CN110261437A - A kind of natural gas station press device defect census method based on infrared thermal imagery - Google Patents
A kind of natural gas station press device defect census method based on infrared thermal imagery Download PDFInfo
- Publication number
- CN110261437A CN110261437A CN201910583100.4A CN201910583100A CN110261437A CN 110261437 A CN110261437 A CN 110261437A CN 201910583100 A CN201910583100 A CN 201910583100A CN 110261437 A CN110261437 A CN 110261437A
- Authority
- CN
- China
- Prior art keywords
- infrared
- image
- press device
- natural gas
- dst
- 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
- VNWKTOKETHGBQD-UHFFFAOYSA-N methane Chemical compound C VNWKTOKETHGBQD-UHFFFAOYSA-N 0.000 title claims abstract description 44
- 230000007547 defect Effects 0.000 title claims abstract description 25
- 238000000034 method Methods 0.000 title claims abstract description 22
- 239000003345 natural gas Substances 0.000 title claims abstract description 22
- 238000001514 detection method Methods 0.000 claims abstract description 31
- 238000012360 testing method Methods 0.000 claims abstract description 4
- 239000011159 matrix material Substances 0.000 claims description 10
- 238000009826 distribution Methods 0.000 claims description 6
- UPBAOYRENQEPJO-UHFFFAOYSA-N n-[5-[[5-[(3-amino-3-iminopropyl)carbamoyl]-1-methylpyrrol-3-yl]carbamoyl]-1-methylpyrrol-3-yl]-4-formamido-1-methylpyrrole-2-carboxamide Chemical compound CN1C=C(NC=O)C=C1C(=O)NC1=CN(C)C(C(=O)NC2=CN(C)C(C(=O)NCCC(N)=N)=C2)=C1 UPBAOYRENQEPJO-UHFFFAOYSA-N 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 6
- 230000004913 activation Effects 0.000 claims description 3
- 238000013461 design Methods 0.000 claims description 3
- 230000008569 process Effects 0.000 claims description 3
- 230000000644 propagated effect Effects 0.000 claims description 3
- 230000009467 reduction Effects 0.000 claims description 3
- 238000011946 reduction process Methods 0.000 claims description 3
- 238000000926 separation method Methods 0.000 claims description 3
- 238000012549 training Methods 0.000 claims description 3
- 238000013135 deep learning Methods 0.000 abstract description 6
- 230000007812 deficiency Effects 0.000 abstract description 3
- 238000011160 research Methods 0.000 abstract description 3
- 238000004088 simulation Methods 0.000 abstract description 3
- 230000006870 function Effects 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 3
- 238000011835 investigation Methods 0.000 description 3
- 238000003860 storage Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000005096 rolling process Methods 0.000 description 2
- 230000002378 acidificating effect Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000000903 blocking effect Effects 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 230000007797 corrosion Effects 0.000 description 1
- 238000005260 corrosion Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 238000001931 thermography Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N25/00—Investigating or analyzing materials by the use of thermal means
- G01N25/72—Investigating presence of flaws
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G06T5/70—
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/194—Segmentation; Edge detection involving foreground-background segmentation
-
- 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
- G06T2207/10024—Color image
-
- 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
- G06T2207/10048—Infrared image
-
- 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]
Abstract
Detection is artificially sampled in order to solve the detection technique of natural gas station press device in the prior art, it is time-consuming and laborious, and the problem of higher cost, the present invention provides a kind of natural gas station press device defect census method based on infrared thermal imagery that automation fast and low-cost detection may be implemented, it is characterized in that, include the following steps, S1: natural gas station press device infrared chart is obtained by infrared detector；S2: handled infrared chart and identified acquisition detection data；S3: data be will test and be stored in Cloud Server.The present invention can solve the problem of sample set deficiency by analogue simulation and live shooting, to deep learning algorithm is used on the software of independent research, to make entire census method more intelligent, efficient.
Description
Technical field
The present invention relates to equipment detection fields, lack more particularly to a kind of natural gas station press device based on infrared thermal imagery
Fall into census method.
Background technique
Since natural gas is in recovery process, it is often associated with a large amount of acidic materials, magazine, vapor etc., be easy to cause pressure
The problems such as corrosion, wear of hydraulic piping and press device, blocking, ponding.Press device is to belong to high-risk equipment, according to country
It is required that needing supervision and inspection and periodically generaI investigation.However, since yard number of devices is numerous, and structure is complicated for equipment component, in addition
The limitation of personnel and detection device bring great difficulty to census operations.
The all generally existing following disadvantage of the detection technique of conventional natural gas yard press device:
1. periodically, fixed point, is artificially sampled detection, time-consuming and laborious, and higher cost；
2. the data after detection are stored in local server, safety is low, and can not efficiently utilize detection data.
And Infrared Thermography Technology be it is a kind of do not stop transport rapidly and efficiently, or not do not sample, is pollution-free, non-contact and imaging it is intuitive
Technology.It is easy to operate, easy to carry, it can be achieved that a large amount of press devices of yard using infrared detector in conjunction with mobile phone
Quickly detection, greatly reduces drain on manpower and material resources；The infrared chart of acquisition is directly transmitted or copied to computer end simultaneously, is tied
The software systems for running hair jointly rapidly realize the intelligent recognition of image by deep learning algorithm, carry out to defect tracking dangerous
Grade classification, and data sheet is generated, dramatically reduce artificial workload；The data obtained after processing are uploaded to cloud
Database increases the safety of data, at the same between data share can also greatly promote its value.
It is therefore proposed that a kind of novel natural gas station press device defect based on infrared thermal imagery generally investigates technology, it is intended to
The shortcoming of traditional detection method is made up, bigger interests are created for enterprise.
Summary of the invention
It is artificially sampled detection in order to solve the detection technique of natural gas station press device in the prior art, it is time-consuming to take
Power, and the problem of higher cost, the present invention provide it is a kind of may be implemented automation fast and low-cost detection based on infrared thermal imagery
Natural gas station press device defect census method, which is characterized in that include the following steps,
S1: natural gas station press device infrared chart is obtained by infrared detector；
S2: handled infrared chart and identified acquisition detection data；
S3: data be will test and be stored in Cloud Server.
Further, the step S1 includes,
Infrared detector is connected with mobile phone and infrared detector shooting natural gas station pressure is held by user
Infrared chart is simultaneously sent directly to computer end progress subsequent operation by copy or network by the infrared chart of equipment；
Alternatively,
Infrared detector is carried on unmanned plane and the infrared of yard press device is realized by control unmanned plane
Thermal map detection simultaneously sends computer end progress subsequent processing by copy or wireless network for infrared chart；
Alternatively,
It shoots infrared chart using infrared video camera and computer end is sent by copy by infrared chart and carry out subsequent place
Reason.
Further, the step S2 includes,
S21: carrying out noise reduction process to image is imported using following formula,
dst1(x, y)=mid { src (x+x ', y+y ') }, (x ', y ') ∈ K1,
Wherein: src is input picture；dst1To export image；Mid is that sequence takes intermediate value；K1For customized matrix；
S22: carrying out binaryzation to image using following formula, realize the separation of image deflects and background,
Wherein: T is to set threshold value certainly；
S23: obtaining two-value defect image using following formula,
dst3(x, y)=min dst2(x+x ', y+y '), (x ', y ') ∈ K2, dst4(x, y)=max dst3(x+x ', y+
Y '),
In formula: K2For customized matrix；
S24: logic and operation is carried out using following formula and original image, obtains color defect figure:
dst5=dst4∩ src,
In formula: dst5Gained defect map as after final process.
Further, the step S2 includes,
S25: obtaining the characteristic point of image by convolutional layer,
Wherein: i, j indicate the i row j column of input picture dst5；K is the line number and columns of convolution kernel ω；M, n are expressed as rolling up
The m row n column of product core ω；B is customized bias term；F is activation primitive；G is then output result；
S26: by pond layer, reducing the inessential sample information of image,
h(i, j)=max (Kel(m, n)g(i+m, j+n)),
Wherein: Kel is customized convolution kernel, element all 1；
After S27 repeats step s25 and s26 several times, the classification of image is then carried out using following steps:
A, use following formula by above-mentioned output h dimensionality reduction for h '1×mThe matrix of (m=i × j) obtains real output value,
B, network propagated forward output result is returned as using softmax function by probability distribution using following formula:
C, the design of loss function
The gap between predicted value and true value is shown with quantized versions using intersection entropy function using following formula
To come, input value is probability required by above formula,
D, backpropagation
Weight is updated by the method that gradient declines using following formula, cross entropy is made to get minimum value
By constantly training, is restrained until cross entropy and meet preset value, obtain the distribution probability of image.
The invention has the advantages that
1 passes through on-the-spot investigation using technical solution of the present invention, and the staff of yard is usually by the detection data of acquisition
It is locally stored with file mode storage or Excel table.In case of accident, it is likely that cause loss of data and be difficult to look for
It returns, and the data that this mode stores easily are stolen by others.And stored by cloud database, then it can greatly promote
The safety of data, it is difficult to lose, and will not be stolen.And the data of storage beyond the clouds, it can be checked whenever and wherever possible
With share, greatly improve the use value of data.Therefore, user is after realizing video procession, by the number of acquisition
It is stored according to report by can directly upload cloud database after data copy either mobile phone shooting, guarantees the safety of data
Property and value maximization.
2 present invention can solve the problem of sample set deficiency, thus by deep learning by analogue simulation and live shooting
Algorithm is used on the software of independent research, to make entire census method more intelligent, efficient.
Detailed description of the invention
Fig. 1 is a kind of natural gas station press device defect census method flow chart based on infrared thermal imagery of the present invention.
Fig. 2 is one embodiment of the invention exploitation route map.
Specific embodiment
As shown in Figure 1, the present invention provides a kind of natural gas station press device defect census method based on infrared thermal imagery,
Include the following steps,
S1: natural gas station press device infrared chart is obtained by infrared detector；
S2: handled infrared chart and identified acquisition detection data；
S3: data be will test and be stored in Cloud Server.
Step s1 is specifically described below.
Infrared detector is connected with mobile phone and by the infrared chart of user's handheld device subject body, acquisition it is red
Outer thermal map is sent to computer end by copy or network and carries out subsequent operation；
Alternatively,
Infrared detector is carried on unmanned plane to and is realized by control unmanned plane the detection of yard press device,
It obtains image and computer end progress subsequent processing is sent to by copy or wireless network.This method compensates for tradition fixed point detection
Time-consuming and laborious defect, greatly improves detection efficiency.
Step s2 is specifically described below
After realizing quickly detection, the picture of acquisition is passed to the end the PC software systems of independent development, carries out the place of picture
Reason and identification.By extracting the defects of original image position, operand can be greatly reduced, arithmetic speed and accuracy are increased.
(1) the quick processing of infrared image.
1. image is all handled in a computer with the matrix of pixel value, noise reduction process is carried out to importing image first.
dst1(x, y)=mid { src (x+x ', y+y ') }, (x ', y ') ∈ K1 (1)
Wherein: src is input picture；dst1To export image；Mid is that sequence takes intermediate value；K1For customized matrix.
2. after noise reduction, carrying out binaryzation to image, realizing the separation of image deflects and background, calculate as follows.
Wherein: T is to set threshold value certainly.
3. carrying out morphological image operation, the two-value defect image of needs is obtained, formula is as follows.
dst3(x, y)=min dst2(x+x ', y+y '), (x ', y ') ∈ K2 (3)
dst4(x, y)=max dst3(x+x ', y+y ') (4)
In formula: K2For customized matrix.
4. carrying out logic and operation with original image, color defect figure can be obtained:
dst5=dst4∩src (5)
In formula: dst5Gained defect map as after final process.
(2) the quick identification of infrared image
After obtaining image deflects figure, defect recognition can be carried out to it by image recognition algorithm.Common image is known
Other algorithm has inverting, region growing, traditional neural network, deep learning etc..This detection method mainly uses deep learning algorithm
In convolutional neural networks carry out image quick identification.
1. obtaining the characteristic point of image by convolutional layer.
Wherein: i, j indicate the i row j column of input picture dst5；K is the line number and columns of convolution kernel ω；M, n are expressed as rolling up
The m row n column of product core ω；B is customized bias term；F is activation primitive；G is then output result.
2. being further reduced the inessential sample information of image by pond layer.
h(i, j)=max (Kel(m, n)g(i+m, j+n) (7)
Wherein: Kel is customized convolution kernel, but element all 1.
3. then carrying out the classification of image after repeating above-mentioned two steps operation.
It A, is h ' by above-mentioned output h dimensionality reduction1×mThe matrix of (m=i × j) obtains real output value.
B, network propagated forward output result is returned as probability distribution using softmax function:
C, the design of loss function
The gap between predicted value and true value can be showed with quantized versions by intersecting entropy function, and input value is
Probability required by above formula.
D, backpropagation
Weight is updated by the method for gradient decline, cross entropy is made to get minimum value
By constantly training, restrains and meet the requirements until cross entropy, can be obtained the distribution probability of image.
Currently, deep learning algorithm uses widely in recognition of face, vehicle detection, but in industrial detection
With or it is less, be primarily present following two reason:
(1) white silk sample is less, and detection accuracy is difficult to reach requirement；
(2) the detection difference of jljl volume defect is larger, and the network model that can be used for reference is less, and independent development is at high cost.
But by analogue simulation and live shooting, the present invention can solve the problem of sample set deficiency, thus by depth
Learning algorithm is used on the software of independent research, to make entire census method more intelligent, efficient.
Step s3 is specifically described below.
By on-the-spot investigation, the staff of yard usually the detection data of acquisition is stored with file mode or
Excel table is locally stored.In case of accident, it is likely that cause loss of data and be difficult to give for change, and this mode stores
Data be easily stolen by others.And stored by cloud database, then it can greatly promote the safety of data, it is difficult to
It loses, and will not be stolen.And the data of storage beyond the clouds, it can be checked and be shared whenever and wherever possible, greatly be promoted
The use value of data.Therefore, the data sheet of acquisition is passed through data copy after realizing video procession by user
Cloud database can be directly either uploaded after mobile phone shooting to be stored, and guarantee safety and the value maximization of data.It should
The exploitation route map of detection technique, as shown in Figure 2.
The foregoing is only a preferred embodiment of the present invention, is not intended to restrict the invention, for the skill of this field
For art personnel, the invention may be variously modified and varied.All within the spirits and principles of the present invention, made any to repair
Change, equivalent replacement, improvement etc., should all be included in the protection scope of the present invention.
Claims (4)
1. a kind of natural gas station press device defect census method based on infrared thermal imagery, which is characterized in that including following step
Suddenly,
S1: natural gas station press device infrared chart is obtained by infrared detector；
S2: handled infrared chart and identified acquisition detection data；
S3: data be will test and be stored in Cloud Server.
2. a kind of natural gas station press device defect census method based on infrared thermal imagery as described in claim 1, special
Sign is, the step S1 includes,
Infrared detector is connected with mobile phone and infrared detector shooting natural gas station press device is held by user
Infrared chart and by infrared chart by copy or network be sent directly to computer end carry out subsequent operation；
Alternatively,
Infrared detector is carried on unmanned plane to and is realized by control unmanned plane the infrared chart of yard press device
It detects and computer end is sent by copy or wireless network by infrared chart and carry out subsequent processing；
Alternatively,
It shoots infrared chart using infrared video camera and computer end is sent by copy by infrared chart and carry out subsequent processing.
3. a kind of natural gas station press device defect census method based on infrared thermal imagery as described in claim 1, special
Sign is, the step S2 includes,
S21: carrying out noise reduction process to image is imported using following formula,
dst1(x, y)=mid { src (x+x ', y+y ') }, (x ', y ') ∈ K1,
Wherein: src is input picture；dst1To export image；Mid is that sequence takes intermediate value；K1For customized matrix；
S22: carrying out binaryzation to image using following formula, realize the separation of image deflects and background,
Wherein: T is to set threshold value certainly；
S23: obtaining two-value defect image using following formula,
dst3(x, y)=min dst2(x+x ', y+y '), (x ',y′)∈K2,
dst4(x, y)=max dst3(x+x ', y+y '),
In formula: K2For customized matrix；
S24: logic and operation is carried out using following formula and original image, obtains color defect figure:
dst5=dst4∩ src,
In formula: dst5Gained defect map as after final process.
4. a kind of natural gas station press device defect census method based on infrared thermal imagery as claimed in claim 3, special
Sign is, the step S2 includes,
S25: obtaining the characteristic point of image by convolutional layer,
Wherein: i, j indicate the i row j column of input picture dst5；K is the line number and columns of convolution kernel ω；M, n are expressed as convolution kernel
The m row n of ω is arranged；B is customized bias term；F is activation primitive；G is then output result；
S26: by pond layer, reducing the inessential sample information of image,
h(i, j)=max (Kel(m, n)g(i+m, j+n)),
Wherein: Kel is customized convolution kernel, element all 1；
After S27 repeats step s25 and s26 several times, the classification of image is then carried out using following steps:
A, use following formula by above-mentioned output h dimensionality reduction for h '1×mThe matrix of (m=i × j) obtains real output value,
B, network propagated forward output result is returned as using softmax function by probability distribution using following formula:
C, the design of loss function
The gap between predicted value and true value is showed with quantized versions using intersection entropy function using following formula,
Input value is probability required by above formula,
D, backpropagation
Weight is updated by the method that gradient declines using following formula, cross entropy is made to get minimum value
By constantly training, is restrained until cross entropy and meet preset value, obtain the distribution probability of image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910583100.4A CN110261437B (en) | 2019-07-01 | 2019-07-01 | Natural gas station pressure equipment defect general investigation method based on infrared thermal image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910583100.4A CN110261437B (en) | 2019-07-01 | 2019-07-01 | Natural gas station pressure equipment defect general investigation method based on infrared thermal image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110261437A true CN110261437A (en) | 2019-09-20 |
| CN110261437B CN110261437B (en) | 2021-10-29 |
Family
ID=67923434
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910583100.4A Active CN110261437B (en) | 2019-07-01 | 2019-07-01 | Natural gas station pressure equipment defect general investigation method based on infrared thermal image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110261437B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116625582A (en) * | 2023-07-24 | 2023-08-22 | 上海安宸信息科技有限公司 | Movable gas leakage monitoring system for petroleum and petrochemical gas field station |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP0819244A1 (en) * | 1995-04-04 | 1998-01-21 | Bacharach, Inc. | Apparatus for imaging gas |
| CN102692429A (en) * | 2011-03-24 | 2012-09-26 | 中国科学院沈阳自动化研究所 | Method for automatic identification and detection of defect in composite material |
| US8759770B1 (en) * | 2013-04-08 | 2014-06-24 | General Electric Company | System and method for qualifying usability risk associated with subsurface defects in a multilayer coating |
| CN104122054A (en) * | 2014-07-27 | 2014-10-29 | 北京航星网讯技术股份有限公司 | Infrared thermal imaging type laser gas leak detection method and apparatus |
| CN105758889A (en) * | 2016-03-30 | 2016-07-13 | 哈尔滨工业大学 | Oil and gas pipeline infrared thermal wave imaging detection system and method |
| CN107677700A (en) * | 2017-08-08 | 2018-02-09 | 上海交通大学 | PCB single board failure detector |
| CN107957434A (en) * | 2017-12-27 | 2018-04-24 | 电子科技大学 | A kind of Non-Destructive Testing Enhancement Method of compound carbon fiber board internal flaw |
| CN108717069A (en) * | 2018-05-29 | 2018-10-30 | 电子科技大学 | A kind of high-pressure bottle thermal imaging imperfection detection method based on the segmentation of row variable step |
| CN109358093A (en) * | 2018-11-28 | 2019-02-19 | 武汉市锅炉压力容器检验研究所 | A kind of fiber-wound gas cylinder infrared thermal imaging detection system and its detection method |
| CN109740603A (en) * | 2019-01-21 | 2019-05-10 | 闽江学院 | Based on the vehicle character identifying method under CNN convolutional neural networks |
| CN109919905A (en) * | 2019-01-08 | 2019-06-21 | 浙江大学 | A kind of Infrared Non-destructive Testing method based on deep learning |
-
2019
- 2019-07-01 CN CN201910583100.4A patent/CN110261437B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP0819244A1 (en) * | 1995-04-04 | 1998-01-21 | Bacharach, Inc. | Apparatus for imaging gas |
| CN102692429A (en) * | 2011-03-24 | 2012-09-26 | 中国科学院沈阳自动化研究所 | Method for automatic identification and detection of defect in composite material |
| US8759770B1 (en) * | 2013-04-08 | 2014-06-24 | General Electric Company | System and method for qualifying usability risk associated with subsurface defects in a multilayer coating |
| CN104122054A (en) * | 2014-07-27 | 2014-10-29 | 北京航星网讯技术股份有限公司 | Infrared thermal imaging type laser gas leak detection method and apparatus |
| CN105758889A (en) * | 2016-03-30 | 2016-07-13 | 哈尔滨工业大学 | Oil and gas pipeline infrared thermal wave imaging detection system and method |
| CN107677700A (en) * | 2017-08-08 | 2018-02-09 | 上海交通大学 | PCB single board failure detector |
| CN107957434A (en) * | 2017-12-27 | 2018-04-24 | 电子科技大学 | A kind of Non-Destructive Testing Enhancement Method of compound carbon fiber board internal flaw |
| CN108717069A (en) * | 2018-05-29 | 2018-10-30 | 电子科技大学 | A kind of high-pressure bottle thermal imaging imperfection detection method based on the segmentation of row variable step |
| CN109358093A (en) * | 2018-11-28 | 2019-02-19 | 武汉市锅炉压力容器检验研究所 | A kind of fiber-wound gas cylinder infrared thermal imaging detection system and its detection method |
| CN109919905A (en) * | 2019-01-08 | 2019-06-21 | 浙江大学 | A kind of Infrared Non-destructive Testing method based on deep learning |
| CN109740603A (en) * | 2019-01-21 | 2019-05-10 | 闽江学院 | Based on the vehicle character identifying method under CNN convolutional neural networks |
Non-Patent Citations (4)
| Title |
|---|
| 万帅: "典型石化高温设备监测数据的处理研究", 《化工装备技术》 * |
| 刘炳集: "基于时频图和CNN的滚动轴承故障诊断", 《南昌航空大学学报:自然科学版》 * |
| 王威: "基于卷积神经网络的红外热成像罐车内壁裂纹识别", 《红外技术》 * |
| 韩燮: "基于卷积神经网络的棋子定位和识别方法", 《激光与光电子学进展》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116625582A (en) * | 2023-07-24 | 2023-08-22 | 上海安宸信息科技有限公司 | Movable gas leakage monitoring system for petroleum and petrochemical gas field station |
| CN116625582B (en) * | 2023-07-24 | 2023-09-19 | 上海安宸信息科技有限公司 | Movable gas leakage monitoring system for petroleum and petrochemical gas field station |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110261437B (en) | 2021-10-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106980641A (en) | The quick picture retrieval system of unsupervised Hash and method based on convolutional neural networks | |
| He et al. | Design of multi-scale receptive field convolutional neural network for surface inspection of hot rolled steels | |
| Hu et al. | CDUNet: Cloud detection UNet for remote sensing imagery | |
| CN109284741A (en) | A kind of extensive Remote Sensing Image Retrieval method and system based on depth Hash network | |
| Guo et al. | DF-SSD: a deep convolutional neural network-based embedded lightweight object detection framework for remote sensing imagery | |
| Mo et al. | Point-by-point feature extraction of artificial intelligence images based on the Internet of Things | |
| CN114550053A (en) | Traffic accident responsibility determination method, device, computer equipment and storage medium | |
| CN116468392A (en) | Method, device, equipment and storage medium for monitoring progress of power grid engineering project | |
| Kun et al. | A deep learning‐based method for pixel‐level crack detection on concrete bridges | |
| Chen et al. | Multi-scale attention feature aggregation network for cloud and cloud shadow segmentation | |
| CN110261437A (en) | A kind of natural gas station press device defect census method based on infrared thermal imagery | |
| Ma et al. | CNN-TransNet: A Hybrid CNN-Transformer Network with Differential Feature Enhancement for Cloud Detection | |
| Peng et al. | New network based on D-LinkNet and densenet for high resolution satellite imagery road extraction | |
| CN116383470B (en) | Image searching method with privacy protection function | |
| CN115546569B (en) | Attention mechanism-based data classification optimization method and related equipment | |
| Xie et al. | Yolov4-mobilenetv2-dw-lcarm: A real-time ship detection network | |
| Pan et al. | Land cover classification based on machine learning using UAV multi-spectral images | |
| CN112633116B (en) | Method for intelligently analyzing PDF graphics context | |
| Fan et al. | CM-YOLOv8: Lightweight YOLO for Coal Mine Fully Mechanized Mining Face | |
| Wang et al. | Optimized faster R-CNN for oil wells detection from high-resolution remote sensing images | |
| Liu et al. | Research on object detection algorithm based on uva aerial image | |
| Wang et al. | Efficient and robust homography estimation using compressed convolutional neural network | |
| Niu et al. | Real-time recognition and location of indoor objects | |
| Wang | Combining UPerNet and ConvNeXt for Contrails Identification to reduce Global Warming | |
| Jaju et al. | Detection of cloud cover in satellite imagery using semantic segmentation |
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 |