CN112198170A - Detection method for identifying water drops in three-dimensional detection of outer surface of seamless steel pipe - Google Patents

Detection method for identifying water drops in three-dimensional detection of outer surface of seamless steel pipe Download PDF

Info

Publication number
CN112198170A
CN112198170A CN202011048548.5A CN202011048548A CN112198170A CN 112198170 A CN112198170 A CN 112198170A CN 202011048548 A CN202011048548 A CN 202011048548A CN 112198170 A CN112198170 A CN 112198170A
Authority
CN
China
Prior art keywords
layer
detection
steel pipe
image
region
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
Application number
CN202011048548.5A
Other languages
Chinese (zh)
Other versions
CN112198170B (en
Inventor
窦曼莉
刘小楠
聂建华
聂斌
鲍磊
杨利红
时亚丽
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
ACADEMY OF PUBLIC SECURITY TECHNOLOGY HEFEI
Original Assignee
ACADEMY OF PUBLIC SECURITY TECHNOLOGY HEFEI
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by ACADEMY OF PUBLIC SECURITY TECHNOLOGY HEFEI filed Critical ACADEMY OF PUBLIC SECURITY TECHNOLOGY HEFEI
Priority to CN202011048548.5A priority Critical patent/CN112198170B/en
Publication of CN112198170A publication Critical patent/CN112198170A/en
Application granted granted Critical
Publication of CN112198170B publication Critical patent/CN112198170B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/84—Systems specially adapted for particular applications
    • G01N21/88—Investigating the presence of flaws or contamination
    • G01N21/95—Investigating the presence of flaws or contamination characterised by the material or shape of the object to be examined
    • G01N21/952—Inspecting the exterior surface of cylindrical bodies or wires
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
    • G01B11/00—Measuring arrangements characterised by the use of optical techniques
    • G01B11/24—Measuring arrangements characterised by the use of optical techniques for measuring contours or curvatures
    • 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
    • G06N3/08—Learning methods
    • 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
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/12—Edge-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
    • G06T7/181—Segmentation; Edge detection involving edge growing; involving edge linking
    • 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
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10028—Range image; Depth image; 3D point clouds
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—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

Abstract

The invention relates to the technical field of three-dimensional detection of the outer surface of a steel pipe, in particular to a detection method for identifying water drops in the three-dimensional detection of the outer surface of a seamless steel pipe. In the invention, three steps of data preprocessing, model training and online detection are used for detecting the seamless steel tube, in the actual production, a three-dimensional laser sensor splices the contour line of the tube body and converts the contour line into a depth map of the surface of the steel tube to be detected, a water drop detection model is used for detecting the depth map, the depth map is firstly subjected to morphological operation and searching outside the region to obtain a region to be detected which is suspected to be a water drop, and then a convolutional neural network is used for identifying the region to be detected.

Description

Detection method for identifying water drops in three-dimensional detection of outer surface of seamless steel pipe
Technical Field
The invention relates to the technical field of three-dimensional detection of the outer surface of a steel pipe, in particular to a detection method for identifying water drops in nondestructive detection of the outer surface of a seamless steel pipe.
Background
The hot-rolled seamless steel pipe needs to be sprayed and cooled in the production process, and particularly needs to be sprayed as a coupling agent when passing through ultrasonic detection equipment, so that water drops sometimes exist on the surface of a pipe body. When the three-dimensional sensor is used for collecting pipe body contour data, water drops are concave, convex, concave and convex in appearance on a depth map, the shape mainly takes the center as the oval and has transverse stripes on two sides, the shapes of the water drops on the surfaces of different steel pipes with iron sheets, paint spraying and the like are variable, the water drops are easily confused with real defects, serious interference is caused to detection, and frequent false alarm of detection equipment is caused. At present, a detection method for identifying water drops with high accuracy does not exist, so that the detection method for identifying the water drops in the three-dimensional detection of the outer surface of the seamless steel pipe is provided for solving the problems.
Disclosure of Invention
The invention aims to provide a detection method for identifying water drops in three-dimensional detection of the outer surface of a seamless steel pipe, which comprises the steps of data preprocessing, model training and online detection, wherein in actual production, a three-dimensional laser sensor splices contour lines of a pipe body into an arc, the surface of the steel pipe is fitted by using a least square method, contour data is projected into a two-dimensional gray image, a depth image is subjected to morphological operation and searching outside the area to obtain a to-be-detected area suspected as water drops, and then a convolutional neural network is used for identifying the to-be-detected area. In order to achieve the purpose, the invention provides the following technical scheme:
a detection method for identifying water drops in three-dimensional detection of the outer surface of a seamless steel pipe comprises the following steps:
step 1: data preprocessing: generating a two-dimensional gray scale image of the surface of the pipe body according to pipe body profile data obtained by the three-dimensional sensor, and preprocessing the gray scale image to obtain all areas to be detected;
step 2: model training: constructing a deep learning network suitable for detecting water drops on the surface of the steel pipe in an off-line manner, and training a model in an off-line manner by using the image of the area to be detected collected in the step (1);
and step 3: online detection: and (3) loading the model obtained in the step (2) into detection equipment, and when the detection equipment works, obtaining all areas to be detected according to the step (1), and detecting water drops on the surface of the pipe body on line.
Preferably, the data preprocessing part is specifically:
1. when the steel pipe passes through the detection device, the three-dimensional sensor acquires the profile data of the surface of the steel pipe to obtain three-dimensional point cloud data of the profile, a least square method is utilized to fit a reference plane of the surface of the steel pipe, points on the surface of the steel pipe, points protruding from the surface of the pipe body and points sinking are respectively projected into different numerical values, and the three-dimensional point cloud data are converted into two-dimensional gray images;
2. when the three-dimensional sensor collects the contour data of the surface of the steel pipe, the image uploaded to the industrial personal computer each time is only a part of the steel pipe, so that water drops on the image can be cut and distributed on two images, and in order to ensure the detection accuracy, the rear half part of the previous frame of image is spliced on the top of the current image;
3. performing corrosion-expansion-after-opening operation on the gray-scale image by using a 3 x 3 square structural element, removing isolated noise points, and identifying a non-zero area on the image after the noise points are removed as a suspected area region;
4. and extracting an edge outline edge of the suspected region, and searching the region outside the edge for points on the edge. Taking a point P (x, y) on the edge contour as an origin, filtering the gray level image by using a 3 x 3 matrix, filtering out points in the region, searching the region out of the region according to the formula (1), changing P1(x, y) into a new origin when a pixel point P1(x, y) with a difference of 0.5 with the gray level of P (x, y) is searched, continuing searching the region out of the region for a pixel point with a difference of 0.5 with the gray level of P1(x, y) until the point meeting the condition is not searched, and adding all searched points meeting the condition into the original suspected region;
Figure BDA0002708800370000021
edge＝{...，P(x，y)，P1(x，y)|abs(Pi(x, y) -P (x, y)) > 0.5} formula (2)
5. And combining the adjacent suspected areas to be the areas to be detected. The merging method is to calculate the intersection ratio of the adjacent suspected areas, and the suspected area with the intersection ratio larger than 0.85 is the area to be detected. The intersection ratio IoU of the suspected regions regionA and regionB is calculated as: the intersection of regionA and regionB is a proportion of the union. If the merge IoU of the region A and the region B is more than 0.85, the region A and the region B are merged into a region1 to be detected, and the region1 is the union of the region A and the region B.
Preferably, the model training part is specifically:
1. constructing a deep learning network suitable for detecting water drops on the surface of the steel pipe, wherein the deep learning network comprises a convolutional neural network for extracting characteristics and a regional suggestion network for accurately generating a target;
2. taking the image of the area to be detected as a training image, and training a water drop detection model off line: and collecting more than 100 pictures with water drops and more than 100 pictures without water drops as samples, and training by using a deep learning network until the loss function is less than 0.001 to obtain a water drop detection model.
Preferably, the online detection part specifically comprises:
1. when the steel pipe passes through the detection device, the three-dimensional sensor acquires the profile data of the surface of the steel pipe to obtain the three-dimensional point cloud data of the profile. Fitting the surface of the steel pipe by using a least square method, and projecting the profile data into a two-dimensional gray image;
2. splicing the back half part of the previous frame of image on the top of the current image;
3. performing morphological operation on the image, and removing noise points to obtain a suspected area;
4. calculating to search the outer area of the points at the edge of the suspected area, and expanding the suspected area;
5. combining two adjacent suspected areas into a to-be-detected area;
6. and taking all the areas to be detected on the picture as the input of the water drop detection model once, judging whether the areas to be detected are water or not, and displaying the identification result on a software interface.
The convolutional neural network is used for extracting a convolutional neural network of features, normalizing input image data, adjusting the size of an image to 224 multiplied by 224, and extracting the features of each image by using 8 layers of convolutional neural networks, and specifically comprises the following steps: the first layer is a convolution layer, the convolution kernel is 11 multiplied by 11, the step length is 2, the second layer is a maximum pooling layer, the size is 3 multiplied by 3, the step length is 2, and the input of the second layer is the output of the first layer; the third layer is a convolutional layer, the convolutional kernel is 5 multiplied by 5, the step length is 2, the input of the third layer is the weighted sum of the output of the first layer and the output of the second layer, the fourth layer is a maximum pooling layer, the size is 3 multiplied by 3, the step length is 2, and the input of the fourth layer is the output of the third layer; the fifth layer is a convolutional layer, the size of the convolutional core is 3 multiplied by 3, the step length is 1, and the input of the fifth layer is the weighted sum of the output of the first layer, the output of the third layer and the output of the fourth layer; the sixth layer is a convolutional layer, the size and the step length of the convolutional kernel of the fifth layer are the same, and the input of the sixth layer is the weighted sum of the output of the first layer, the output of the third layer, the output of the fourth layer and the output of the fifth layer; the seventh layer is a convolutional layer, the size and the step length of the convolutional kernel are the same as those of the fifth layer, the input of the seventh layer is the weighted sum of the output of the first layer, the output of the third layer, the output of the fourth layer, the output of the fifth layer and the output of the sixth layer, the eighth layer is a maximum pooling layer, the size is 3 multiplied by 3, the step length is 2, and the input of the eighth layer is the output of the seventh layer, so that the characteristic diagram of the original image is obtained.
The regional suggestion network for accurately generating the target consists of a regional pyramid network, a pooling layer and a full-connection layer, and specifically comprises the following steps: processing the feature information extracted by the convolutional neural network by using a feature pyramid network FPN, and integrating the high-level features of the low-resolution and high-semantic information and the low-level features of the high-resolution and low-semantic information to obtain a candidate region ROI; using an ROI pooling layer, mapping a candidate region ROI obtained by a previous layer of network to a corresponding position on a feature map, dividing the mapped region into regions with the same size, and performing maximum pooling operation on each region to obtain the feature map with a fixed size; and classifying and predicting the output result of the pooling layer by using the full-connection layer to obtain the type of each pixel on the original picture, thereby realizing the detection of the water drops.
Compared with the prior art, the invention has the beneficial effects that: the detection of the hot-rolled steel pipe is carried out by using three steps of data preprocessing, model training and online detection, in actual production, a three-dimensional laser sensor splices the contour lines of a pipe body into an arc, the surface of the steel pipe is fitted by using a least square method, the contour data is projected into a two-dimensional gray image, the image is preprocessed, a region to be detected is obtained, a water drop detection model is used for detection, the detection error is greatly reduced, and the method is suitable for popularization and use.
Drawings
FIG. 1 is a schematic view of the overall structure of a detection method for identifying water drops in three-dimensional detection of the outer surface of a seamless steel pipe according to the invention;
FIG. 2 is a schematic diagram of a feature extraction network in the detection method for identifying water drops in the three-dimensional detection of the outer surface of a seamless steel pipe according to the invention;
FIG. 3 is a schematic diagram of a water drop picture and a mark in the detection method for identifying water drops in the three-dimensional detection of the outer surface of the seamless steel pipe.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Examples
Referring to fig. 1-3, the present invention provides a technical solution:
a detection method for identifying water drops in three-dimensional detection of the outer surface of a seamless steel pipe comprises the following steps:
step 1: data preprocessing: generating a two-dimensional gray scale image of the surface of the pipe body according to pipe body profile data obtained by the three-dimensional sensor, and preprocessing the gray scale image to obtain all areas to be detected;
step 2: model training: constructing a deep learning network suitable for detecting water drops on the surface of the steel pipe in an off-line manner, and training a model in an off-line manner by using the image of the area to be detected collected in the step (1);
and step 3: online detection: and (3) loading the model obtained in the step (2) into detection equipment, and when the detection equipment works, obtaining all areas to be detected according to the step (1), and detecting water drops on the surface of the pipe body on line.
The following operations are carried out according to the method:
off-line training of a water drop detection model: 230 pictures obtained by pipe body contour line projection are collected, 119 water drops are marked on the pictures, 111 pictures without water drops are marked, 1080Ti is used as a GPU for training, and the training time is 12 hours.
Detecting water drops on line: a gotacor 2340 model three-dimensional laser sensor of lmi company is selected to collect pictures, the contour lines of the pipe body are spliced into circular arcs, the surface of the steel pipe is fitted by a least square method, and contour data are projected into a two-dimensional gray image.
The image is preprocessed to obtain an area to be detected, a water drop detection model is used for detecting, 223 pictures with water drops and 2675 pictures without water drops are collected, 208 pictures are correctly detected by an algorithm, 15 pictures are missed and 0 picture is mistakenly detected, the detection time of each frame of picture is 1.5ms, and the actual use requirement is met.
Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.

Claims (6)

1. A detection method for identifying water drops in three-dimensional detection of the outer surface of a seamless steel pipe comprises the following steps:
step 1: data preprocessing: generating a two-dimensional gray scale image of the surface of the pipe body according to pipe body profile data obtained by the three-dimensional sensor, and preprocessing the gray scale image to obtain all areas to be detected;
step 2: model training: constructing a deep learning network suitable for detecting water drops on the surface of the steel pipe in an off-line manner, and training a model in an off-line manner by using the image of the area to be detected collected in the step (1);
and step 3: online detection: and (3) loading the model obtained in the step (2) into detection equipment, and when the detection equipment works, obtaining all areas to be detected according to the step (1), and detecting water drops on the surface of the pipe body on line.
2. The detection method for identifying water drops in the three-dimensional detection of the outer surface of the seamless steel pipe according to claim 1, wherein the detection method comprises the following steps: the data preprocessing part is specifically as follows:
1) when the steel pipe passes through the detection device, the three-dimensional sensor acquires the profile data of the surface of the steel pipe to obtain three-dimensional point cloud data of the profile, a least square method is utilized to fit a reference plane of the surface of the steel pipe, points on the surface of the steel pipe, points protruding from the surface of the pipe body and points sinking are respectively projected into different numerical values, and the three-dimensional point cloud data are converted into two-dimensional gray images;
2) when the three-dimensional sensor collects the contour data of the surface of the steel pipe, the image uploaded to the industrial personal computer each time is only a part of the steel pipe, so that water drops on the image can be cut and distributed on two images, and in order to ensure the detection accuracy, the rear half part of the previous frame of image is spliced on the top of the current image;
3) performing corrosion-expansion-after-opening operation on the gray-scale image by using a 3 x 3 square structural element, removing isolated noise points, and marking a non-zero area on the image after the noise points are removed as a suspected area region;
4) extracting an edge contour edge of a suspected region, searching points on the edge out of the region, filtering a gray image by using a 3 x 3 matrix with a point P (x, y) on the edge contour as an origin, filtering the gray image, filtering out the points in the region, searching out of the region according to a formula (1), changing P1(x, y) into a new origin when a pixel point P1(x, y) with a gray value difference of 0.5 with the gray value of P (x, y) is searched, continuously searching out of the region for a pixel point with a gray value difference of 0.5 with the gray value of P1(x, y) until the searched points with the conditions are not met, and adding all searched points meeting the conditions into the original suspected region;
5) combining adjacent suspected areas to be detected, wherein the combining method comprises the steps of calculating the intersection ratio of the adjacent suspected areas, combining the suspected areas with the intersection ratio being more than 0.85 to be the areas to be detected, and the calculation mode of the intersection ratio IoU of the suspected areas regionA and regionB is as follows: the intersection of regionA and regionB is a proportion of the union. If the merge IoU of the region A and the region B is more than 0.85, the region A and the region B are merged into a region1 to be detected, and the region1 is the union of the region A and the region B.
3. The detection method for identifying water drops in the three-dimensional detection of the outer surface of the seamless steel pipe according to claim 1, wherein the detection method comprises the following steps: the model training part comprises the following concrete steps:
1) constructing a deep learning network suitable for detecting water drops on the surface of the steel pipe, wherein the deep learning network comprises a convolutional neural network for extracting characteristics and a regional suggestion network for accurately generating a target;
2) taking the image of the area to be detected as a training image, and training a water drop detection model off line: and collecting more than 100 pictures with water drops and more than 100 pictures without water drops as samples, and training by using a deep learning network until the loss function is less than 0.001 to obtain a water drop detection model.
4. The detection method for identifying water drops in the three-dimensional detection of the outer surface of the seamless steel pipe according to claim 1, wherein the detection method comprises the following steps: the online detection part specifically comprises:
1) when the steel pipe passes through the detection device, the three-dimensional sensor acquires the profile data of the surface of the steel pipe to obtain three-dimensional point cloud data of the profile, the surface of the steel pipe is fitted by using a least square method, and the profile data is projected into a two-dimensional gray image;
2) splicing the back half part of the previous frame of image on the top of the current image;
3) performing morphological operation on the image, and removing noise points to obtain a suspected area;
4) searching out of the area of points at the edge of the suspected area, and expanding the suspected area;
5) combining two adjacent suspected areas to be an area to be detected;
6) and taking all the areas to be detected on the picture as the input of a water drop detection model once, judging whether the areas to be detected are water drops or not, and displaying the identification result on a software interface.
5. The detection method for identifying water drops in the three-dimensional detection of the outer surface of the seamless steel pipe according to claim 3, wherein the detection method comprises the following steps: the convolutional neural network is used for extracting a convolutional neural network of features, normalizing input image data, adjusting the size of an image to 224 multiplied by 224, and extracting the features of each image by using 8 layers of convolutional neural networks, and specifically comprises the following steps: the first layer is a convolution layer, the convolution kernel is 11 multiplied by 11, the step length is 2, the second layer is a maximum pooling layer, the size is 3 multiplied by 3, the step length is 2, and the input of the second layer is the output of the first layer; the third layer is a convolutional layer, the convolutional kernel is 5 multiplied by 5, the step length is 2, the input of the third layer is the weighted sum of the output of the first layer and the output of the second layer, the fourth layer is a maximum pooling layer, the size is 3 multiplied by 3, the step length is 2, and the input of the fourth layer is the output of the third layer; the fifth layer is a convolutional layer, the size of the convolutional core is 3 multiplied by 3, the step length is 1, and the input of the fifth layer is the weighted sum of the output of the first layer, the output of the third layer and the output of the fourth layer; the sixth layer is a convolutional layer, the size and the step length of the convolutional kernel of the fifth layer are the same, and the input of the sixth layer is the weighted sum of the output of the first layer, the output of the third layer, the output of the fourth layer and the output of the fifth layer; the seventh layer is a convolutional layer, the size and the step length of the convolutional kernel are the same as those of the fifth layer, the input of the seventh layer is the weighted sum of the output of the first layer, the output of the third layer, the output of the fourth layer, the output of the fifth layer and the output of the sixth layer, the eighth layer is a maximum pooling layer, the size is 3 multiplied by 3, the step length is 2, and the input of the eighth layer is the output of the seventh layer, so that the characteristic diagram of the original image is obtained.
6. The detection method for identifying water drops in the nondestructive testing of the outer surface of the seamless steel pipe according to claim 3, characterized in that: the regional suggestion network for accurately generating the target consists of a regional pyramid network, a pooling layer and a full-connection layer, and specifically comprises the following steps: processing the feature information extracted by the convolutional neural network by using a feature pyramid network FPN, and integrating the high-level features of the low-resolution and high-semantic information and the low-level features of the high-resolution and low-semantic information to obtain a candidate region ROI; using an ROI pooling layer, mapping a candidate region ROI obtained by a previous layer of network to a corresponding position on a feature map, dividing the mapped region into regions with the same size, and performing maximum pooling operation on each region to obtain the feature map with a fixed size; and classifying and predicting the output result of the pooling layer by using the full-connection layer to obtain the type of each pixel on the original picture, thereby realizing the detection of the water drops.
CN202011048548.5A 2020-09-29 2020-09-29 Detection method for identifying water drops in three-dimensional detection of outer surface of seamless steel tube Active CN112198170B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011048548.5A CN112198170B (en) 2020-09-29 2020-09-29 Detection method for identifying water drops in three-dimensional detection of outer surface of seamless steel tube

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011048548.5A CN112198170B (en) 2020-09-29 2020-09-29 Detection method for identifying water drops in three-dimensional detection of outer surface of seamless steel tube

Publications (2)

Publication Number Publication Date
CN112198170A true CN112198170A (en) 2021-01-08
CN112198170B CN112198170B (en) 2023-09-29

Family

ID=74007846

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011048548.5A Active CN112198170B (en) 2020-09-29 2020-09-29 Detection method for identifying water drops in three-dimensional detection of outer surface of seamless steel tube

Country Status (1)

Country Link
CN (1) CN112198170B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114399467A (en) * 2021-12-15 2022-04-26 北京德风新征程科技有限公司 Case shell detection method and device, electronic equipment and computer readable medium
CN114820379B (en) * 2022-05-12 2024-04-26 中南大学 Image rain-like layer removing method for generating countermeasure network based on attention dual residual error

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109241818A (en) * 2017-07-11 2019-01-18 松下电器（美国）知识产权公司 Attachment detection method and device and system, attachment learning method and device, program
CN109447071A (en) * 2018-11-01 2019-03-08 博微太赫兹信息科技有限公司 A kind of mm-wave imaging dangerous goods detection method based on FPGA and deep learning
CN109598891A (en) * 2018-12-24 2019-04-09 中南民族大学 A kind of method and system for realizing Smoke Detection using deep learning disaggregated model
US20200065606A1 (en) * 2018-08-24 2020-02-27 Petrochina Company Limited Method and apparatus for automatically extracting image features of electrical imaging well logging

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109241818A (en) * 2017-07-11 2019-01-18 松下电器（美国）知识产权公司 Attachment detection method and device and system, attachment learning method and device, program
US20200065606A1 (en) * 2018-08-24 2020-02-27 Petrochina Company Limited Method and apparatus for automatically extracting image features of electrical imaging well logging
CN109447071A (en) * 2018-11-01 2019-03-08 博微太赫兹信息科技有限公司 A kind of mm-wave imaging dangerous goods detection method based on FPGA and deep learning
CN109598891A (en) * 2018-12-24 2019-04-09 中南民族大学 A kind of method and system for realizing Smoke Detection using deep learning disaggregated model

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114399467A (en) * 2021-12-15 2022-04-26 北京德风新征程科技有限公司 Case shell detection method and device, electronic equipment and computer readable medium
CN114399467B (en) * 2021-12-15 2022-11-11 北京德风新征程科技有限公司 Case shell detection method and device, electronic equipment and computer readable medium
CN114820379B (en) * 2022-05-12 2024-04-26 中南大学 Image rain-like layer removing method for generating countermeasure network based on attention dual residual error

Also Published As

Publication number Publication date
CN112198170B (en) 2023-09-29

Similar Documents

Publication Publication Date Title
Li et al. Recognition and evaluation of bridge cracks with modified active contour model and greedy search-based support vector machine
CN101609504B (en) Method for detecting, distinguishing and locating infrared imagery sea-surface target
CN109191459B (en) Automatic identification and rating method for continuous casting billet macrostructure center segregation defect
US20230108634A1 (en) Ground penetrating radar and deep learning-based underground pipeline detection method and system
Rau et al. Bridge crack detection using multi-rotary UAV and object-base image analysis
CN104574393A (en) Three-dimensional pavement crack image generation system and method
CN110473187B (en) Object-oriented line scanning three-dimensional pavement crack extraction method
CN111462140B (en) Real-time image instance segmentation method based on block stitching
CN113240623B (en) Pavement disease detection method and device
CN109584206B (en) Method for synthesizing training sample of neural network in part surface flaw detection
CN113393426A (en) Method for detecting surface defects of rolled steel plate
CN114596500A (en) Remote sensing image semantic segmentation method based on channel-space attention and DeeplabV3plus
CN111612747A (en) Method and system for rapidly detecting surface cracks of product
CN110147748A (en) A kind of mobile robot obstacle recognition method based on road-edge detection
CN110348307B (en) Path edge identification method and system for crane metal structure climbing robot
CN116645616A (en) Method for evaluating damage degree of building in natural disasters by using remote sensing images
CN111582270A (en) Identification tracking method based on high-precision bridge region visual target feature points
CN111476157A (en) Lane guide arrow recognition method under intersection monitoring environment
CN112198170B (en) Detection method for identifying water drops in three-dimensional detection of outer surface of seamless steel tube
CN115272189A (en) Pipeline defect rating method, terminal device and storage medium
CN115294541A (en) Local feature enhanced Transformer road crack detection method
Zhou et al. Road detection based on edge feature with GAC model in aerial image
CN114565753A (en) Unmanned aerial vehicle small target identification method based on improved YOLOv4 network
Han et al. Damage detection of quayside crane structure based on improved faster R-CNN
Zhang Online defect detection method of array optical sensor based on watershed image processing algorithm

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant