CN111382773A - Image matching method based on nine-grid principle for monitoring inside of pipeline - Google Patents
Image matching method based on nine-grid principle for monitoring inside of pipeline Download PDFInfo
- Publication number
- CN111382773A CN111382773A CN201811651344.3A CN201811651344A CN111382773A CN 111382773 A CN111382773 A CN 111382773A CN 201811651344 A CN201811651344 A CN 201811651344A CN 111382773 A CN111382773 A CN 111382773A
- Authority
- CN
- China
- Prior art keywords
- image
- pipeline
- matching
- similarity
- palace
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000012544 monitoring process Methods 0.000 title claims abstract description 23
- 238000000034 method Methods 0.000 title claims abstract description 20
- 239000013598 vector Substances 0.000 claims abstract description 31
- 238000010801 machine learning Methods 0.000 claims abstract description 10
- 238000010606 normalization Methods 0.000 claims abstract description 7
- 238000012545 processing Methods 0.000 claims description 10
- 230000002159 abnormal effect Effects 0.000 claims description 5
- 239000000284 extract Substances 0.000 claims description 5
- 239000003086 colorant Substances 0.000 claims description 4
- 210000004291 uterus Anatomy 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 abstract description 7
- 238000006243 chemical reaction Methods 0.000 abstract 1
- 238000005516 engineering process Methods 0.000 description 7
- 238000004458 analytical method Methods 0.000 description 5
- 238000001514 detection method Methods 0.000 description 3
- 238000013459 approach Methods 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- VNWKTOKETHGBQD-UHFFFAOYSA-N methane Chemical compound C VNWKTOKETHGBQD-UHFFFAOYSA-N 0.000 description 2
- 238000004891 communication Methods 0.000 description 1
- 238000005260 corrosion Methods 0.000 description 1
- 230000007797 corrosion Effects 0.000 description 1
- 239000013078 crystal Substances 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 239000003345 natural gas Substances 0.000 description 1
- 239000003208 petroleum Substances 0.000 description 1
- 238000012805 post-processing Methods 0.000 description 1
- 230000002285 radioactive effect Effects 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/06—Recognition of objects for industrial automation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention provides an image matching method based on the nine-grid principle for monitoring the interior of a pipeline. Firstly, the system performs machine learning on picture sets of various conditions in the pipeline, each picture is sequentially subjected to gray level conversion, normalization, image division, calculation of pixel characteristic values of each palace and embedding of the characteristic values of each palace into a one-dimensional nine-palace vector, and light weight storage in a robot singlechip is realized. And after the robot carries a camera and enters a pipeline, the acquired video is subjected to frame taking in real time, the frame picture is quickly converted into a one-dimensional nine-palace vector, the similarity is calculated with a nine-palace vector set stored in a single chip microcomputer, and when the similarity is greater than the preset matching degree, the matching is successful. The image matching method has simple machine learning process, does not need complex mathematical operation, can realize lightweight algorithm and data set storage, reduces equipment cost, simultaneously improves monitoring efficiency in the pipeline, accelerates project progress, and also provides powerful guarantee for project safety.
Description
Technical Field
The invention relates to the field of image processing, in particular to a video image matching method based on the nine-grid principle for monitoring the interior of a pipeline.
Background
Modern industries and lives are not separated from various pipelines, such as petroleum pipelines, natural gas pipelines, power transmission pipelines and water transmission pipelines. The actual demands on the automation of pipeline monitoring nowadays create conditions for the development and application of pipeline monitoring technology. However, many of the related technologies are still immature, dangerous to operate, and expensive. For example, on-line corrosion monitoring, radioactive source-based monitoring and magnetic flaw detection-based monitoring have the advantages that although the monitoring effect can meet the requirement, the equipment and the technology are complex, monopolized by large-scale companies, the matched equipment is expensive, and the use cost is extremely high. Although the price of equipment using a camera for video monitoring in a pipeline is relatively low, the equipment is limited by various factors, for example, image processing, machine learning and real-time video processing need a powerful computer, a small-sized single chip microcomputer cannot process the equipment, and how the equipment and a pipeline robot are carried to enable the equipment and the pipeline robot to enter the pipeline is often a problem. Meanwhile, because of the large amount of calculation, the amount of power consumed by the mounted robot is increased, which makes the robot have to mount more batteries. This results in the whole weight and the volume increase of robot, has not only seriously influenced robot's work efficiency, has reduced the flexibility ratio of robot in the pipeline, has also increased the possibility of meeting danger in the work moreover. Because the pipeline internal conditions are bad, the condition that the video collection process is interrupted and resumed often appears, so if the robot only collects images in the pipeline and does not perform real-time processing of data, the robot takes out the video for analysis by the staff after cruising, the exact position corresponding to the video picture for finding the pipeline problem is difficult to confirm when the post-processing is often caused. This approach also does not fully function as an automatic monitoring system. At present, there is also a technology that a machine moves in parallel outside a pipeline, the robot inside the pipeline does not perform data processing, video data is sent to the machine outside the pipeline only through wireless equipment in real time, and the machine outside the pipeline performs real-time operation. Although the workload of the robot in the pipeline is reduced, the pipeline is buried deeply or is in a harsh natural environment or geographical environment outside the pipeline, so that the robot outside the pipeline cannot synchronize with the robot in the pipeline or cannot receive wireless communication data sent by the robot in the pipeline in actual operation.
If the light-weighted image calculation can complete the analysis of the conditions in the pipeline, the performance of the robot in the pipeline and the monitoring working efficiency of the robot can be greatly improved, the automation degree of the robot is increased, and the cost is reduced. The image matching is an image processing technology core technology for finding abnormal conditions in the pipeline, and the robot can automatically judge the conditions in the pipeline and store information through matching with normal condition pictures and various abnormal condition pictures in a learning library. In addition, image matching technology is widely applied in many fields such as satellite remote sensing, automatic navigation of space vehicles, terminal guidance and target seeking of weapon projection systems, image target tracking of optics and radars, analysis and detection of earth resources, weather forecast, medical diagnosis, character reading, change detection in scene analysis and the like. Today's image matching algorithms are mainly divided into two categories: one is a grayscale matching based approach; another class is feature matching based methods. The former mainly uses a spatial one-dimensional or two-dimensional sliding template for image matching, and the difference of different algorithms is mainly in the aspect of selection of the template and related criteria, and the method generally has high matching rate, large calculation amount and low speed; in the latter, salient features such as points, lines, areas and the like are extracted from an original image and used as matching primitives, and then the salient features are used for feature matching, and the matching speed is generally high, but the matching precision is not necessarily high.
Disclosure of Invention
The invention aims to provide a video image matching method for pipeline internal monitoring based on the nine-grid principle and only requiring lightweight computation.
In order to achieve the purpose, the invention provides a video image matching method for monitoring the interior of a pipeline based on the nine-grid principle, which comprises the steps of performing machine learning on images in front of the pipeline under a robot and matching the images behind the pipeline under the robot in real time.
The image machine learning in front of the pipeline under the robot comprises the following steps:
leading in an internal picture set and an abnormal picture set of a normal condition pipeline by a system;
secondly, converting the image into a gray scale, and converting the three primary colors or four primary colors of images with different formats into a uniform gray scale image;
normalizing the gray scale of the image, normalizing the pixel value of the image to be within the range of 0-255, wherein Xnew = ((X-Xmin)/(Xmax-Xmin)) × 255, X is the pixel value of the original image, Xmin is the minimum value of the pixel value of the original image, Xmax is the maximum value of the pixel value of the original image, Xnew is the pixel value after normalization, and the image which is too bright or too dark can be adjusted into a unified normal image through normalization;
step four: dividing the image into a plurality of parts, namely dividing the image into n x n parts, wherein n is a nine-grid parameter;
calculating the characteristic value of each uterus;
embedding the characteristic values of each palace into a one-dimensional vector;
step seven: and storing the one-dimensional vector of each image into the single chip microcomputer database.
The image matching work after the robot descends the pipeline comprises the following steps:
the robot singlechip takes out images acquired by the camera in real time, namely images to be matched;
nine-grid processing is carried out on the image to be matched to form a one-dimensional vector with the same length as the image in the single chip microcomputer database;
step ten, calculating the similarity between the picture to be matched and the image in the singlechip database;
step eleven: if the similarity is higher than a preset threshold value R1, the matching is successful, and the storage information is as follows: matching with the X condition successfully, wherein the X condition is the condition corresponding to the matched image in the database;
step twelve: if the similarity is lower than a preset threshold value R1, respectively taking out numerical values lower than the average values of all the vectors in the nine-grid one-dimensional vectors of the image to be detected and the database image to recombine into a new one-dimensional vector, wherein the numerical values lower than the average values of all the vectors in the gray-scale image are black and white, and therefore the numerical values lower than the average values of all the vectors are relatively dark parts in the image;
step thirteen: calculating the similarity of the two new one-dimensional vectors;
fourteen steps: if the similarity is higher than a preset threshold value R2, the matching is successful, and the storage information is as follows: matching with the situation X successfully, and finally judging that the information is undetermined; because the image collected in the actual work has deviation with the image in the database, the system extracts and matches the part with higher gray so as to avoid missing judgment;
step fifteen: if the similarity is lower than a preset threshold R2, the matching is not successful, and the system extracts the next picture in the database for matching.
The invention realizes light-weight image matching calculation, the real-time image matching only needs to carry out similarity calculation on one-dimensional vectors, and machine learning only needs to convert the images into the one-dimensional vectors, thereby reducing the calculation amount of a singlechip. In actual operation, a common single chip microcomputer with a 16MHz crystal oscillator can realize all the work of the patent. The analysis of the condition in the pipeline is completed, the energy consumption of the robot in the pipeline is not greatly increased, the monitoring working efficiency is improved, the automation degree is increased, and the cost is reduced.
Drawings
Fig. 1 is a flow chart of image machine learning before a pipeline under a robot for monitoring video image matching method based on the nine-grid principle in the invention.
Fig. 2 is a flow chart of real-time matching of images after pipeline in a robot for monitoring video image matching method based on the nine-grid principle in the pipeline interior.
Detailed description of the preferred embodiments
The technical solutions in the embodiments of the present invention will be clearly and completely described below, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all 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.
Embodiments of the present invention will be described below with reference to the accompanying drawings.
Fig. 1 is an image machine learning part in front of a pipeline under a robot in a video image matching method for monitoring the interior of the pipeline based on the nine-palace principle, which comprises the following steps:
leading in an internal picture set and an abnormal picture set of a normal condition pipeline by a system;
converting the image into a GRAY scale, converting three-primary-color or four-primary-color images with different formats into a unified GRAY scale image, for example, converting an RGB three-channel photo into a GRAY single-channel picture;
normalizing the gray scale of the image, normalizing the pixel value of the image to be within the range of 0-255, wherein Xnew = ((X-Xmin)/(Xmax-Xmin)) > 255, X is the pixel value of the original image, Xmin is the minimum value of the pixel value of the original image, Xmax is the maximum value of the pixel value of the original image, Xnew is the pixel value after normalization, the image which is too bright or too dark can be adjusted to be a unified normal image through normalization, for example, the original pixel value is 122-;
step four: dividing the image into n × n parts, where n is a nine-grid parameter, for example, when n =15, the image is divided into 15 × 15=225 small parts, namely 255 grids;
calculating characteristic values in each palace, namely the average value and median of all pixels, such as 255 palaces, and calculating to obtain 255 average values and 255 medians;
embedding the mean value and median of each palace into a one-dimensional vector, for example, 255 palaces have 510 mean values and median to form a number array with the size of 1 x 510, namely the one-dimensional vector;
step seven: and storing the one-dimensional vector of each image into the single chip microcomputer database.
FIG. 2 is an image matching working part used for pipeline internal monitoring and based on the nine-palace principle after a robot descends a pipeline in a video image matching method, and the image matching working part comprises the following steps:
the robot singlechip takes out images acquired by the camera in real time, namely images to be matched;
nine-grid processing is carried out on the image to be matched to form a one-dimensional vector with the same length as the image in the single chip microcomputer database;
step ten, calculating the similarity between the picture to be matched and the image in the singlechip database,namely cosine Similarity:
；
step eleven: if the similarity is higher than a preset threshold value R1, the matching is successful, and the storage information is as follows: matching with the X condition successfully, wherein the X condition is the condition corresponding to the matched image in the database;
step twelve: if the similarity is lower than a preset threshold value R1, respectively taking out numerical values lower than the average values of all the vectors in the nine-grid one-dimensional vectors of the image to be detected and the database image to recombine into a new one-dimensional vector, wherein the numerical values lower than the average values of all the vectors in the gray-scale image are black and white, and therefore the numerical values lower than the average values of all the vectors are relatively dark parts in the image;
step thirteen: calculating the cosine similarity of the two new one-dimensional vectors;
fourteen steps: if the cosine similarity is higher than a preset threshold value R2, matching is successful, and the stored information is as follows: matching with the situation X successfully, and finally judging that the information is undetermined; because the image collected in the actual work has deviation with the image in the database, the system extracts and matches the part with higher gray so as to avoid missing judgment;
step fifteen: if the cosine similarity is lower than a preset threshold value R2, the matching is unsuccessful, and the system extracts the next picture in the database for matching.
Claims (4)
1. A video image matching method based on the nine-grid principle for monitoring the interior of a pipeline is characterized by comprising the following processing steps of:
firstly, starting an image machine learning part in front of a pipeline under a robot, and leading a system into an internal picture set and an abnormal picture set of the pipeline under a normal condition;
secondly, converting the image into a gray scale, and converting the three primary colors or four primary colors of images with different formats into a uniform gray scale image;
thirdly, normalizing the image gray level;
step four: dividing the image into a plurality of parts, namely dividing the image into n x n parts, wherein n is a nine-grid parameter;
calculating the characteristic value of each palace;
embedding the characteristic values of each palace into a one-dimensional vector;
step seven: storing the one-dimensional vector of each image into a single-chip microcomputer database, and finishing the machine learning part of the image in front of the pipeline under the robot;
starting image matching work after the robot enters the pipeline, and taking out images acquired by a camera in real time by a single chip microcomputer of the robot, namely images to be matched;
nine-grid processing is carried out on the image to be matched to form a one-dimensional vector with the same length as the image in the single chip microcomputer database;
step ten, calculating the similarity between the picture to be matched and the image in the singlechip database;
step eleven: if the similarity is higher than a preset threshold value R1, the matching is successful, and the storage information is as follows: matching with the X condition successfully, wherein the X condition is the condition corresponding to the matched image in the database;
step twelve: if the similarity is lower than a preset threshold value R1, respectively taking out the values which are lower than the average values of all the vectors in the nine-grid one-dimensional vectors of the image to be detected and the database image, and recombining the values into a new one-dimensional vector;
step thirteen: calculating the similarity of the two new one-dimensional vectors;
fourteen steps: if the similarity is higher than a preset threshold value R2, the matching is successful, and the storage information is as follows: matching with the situation X successfully, and finally judging that the information is undetermined;
step fifteen: if the similarity is lower than a preset threshold R2, the matching is not successful, and the system extracts the next picture in the database for matching.
2. The method of claim 1, wherein in the third step, the normalization of the pixel values of the image is performed to normalize all the pixel values of the image to a range of 0-255, Xnew = ((X-Xmin)/(Xmax-Xmin)) = 255, X is the original pixel value, Xmin is the original pixel value minimum, Xmax is the original pixel value maximum, and Xnew is the pixel value after the normalization.
3. The method as claimed in claim 1, wherein in step five, the characteristic value of each uterus includes the average value and median of all pixels in the uterus.
4. The nine-palace-principle-based video image matching method for pipeline interior monitoring according to claim 1, wherein in the ten to fifteen steps, the similarity comprises a cosine similarity.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811651344.3A CN111382773A (en) | 2018-12-31 | 2018-12-31 | Image matching method based on nine-grid principle for monitoring inside of pipeline |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811651344.3A CN111382773A (en) | 2018-12-31 | 2018-12-31 | Image matching method based on nine-grid principle for monitoring inside of pipeline |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111382773A true CN111382773A (en) | 2020-07-07 |
Family
ID=71218335
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811651344.3A Pending CN111382773A (en) | 2018-12-31 | 2018-12-31 | Image matching method based on nine-grid principle for monitoring inside of pipeline |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111382773A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115265669A (en) * | 2022-09-19 | 2022-11-01 | 博格达智能装备(南通)有限公司 | Pipe cutting hot melting process quality detection system based on two classifiers |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103116890A (en) * | 2013-02-27 | 2013-05-22 | 中山大学 | Video image based intelligent searching and matching method |
| CN106503605A (en) * | 2015-09-01 | 2017-03-15 | 南京理工大学 | Human body target recognition methods based on stereovision technique |
| US20170322951A1 (en) * | 2010-03-29 | 2017-11-09 | Ebay Inc. | Finding products that are similar to a product selected from a plurality of products |
-
2018
- 2018-12-31 CN CN201811651344.3A patent/CN111382773A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170322951A1 (en) * | 2010-03-29 | 2017-11-09 | Ebay Inc. | Finding products that are similar to a product selected from a plurality of products |
| CN103116890A (en) * | 2013-02-27 | 2013-05-22 | 中山大学 | Video image based intelligent searching and matching method |
| CN106503605A (en) * | 2015-09-01 | 2017-03-15 | 南京理工大学 | Human body target recognition methods based on stereovision technique |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115265669A (en) * | 2022-09-19 | 2022-11-01 | 博格达智能装备(南通)有限公司 | Pipe cutting hot melting process quality detection system based on two classifiers |
| CN115265669B (en) * | 2022-09-19 | 2023-10-13 | 博格达智能装备(南通)有限公司 | Pipe cutting hot melting process quality detection system based on two classifiers |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107808133B (en) | Unmanned aerial vehicle line patrol-based oil and gas pipeline safety monitoring method and system and software memory | |
| CN106295613A (en) | A kind of unmanned plane target localization method and system | |
| CN116579616B (en) | Risk identification method based on deep learning | |
| CN103679740B (en) | ROI (Region of Interest) extraction method of ground target of unmanned aerial vehicle | |
| CN112652020A (en) | Visual SLAM method based on AdaLAM algorithm | |
| CN114092793A (en) | End-to-end biological target detection method suitable for complex underwater environment | |
| CN116310997A (en) | Deep learning-based marine small target detection method | |
| CN111382773A (en) | Image matching method based on nine-grid principle for monitoring inside of pipeline | |
| CN111767826A (en) | Timing fixed-point scene abnormity detection method | |
| CN115471782B (en) | Unmanned ship-oriented infrared ship target detection method and device | |
| CN111127355A (en) | Method for finely complementing defective light flow graph and application thereof | |
| CN110826478A (en) | Aerial photography illegal building identification method based on countermeasure network | |
| CN110345919A (en) | Space junk detection method based on three-dimensional space vector and two-dimensional plane coordinate | |
| CN112800932B (en) | Method for detecting remarkable ship target in offshore background and electronic equipment | |
| CN110991387B (en) | Distributed processing method and system for robot cluster image recognition | |
| CN114219989A (en) | Foggy scene ship instance segmentation method based on interference suppression and dynamic contour | |
| Mai et al. | Method to perform 3d localization of text in shipboard point cloud data using corresponding 2d image | |
| CN112906538A (en) | Ship target image recognition system and method based on unattended platform | |
| Özer et al. | An approach based on deep learning methods to detect the condition of solar panels in solar power plants | |
| CN210488578U (en) | Data processing device of unmanned aerial vehicle image data algorithm | |
| CN113971679B (en) | Ocean tide measuring method based on computer vision and image processing | |
| Zhang et al. | Power Plant Indicator Light Detection System Based on Improved YOLOv5 | |
| Ouyang et al. | A Lightweight Object Detector with Deformable Upsampling for Marine Organism Detection | |
| Lu et al. | Two-stage Infrared Images Photovoltaic Panel Extraction Based on Deep Semantic Segmentation | |
| Wang et al. | Structures Detection Based on CLSK Model Combined with Shadow Information Using High Resolution Remote Sensing Images |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20200707 |