disclosure of Invention
The invention aims to overcome the defects in the prior art and provide a nut loosening detection method for a wind power tower.
The purpose of the invention can be realized by the following technical scheme:
a nut loosening detection method for a wind power tower cylinder specifically comprises the following steps:
s1, collecting two-dimensional image data;
s2, detecting the position of the nut in the two-dimensional image through a random forest classification model;
s3, acquiring a primary translation vector of each nut in the two-dimensional image relative to a camera coordinate system through a regression model;
s4, selecting a preset nut posture template closest to each nut according to the initial translation vector of each nut, and taking the nut posture template as the initial position of each nut, thereby obtaining the initial rotation vector R and the initial translation vector T of each nut relative to a camera coordinate system in the two-dimensional image;
s5, bringing the initial rotation vector R and the initial translation vector T into an objective function, and optimizing the objective function to obtain the current rotation vector R 'and the current translation vector T' of each nut;
s6, setting the rotation vector R 'and the current translation vector T' of each nut and the preset original rotation vector R thereof0And the original translation vector T0And comparing, and if the difference is larger than a set threshold, judging that the nut is loosened.
Further, in step S5, the objective function expression is:
wherein g (T, R) represents the three-dimensional pose of the nut, oiRepresenting the extracted grating points of the nut CAD model in three-dimensional space, pi (·) representing the camera projection model,representing the edge direction of the projection model after projection of the grating points onto a two-dimensional plane, DT3VIndicates the direction chamfer distance, EDCMRepresenting the residual sum function.
Further, in step S2, the training step of the random forest classification model includes:
a1, acquiring a large amount of two-dimensional image data containing nuts, wherein the two-dimensional image data contains various postures of the nuts;
a2, acquiring the pose of the nut in each frame of image by using a tracking algorithm, projecting edge grating points onto a two-dimensional image after rasterizing a CAD (computer-aided design) model of the nut, and continuously optimizing a target function so as to adjust the projection position of the grating points; when the optimal pose is reachedThen, all raster points are traversed to find xmin,xmax,yminAnd ymaxFour coordinate values constituting two index points (x)min,ymin) And (x)max,ymax) Constructing a rectangular area according to the two mark points, wherein the image characteristics in the rectangular area are the image characteristics of the nut;
a3, dividing the direction chamfer matching tensor of the two-dimensional image by using the rectangular region in the step A2, and extracting an image region only containing the edge of the nut as a positive sample of training;
a4, randomly intercepting other areas of the current two-dimensional image, and obtaining an image area with the same size and direction chamfer matching tensor as a detection database as a negative sample;
and A5, inputting the positive sample and the negative sample into a random forest machine learning model for training.
Further, the tracking algorithm is extracted by adopting an LSD edge extraction algorithm or a Canny edge extraction algorithm.
Further, the features of the same nut under different postures are classified, and the features under different postures are used as independent detection classes to train a random forest classification model.
6. The method for detecting loosening of nuts for a wind power tower as claimed in claim 1, wherein in step S3, the training step of the regression model includes:
b1, acquiring the pose of the nut in each frame of image by using a tracking algorithm, projecting edge grating points onto a two-dimensional image after rasterizing the CAD model of the nut, and continuously optimizing a target function so as to adjust the projection position of the grating points; when the optimal pose is reached, all the raster points are traversed to find xmin,xmax,yminAnd ymaxFour coordinate values constituting two index points (x)min,ymin) And (x)max,ymax) Constructing a rectangular area according to the two mark points, segmenting the direction chamfer matching tensor of each frame of image according to the position and the size of the rectangular area, and extracting the image area only containing the edge of the nut as a special image for trainingPerforming sign;
b2, acquiring position characteristics according to the rectangular area: coordinate value x of upper left corner of rectangleminAnd ymin,xmax-xminLength of (a), and ymax-yminThe width of (d);
b3, acquiring a translation vector of each bolt according to the optimal position in the step B1 as a true pose value;
and B4, inputting the image characteristics, the position characteristics and the pose truth values into a regression model for training.
Further, the establishing step of the objective function is as follows:
c1, acquiring a scene image containing the nut;
c2, converting the scene image into a gray image and extracting a scene edge image by using an extraction algorithm;
c3, calculating a direction chamfer matching tensor corresponding to the scene image;
c4, extracting a grating point model of the nut from the CAD model of the nut, and projecting the grating point model to an image plane through a preset nut posture template and a vision sensor projection model;
and C5, constructing an optimized objective function according to the model edge direction of the projection point and the direction chamfer matching tensor.
Further, the two-dimensional image data is acquired by an industrial camera or a monitoring camera.
Compared with the prior art, the invention has the following advantages:
the invention provides a method for inspecting and judging whether the nut in the fan tower barrel is loosened or not for inspection personnel by using computer vision and machine learning technology. According to the method, the initial positioning of the pose of the nut is realized by utilizing random forest classification and regression models, the pose of the nut is matched with the closest nut pose template, and the accurate pose of the nut is obtained through the objective function and is used for judging whether the nut is loosened or not.
Drawings
FIG. 1 is a schematic structural diagram of the present invention.
Detailed Description
The invention is described in detail below with reference to the figures and specific embodiments. The present embodiment is implemented on the premise of the technical solution of the present invention, and a detailed implementation manner and a specific operation process are given, but the scope of the present invention is not limited to the following embodiments.
As shown in fig. 1, the embodiment provides a nut loosening detection method for a wind power tower, which specifically includes the following steps:
step S1, acquiring two-dimensional image data containing the target object through a visual sensor, such as an industrial camera or a monitoring camera;
s2, detecting the position of the nut in the two-dimensional image through a random forest classification model;
s3, acquiring a primary translation vector of each nut in the two-dimensional image relative to a camera coordinate system through a regression model;
s4, selecting a preset nut posture template which is closest to each nut according to the initial translation vector of each nut, and taking the nut posture template as the initial position of each nut, thereby obtaining the initial rotation vector R and the initial translation vector T of each nut relative to a camera coordinate system in the two-dimensional image; the nut posture template is obtained by rotating the z-axis at equal angular intervals, and the 180-degree angle range of the front face is traversed to ensure that the pose of the front face can be successfully matched.
Step S5, bringing the initial rotation vector R and the initial translation vector T into an objective function, and optimizing the objective function to obtain the current rotation vector R 'and the current translation vector T' of each nut;
step S6, the rotation vector R 'and the current translation vector T' of each nut are compared with the preset original rotation vector R0And the original translation vector T0And comparing, and judging that the nut is loosened if the difference value is larger than a set threshold value (at least one of the difference values of the components (tx, ty, tz) and (rx, ry, rz) of the translation vector and the rotation vector is larger than 0.01).
First, the establishment of the detection objective function
The first step is as follows: an image of a scene containing an object is acquired.
The second step is that: and (4) converting the scene image into a gray image, and then extracting a scene edge image by using an LSD edge extraction algorithm or a Canny edge extraction algorithm.
The second step is that: calculating a Direction Chamfer Matching (DCM) tensor corresponding to the scene image, wherein the tensor is used for representing the corresponding relation between each point in the image coordinate and the edge, and in the DCM tensor image, the closer the pixel point to the image edge is, the smaller the gray value is, and the darker the image is displayed; the farther the pixel points are from the edge of the image, the larger the gray value of the pixel points is, and the brighter the image is displayed.
The third step: and extracting a target object grating point model from the CAD model of the target object, and projecting the grating point model to an image plane through a preset target object initial pose and a visual sensor projection model.
The fourth step: and constructing an optimized objective function according to the model edge direction of the projection point and the direction chamfer matching tensor. Let the pose of the object relative to the camera contain a rotation vector R ═ Rxryrz]TAnd translation vector T ═ Txtytz]TAnd obtaining an objective function:
wherein g (T, R) represents the three-dimensional pose of the nut, oiRepresenting the extracted grating points of the nut CAD model in three-dimensional space, pi (·) representing the camera projection model,representing the edge direction of the projection model after projection of the grating points onto a two-dimensional plane, DT3VDirection chamfer distance, EDCMA residual sum function.
And obtaining the accurate pose transformation relation of the object in the current image relative to the sensor through analytic partial derivative optimization E (T, R): analysis of the target function E (T)R), which is a two-dimensional image coordinate point set x mapped into the image by the scene DT3v tensor, modeli(ii,vi) And the edge direction of the point setAnd determining the variable to be optimized as pose transformation g (T, R) of the target object relative to the camera. The object edge projection point coordinates directly determine an objective function for pose optimization, so that the accuracy of pose matching is improved by analyzing and solving the partial derivatives of pose changes relative to the projection point coordinates. By giving a certain perturbation of the relative pose, firstly calculating the relative change of the coordinates of the projection point, then calculating the pose optimization residual error through the change of the coordinates, searching the minimum value in a certain step length in the corresponding DCM tensor as the optimization residual error of the grating point, and optimizing the objective function by using a nonlinear least square method to obtain the optimization pose.
Second, training and detecting classification model
A. A large number of gray level videos containing target objects are acquired through the sensor, and various poses of the objects should appear in the videos as much as possible.
B. Acquiring the pose of a target object in each frame of image by using a tracking algorithm, projecting edge grating points onto a two-dimensional image after rasterizing a CAD (computer-aided design) model of the nut, and continuously optimizing a target function so as to adjust the projection positions of the grating points; when the optimal pose is reached, all the raster points are traversed to find xmin,xmax,yminAnd ymaxFour coordinate values constituting two index points (x)min,ymin) And (x)max,ymax) And constructing a rectangular area according to the two mark points, wherein the image characteristics in the rectangular area are the image characteristics of the nut.
C. The direction chamfer matching tensor of the two-dimensional image is segmented by using the rectangular region in the step a2, and an image region including only the nut edge is extracted as a positive sample for training.
D. And randomly intercepting other areas of the current two-dimensional image, and obtaining an image area with the same size of the direction chamfer matching tensor as a detection database serving as a negative sample.
E. And classifying the characteristics of the same object under different postures, and taking the characteristics under different postures as independent detection classes to train the model.
F. The collected image data is input for training by using a random forest machine learning model carried by sklern. For example, according to the symmetry characteristics of the nut, the poses of the nut in different scenes are roughly classified into two types, namely 1 and 2, and the negative sample is marked as 0.
In the detection stage, the input picture is detected in a sliding window mode, namely, the picture in a certain area is selected through the sliding window to be cut, and the cut picture is classified and judged by using a trained classification model.
Thirdly, training a model for estimating a translation vector T
A. Acquiring the pose of the nut in each frame of image by using a tracking algorithm, projecting edge grating points onto a two-dimensional image after rasterizing a CAD (computer-aided design) model of the nut, and continuously optimizing a target function so as to adjust the projection position of the grating points; when the optimal pose is reached, all the raster points are traversed to find xmin,xmax,yminAnd ymaxFour coordinate values constituting two index points (x)min,ymin) And (x)max,ymax) And constructing a rectangular area according to the two mark points, segmenting the direction chamfer matching tensor of each frame of image according to the position and the size of the rectangular area, and extracting the image area only containing the edge of the nut as the image characteristic of training.
B. Obtaining position features according to the rectangular area: coordinate value x of upper left corner of rectangleminAnd ymin,xmax-xminLength of (a), and ymax-yminIs measured.
C. And acquiring the translation vector of each bolt as a true pose value according to the optimal position in the step B1.
D. And inputting the image characteristics, the position characteristics and the pose truth values into a regression model for training.
A detection stage: the method comprises the steps of detecting an input picture by using a sliding window mode, namely, selecting a picture in a certain area through a sliding window to cut, finding a detection frame containing a target object after the sliding window detection of a scene picture is completed, and returning a primary translation vector T of the object relative to a camera coordinate system by taking a left vertex coordinate of the frame, the size of the frame and a detection classification result as input.
Fourthly, detecting the looseness of the nut
During actual detection, a nut image is shot, a sliding window is cut out on the image, a plurality of pictures (pictures 1, 2, 3 …, N) with the size of the sliding window are obtained, the pictures are input into a classification model, a picture N containing a target nut is found through the classification model, the picture N, the coordinate and the size of the sliding window are combined and input into a regression model, the translation vector of the nut in the picture N is obtained, a corresponding template is found according to the translation vector, the template is used as an initial pose and input into an optimization function, and the accurate pose (T, R) is obtained through optimization. The same nut in two pictures in different periods is detected according to the detection method, and if the difference of the pose values is larger than a threshold value, the loosening is considered to be generated.
The foregoing detailed description of the preferred embodiments of the invention has been presented. It should be understood that numerous modifications and variations could be devised by those skilled in the art in light of the present teachings without departing from the inventive concepts. Therefore, the technical solutions available to those skilled in the art through logic analysis, reasoning and limited experiments based on the prior art according to the concept of the present invention should be within the scope of protection defined by the claims.
Claims (8)
1. A nut loosening detection method for a wind power tower barrel is characterized by comprising the following steps:
s1, collecting two-dimensional image data;
s2, detecting the position of the nut in the two-dimensional image through a random forest classification model;
s3, acquiring a primary translation vector of each nut in the two-dimensional image relative to a camera coordinate system through a regression model;
s4, selecting a preset nut posture template closest to each nut according to the initial translation vector of each nut, and taking the nut posture template as the initial position of each nut, thereby obtaining the initial rotation vector R and the initial translation vector T of each nut relative to a camera coordinate system in the two-dimensional image;
s5, bringing the initial rotation vector R and the initial translation vector T into an objective function, and optimizing the objective function to obtain the current rotation vector R 'and the current translation vector T' of each nut;
s6, setting the rotation vector R 'and the current translation vector T' of each nut and the preset original rotation vector R thereof0And the original translation vector T0And comparing, and if the difference is larger than a set threshold, judging that the nut is loosened.
2. The method for detecting loosening of nuts used for wind power tower cylinder according to claim 1, wherein in step S5, the objective function expression is:
wherein g (T, R) represents the three-dimensional pose of the nut, oiRepresenting the extracted grating points of the nut CAD model in three-dimensional space, pi (·) representing the camera projection model,representing the edge direction of the projection model after projection of the grating points onto a two-dimensional plane, DT3VIndicates the direction chamfer distance, EDCMRepresenting the residual sum function.
3. The method for detecting loosening of nuts used for a wind power tower as claimed in claim 1, wherein in step S2, the training step of the random forest classification model includes:
a1, acquiring a large amount of two-dimensional image data containing nuts, wherein the two-dimensional image data contains various postures of the nuts;
a2, acquiring the pose of the nut in each frame of image by using a tracking algorithm, rasterizing the CAD model of the nut,projecting the edge grating points on a two-dimensional image, and continuously optimizing a target function so as to adjust the projection positions of the grating points; when the optimal pose is reached, all the raster points are traversed to find xmin,xmax,yminAnd ymaxFour coordinate values constituting two index points (x)min,ymin) And (x)max,ymax) Constructing a rectangular area according to the two mark points, wherein the image characteristics in the rectangular area are the image characteristics of the nut;
a3, dividing the direction chamfer matching tensor of the two-dimensional image by using the rectangular region in the step A2, and extracting an image region only containing the edge of the nut as a positive sample of training;
a4, randomly intercepting other areas of the current two-dimensional image, and obtaining an image area with the same size and direction chamfer matching tensor as a detection database as a negative sample;
and A5, inputting the positive sample and the negative sample into a random forest machine learning model for training.
4. The method as claimed in claim 2, wherein the tracking algorithm is extracted by using an LSD edge extraction algorithm or a Canny edge extraction algorithm.
5. The nut loosening detection method for the wind power tower according to claim 2, characterized in that the features of the same nut in different postures are classified, and the features in different postures are used as separate detection classes to train a random forest classification model.
6. The method for detecting loosening of nuts for a wind power tower as claimed in claim 1, wherein in step S3, the training step of the regression model includes:
b1, acquiring the pose of the nut in each frame of image by using a tracking algorithm, projecting edge grating points onto a two-dimensional image after rasterizing the CAD model of the nut, and continuously optimizing a target function so as to adjust the projection position of the grating points; when the optimum is reachedAfter the pose, all raster points are traversed to find xmin,xmax,yminAnd ymaxFour coordinate values constituting two index points (x)min,ymin) And (x)max,ymax) Constructing a rectangular area according to the two mark points, segmenting the direction chamfer matching tensor of each frame of image according to the position and the size of the rectangular area, and extracting the image area only containing the edge of the nut as the image characteristic of training;
b2, acquiring position characteristics according to the rectangular area: coordinate value x of upper left corner of rectangleminAnd ymin,xmax-xminLength of (a), and ymax-yminThe width of (d);
b3, acquiring a translation vector of each bolt according to the optimal position in the step B1 as a true pose value;
and B4, inputting the image characteristics, the position characteristics and the pose truth values into a regression model for training.
7. The method for detecting loosening of nuts used for a wind power tower according to claim 1, wherein the objective function establishing step is as follows:
c1, acquiring a scene image containing the nut;
c2, converting the scene image into a gray image and extracting a scene edge image by using an extraction algorithm;
c3, calculating a direction chamfer matching tensor corresponding to the scene image;
c4, extracting a grating point model of the nut from the CAD model of the nut, and projecting the grating point model to an image plane through a preset nut posture template and a vision sensor projection model;
and C5, constructing an optimized objective function according to the model edge direction of the projection point and the direction chamfer matching tensor.
8. The method for detecting loosening of nuts for a wind power tower according to claim 1, wherein the two-dimensional image data is collected by an industrial camera or a surveillance camera.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911046957.9A CN110910350B (en) | 2019-10-30 | 2019-10-30 | Nut loosening detection method for wind power tower cylinder |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911046957.9A CN110910350B (en) | 2019-10-30 | 2019-10-30 | Nut loosening detection method for wind power tower cylinder |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110910350A true CN110910350A (en) | 2020-03-24 |
CN110910350B CN110910350B (en) | 2022-06-14 |
Family
ID=69814856
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911046957.9A Active CN110910350B (en) | 2019-10-30 | 2019-10-30 | Nut loosening detection method for wind power tower cylinder |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110910350B (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111442723A (en) * | 2020-03-27 | 2020-07-24 | 江苏理工学院 | Method and device for integrally detecting shape and position of nut on mobile phone middle plate |
CN111830045A (en) * | 2020-05-29 | 2020-10-27 | 华电电力科学研究院有限公司 | Unmanned aerial vehicle wind power detection system and method based on BIM technology |
CN111898575A (en) * | 2020-08-06 | 2020-11-06 | 华北电力大学(保定) | Bolt-nut pair automatic combination method based on fast R-CNN detector |
CN111915632A (en) * | 2020-07-26 | 2020-11-10 | 同济大学 | Poor texture target object truth value database construction method based on machine learning |
CN112986146A (en) * | 2021-05-06 | 2021-06-18 | 上海建工集团股份有限公司 | Image recognition-based guide rail frame bolt connection reliability detection device |
CN113554003A (en) * | 2021-09-18 | 2021-10-26 | 三一汽车制造有限公司 | Mixing drum rotation direction identification method and device and electronic equipment |
CN114061738A (en) * | 2022-01-17 | 2022-02-18 | 风脉能源(武汉)股份有限公司 | Wind turbine tower drum foundation ring vibration monitoring method based on calibration plate pose calculation |
CN116681660A (en) * | 2023-05-18 | 2023-09-01 | 中国长江三峡集团有限公司 | Target object defect detection method and device, electronic equipment and storage medium |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102840971A (en) * | 2012-08-29 | 2012-12-26 | 江苏法尔胜泓昇集团有限公司 | Looseness detecting device of bolt and nut fastening piece |
CN105719285A (en) * | 2016-01-19 | 2016-06-29 | 东南大学 | Pedestrian detection method based on directional chamfering distance characteristics |
CN107650514A (en) * | 2017-09-27 | 2018-02-02 | 何大为 | A kind of track loosens detection automatic marking device |
CN107993287A (en) * | 2017-12-01 | 2018-05-04 | 大唐国信滨海海上风力发电有限公司 | A kind of auto-initiation method of target following |
CN108122256A (en) * | 2017-12-25 | 2018-06-05 | 北京航空航天大学 | It is a kind of to approach under state the method for rotating object pose measurement |
-
2019
- 2019-10-30 CN CN201911046957.9A patent/CN110910350B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102840971A (en) * | 2012-08-29 | 2012-12-26 | 江苏法尔胜泓昇集团有限公司 | Looseness detecting device of bolt and nut fastening piece |
CN105719285A (en) * | 2016-01-19 | 2016-06-29 | 东南大学 | Pedestrian detection method based on directional chamfering distance characteristics |
CN107650514A (en) * | 2017-09-27 | 2018-02-02 | 何大为 | A kind of track loosens detection automatic marking device |
CN107993287A (en) * | 2017-12-01 | 2018-05-04 | 大唐国信滨海海上风力发电有限公司 | A kind of auto-initiation method of target following |
CN108122256A (en) * | 2017-12-25 | 2018-06-05 | 北京航空航天大学 | It is a kind of to approach under state the method for rotating object pose measurement |
Non-Patent Citations (2)
Title |
---|
PRIYANKA MANDIKAL ET AL.: "3D-LMNet:Latent Embedding Matching for Accurate and Diverse 3D Point Cloud Reconstruction from a Single Image", 《ARXIV》 * |
王宝丽 等: "中低速磁浮F轨紧固螺栓松动故障检测", 《信息技术》 * |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111442723A (en) * | 2020-03-27 | 2020-07-24 | 江苏理工学院 | Method and device for integrally detecting shape and position of nut on mobile phone middle plate |
CN111442723B (en) * | 2020-03-27 | 2022-02-11 | 江苏理工学院 | Method and device for integrally detecting shape and position of nut on mobile phone middle plate |
CN111830045A (en) * | 2020-05-29 | 2020-10-27 | 华电电力科学研究院有限公司 | Unmanned aerial vehicle wind power detection system and method based on BIM technology |
CN111915632A (en) * | 2020-07-26 | 2020-11-10 | 同济大学 | Poor texture target object truth value database construction method based on machine learning |
CN111915632B (en) * | 2020-07-26 | 2024-04-26 | 同济大学 | Machine learning-based method for constructing truth database of lean texture target object |
CN111898575A (en) * | 2020-08-06 | 2020-11-06 | 华北电力大学(保定) | Bolt-nut pair automatic combination method based on fast R-CNN detector |
CN111898575B (en) * | 2020-08-06 | 2022-09-02 | 华北电力大学(保定) | Bolt-nut pair automatic combination method based on Faster R-CNN detector |
CN112986146A (en) * | 2021-05-06 | 2021-06-18 | 上海建工集团股份有限公司 | Image recognition-based guide rail frame bolt connection reliability detection device |
CN113554003A (en) * | 2021-09-18 | 2021-10-26 | 三一汽车制造有限公司 | Mixing drum rotation direction identification method and device and electronic equipment |
CN114061738A (en) * | 2022-01-17 | 2022-02-18 | 风脉能源(武汉)股份有限公司 | Wind turbine tower drum foundation ring vibration monitoring method based on calibration plate pose calculation |
CN114061738B (en) * | 2022-01-17 | 2022-05-17 | 风脉能源(武汉)股份有限公司 | Wind turbine tower drum foundation ring vibration monitoring method based on calibration plate pose calculation |
CN116681660A (en) * | 2023-05-18 | 2023-09-01 | 中国长江三峡集团有限公司 | Target object defect detection method and device, electronic equipment and storage medium |
CN116681660B (en) * | 2023-05-18 | 2024-04-19 | 中国长江三峡集团有限公司 | Target object defect detection method and device, electronic equipment and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN110910350B (en) | 2022-06-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110910350B (en) | Nut loosening detection method for wind power tower cylinder | |
CN111563442B (en) | Slam method and system for fusing point cloud and camera image data based on laser radar | |
CN111340797B (en) | Laser radar and binocular camera data fusion detection method and system | |
CN113744270B (en) | Unmanned aerial vehicle visual detection and identification method for crane complex steel structure surface defects | |
CN109615611B (en) | Inspection image-based insulator self-explosion defect detection method | |
CN108038846A (en) | Transmission line equipment image defect detection method and system based on multilayer convolutional neural networks | |
Wang et al. | Measurement for cracks at the bottom of bridges based on tethered creeping unmanned aerial vehicle | |
CN111062915A (en) | Real-time steel pipe defect detection method based on improved YOLOv3 model | |
CN105373135A (en) | Method and system for guiding airplane docking and identifying airplane type based on machine vision | |
CN103727930A (en) | Edge-matching-based relative pose calibration method of laser range finder and camera | |
CN113538503B (en) | Solar panel defect detection method based on infrared image | |
CN113379712A (en) | Steel bridge bolt disease detection method and system based on computer vision | |
CN105447431B (en) | A kind of docking aircraft method for tracking and positioning and system based on machine vision | |
Troncoso-Pastoriza et al. | Use of BIM data as input and output for improved detection of lighting elements in buildings | |
CN115222884A (en) | Space object analysis and modeling optimization method based on artificial intelligence | |
CN107767366B (en) | A kind of transmission line of electricity approximating method and device | |
CN116958837A (en) | Municipal facilities fault detection system based on unmanned aerial vehicle | |
CN116309882A (en) | Tray detection and positioning method and system for unmanned forklift application | |
CN113688819B (en) | Target object expected point tracking and matching method based on marked points | |
CN112116561B (en) | Power grid transmission line detection method and device based on image processing fusion network weight | |
You et al. | A mechanical part sorting system based on computer vision | |
Niblock et al. | Fast model-based feature matching technique applied to airport lighting | |
Majidi et al. | Aerial tracking of elongated objects in rural environments | |
CN117823741B (en) | Pipe network non-excavation repairing method and system combined with intelligent robot | |
Kok et al. | Occluded apples orientation estimator based on deep learning model for robotic harvesting |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |