CN107516315A - A kind of development machine based on machine vision is slagged tap monitoring method - Google Patents

A kind of development machine based on machine vision is slagged tap monitoring method Download PDF

Info

Publication number
CN107516315A
CN107516315A CN201710743406.2A CN201710743406A CN107516315A CN 107516315 A CN107516315 A CN 107516315A CN 201710743406 A CN201710743406 A CN 201710743406A CN 107516315 A CN107516315 A CN 107516315A
Authority
CN
China
Prior art keywords
mrow
image
tap
msup
monitoring
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
CN201710743406.2A
Other languages
Chinese (zh)
Other versions
CN107516315B (en
Inventor
张浪文
何昌传
谢巍
吴伟林
余孝源
何伟
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.)
South China University of Technology SCUT
Original Assignee
South China University of Technology SCUT
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 South China University of Technology SCUT filed Critical South China University of Technology SCUT
Priority to CN201710743406.2A priority Critical patent/CN107516315B/en
Publication of CN107516315A publication Critical patent/CN107516315A/en
Application granted granted Critical
Publication of CN107516315B publication Critical patent/CN107516315B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/0002Inspection of images, e.g. flaw detection
    • G06T7/0004Industrial image inspection
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/23Clustering techniques
    • G06F18/232Non-hierarchical techniques
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/24Classification techniques
    • G06F18/241Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2413Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
    • G06F18/24133Distances to prototypes
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/10Segmentation; Edge detection
    • G06T7/11Region-based segmentation
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/10Segmentation; Edge detection
    • G06T7/187Segmentation; Edge detection involving region growing; involving region merging; involving connected component labelling
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/20Special algorithmic details
    • G06T2207/20112Image segmentation details
    • G06T2207/20152Watershed segmentation
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/30Subject of image; Context of image processing
    • G06T2207/30108Industrial image inspection
    • G06T2207/30132Masonry; Concrete

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Quality & Reliability (AREA)
  • Image Processing (AREA)
  • Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)

Abstract

Slagged tap monitoring method the invention discloses a kind of development machine based on machine vision, including:Country rock is slagged tap degree of crushing monitoring step, conveyer belt load monitoring step and rocvk classification monitoring step;Degree of crushing monitoring step of slagging tap counts to the full-size, minimum dimension and average-size of rock ballast, there is provided the breaking surrounding rock ability under current boring performance;Conveyer belt load monitoring step is slagged tap by detection accounts for the depth of whole belt feeder to assess the load condition of current belt machine;Rocvk classification monitoring step judges the current situation for tunneling country rock by surface characteristics of slagging tap.The problems such as method of present invention machine vision improves efficiency, the accuracy rate for monitoring of slagging tap, and monitoring caused by avoiding severe ring change border is inaccurate, while cost of labor can also be reduced.

Description

A kind of development machine based on machine vision is slagged tap monitoring method
Technical field
The present invention relates to machine vision applications technical field, more particularly to a kind of development machine based on machine vision is slagged tap prison Survey method.
Background technology
21 century is the information-based epoch, realizes that modern production and information system management need to carry out not in the mode of production Reform disconnectedly.Development machine application technology, era step should be closelyed follow, strengthen automatic management, to adapt to informatization requirement.So And current development machine detecting system is to be based on various sensors, data receiver analysis is realized either by wireless device Development machine condition monitoring system based on virtual instrument;Under rugged environment, sensor can be effected by environmental factors, leads The problems such as causing gathered data inaccuracy or device damage, it is unfavorable for situation that is real-time, accurate, stably monitoring development machine;And And it is required for putting into many human and material resources in terms of equipment and maintenance.
In recent years, industry manufacture 2025 imply that Machine Vision Detection be China's industry towards intelligent direction develop must Right trend, have great importance to industry towards intellectuality, automation aspect development.Machine vision technique level is still continuous Lifting, based on Embedded NI Vision Builder for Automated Inspection by as future developing direction;Embedded system can be regarded in real time Feel IMAQ, visual pattern processing control, there is the characteristics of compact-sized, cost is low, low in energy consumption.The base of NI Vision Builder for Automated Inspection This feature is that speed is fast, contain much information, precision is high and non-contact, can be greatly enhanced flexibility and the automation journey of production Degree.Be not suitable for the environment of manual work at some, artificial vision is difficult to meet the occasion for requiring or having high-volume repetitive operation, Artificial vision is replaced to greatly improve production efficiency with machine vision.
Machine vision technique is blended with other sensing technologies and a trend, and multi-sensor technology is in detection, tracking With the reliability that system can be improved in terms of target identification.For the detection of slagging tap of development machine, the environment of detection can compare Badly, acquired picture quality is poor, if machine vision technique combined with other sensors technology, utilizes sensor sheet The superiority of body makes up the deficiency of image, can will preferably complete the detection slagged tap.
The content of the invention
The shortcomings that it is an object of the invention to overcome prior art and deficiency, there is provided a kind of development machine based on machine vision Slag tap monitoring method, improve efficiency, the accuracy rate of monitoring of slagging tap with the method for machine vision, avoid prison caused by severe ring change border The problems such as indeterminacy is true, while cost of labor can also be reduced.
The purpose of the present invention is realized by following technical scheme:
A kind of development machine based on machine vision is slagged tap monitoring method, including:Country rock is slagged tap degree of crushing monitoring step, defeated Send bringing onto load monitoring step and rocvk classification monitoring step;Slag tap degree of crushing monitoring step to the full-size of rock ballast, most Small size and average-size are counted, there is provided the breaking surrounding rock ability under current boring performance;Conveyer belt load monitoring step Slagged tap by detection and account for the depth of whole belt feeder to assess the load condition of current belt machine;Rocvk classification monitoring step passes through Slag tap surface characteristics, judge the situation of current driving country rock.
Specifically, operating procedure is as follows:
S1, image of slagging tap is strengthened, the pretreatment of denoising;
S2, country rock are slagged tap degree of crushing monitoring step:Pass through pretreated image of slagging tap based on above-mentioned, first to image Binary conversion treatment is carried out, then carries out inverse processing, finally by white connected region in mark image, finds out area maximum Connected region, border white block is removed, and outline, middle white block are clinker, calculate all middle white blocks Maximum, minimum value, average value and the standard deviation of area, the degree of crushing slagged tap is assessed by the size for counting clinker;
S3, conveyer belt load monitoring step:It is pretreated based on step S1 after the degree of crushing analysis for completing to slag tap Slag tap image, by identifying position and the line of roller bearing on conveyer belt, obtain belt sideline, identified using line detection algorithm The tangent line slagged tap with conveyer belt boundary, and then the top width L2 and face width L1 that slags tap on belt is calculated, finally calculate what is slagged tap The load of machinery systems:
H is belt depth, can be according to equation below computational load:
Wherein, what d was represented is load percentage, for symbolizing the loading condition of slag crust band;
S4, rocvk classification monitoring step:
For training sample, country rock is classified using clustering algorithm, for the pretreated images of step S1, drawn Gradation of image value histogram, extract the input of the average gray value of gradation of image value histogram, the peak value of histogram as cluster Variable, obtained after training per sorting criterion of a kind of center of mass point as test input;
When new images are inputted, the feature of the image is first extracted, includes the average gray of gradation of image value histogram It is worth, the peak value of histogram, then the class according to belonging to below equation calculates the image:
Wherein, what Pi was represented is every a kind of center of mass point;What P was represented is the characteristic point of current input image;Dis is current The distance of characteristic point and class center of mass point, in this, as the standard of classification.
Preferably, step S1 pre-treatment steps are specially:Gray scale is carried out to the video image of slagging tap of camera collection first Conversion, obtains containing noisy gradient image of slagging tap;Then, using small probability strategy and Two-dimensional Maximum Ostu method to image Split, the regional for the image that obtains slagging tap:Interference noise region, texture region and smooth region;Finally, using difference The fractional calculus mask of order is handled regional, obtains self-adaptive solution and enhanced image of slagging tap.
Specifically, being handled using the fractional calculus mask of different orders regional, adaptively gone Make an uproar and enhanced image of slagging tap:Defined according to fractional order G-L, when the v orders of fractional order are positive numbers to differentiate, when point The v orders of number rank are integral operation when being negative:
As v > 0, the fractional order differential that G-L defines lower v orders is:
Wherein,Represent that G-L defines lower fractional order integration operator, subscript G-L represents G-L definition, subscript v expressions V rank differential is sought, subscript a and t represent the upper bound and the lower bound of integration type, and a is time t initial value;
As v < 0, the fractional order integration formula under G-L is defined is:
The image of the fractional order mask for having change order and camera collection is subjected to convolution algorithm and can be obtained by enhancing Image of slagging tap afterwards.
Preferably, in step S2, it is contemplated that the light lack of uniformity for video image of slagging tap, according to the distribution characteristics of image, Fragmentation threshold binaryzation is used to the different piece for image of slagging tap, to mark the white in video image of slagging tap to connect to greatest extent Region.
Specifically, in step S2, the profile of stone is extracted using the OpenCV operator findContours provided, can be with Select the profile of the image after binary-state threshold;The focus point of profile is determined by extracting the general profile of stone, As the seed point in the stone region, connect according to being obtained after the burn into expansive working of the actual conditions of image progress appropriateness Get off to need the sign image used, the region of needs is isolated using watershed algorithm;Finally, provided using OpenCV ContoursArea functions sort to ask for the area that profile includes to area, find out the stone in each frame input picture Minimum and maximum area, and calculate the standard deviation of average area and area.
Preferably, consider that the color component value of roller bearing is bigger, uses the enhanced roller bearings of step S1 from whole image The single channel image of color component carries out thresholding to image, obtains the profile in roller bearing region, try to achieve roller bearing as process object Two end points of the focus point of profile as straight line, connect into belt sideline.
Preferably, the training sample for needing to provide in step S4 is { x(1),…,x(m), eachDo not correspond to Group indication, it is accurate as the classification of test input in order to which known sample is aggregated into K class and obtains the center of mass point of class Then, the detailed process of image clustering of slagging tap can be expressed as follows:
Step 1) randomly selects K cluster center of mass point (cluster centroids)
Step 2) repeats procedure below until convergence:
To sample i, the class belonging to it is judged by calculating distance:
c(i):=argminj||x(i)j||2
To classification j, center of mass point is calculated again according to all samples for belonging to such:
K is the number that given needs are classified, c(i)What is represented is sample i in the distance versus of the center of mass point with K class That class in small distance, barycenter μjWhat is represented is the center of a sample's point for belonging to same class.
The present invention compared with prior art, has the following advantages that and beneficial effect:
Mechanical vision inspection technology is applied to the slagging tap in terms of monitoring of development machine by the present invention, for transmission on-load, The indexs such as clinker degree of crushing, rocvk classification identification monitor in real time;Efficiency, the standard of monitoring of slagging tap are improved with the method for machine vision True rate, avoid monitoring the problems such as inaccurate caused by severe ring change border, while cost of labor can also be reduced.
Brief description of the drawings
Fig. 1 be embodiment in development machine slag tap monitoring image shoot schematic diagram.
Fig. 2 is that development machine is slagged tap monitoring system general flow chart in embodiment.
Fractional order denoisings and Enhancement Method flow chart of the Fig. 3 for image of being slagged tap in embodiment.
Fig. 4 is degree of crushing monitoring flow chart of being slagged tap in embodiment.
Fig. 5 is degree of crushing monitoring processing schematic diagram of being slagged tap in embodiment.
Fig. 6 is load monitoring flow chart in embodiment.
Fig. 7 is that the belt sideline information based on machine vision identifies schematic diagram in embodiment.
Fig. 8 is the profile analysis figure of slag tap belt conveyer in embodiment.
Fig. 9 is rocvk classification flow chart in embodiment.
Embodiment
With reference to embodiment and accompanying drawing, the present invention is described in further detail, but embodiments of the present invention are unlimited In this.
Embodiment
The vision technique used during video image is handled relates generally to following a few class technologies:1) increasing of image By force:Fractional order algorithm for image enhancement, the marginal information of enhancing detection target, is easy to extract;2) contours extract of image:Extraction inspection The profile of target is surveyed, determines position and the area information of target;3) K-means clustering algorithms:Characteristics of image is extracted, to image Classified;4) straight-line detection of image:Detection conveyer belt and the boundary straight line in region of slagging tap, obtain detection zone, reduce ring The interference that border is brought, improve processing speed etc..
OpenCV image procossing api function storehouses and VS2010 IDEs:OpenCV is at an image increased income Function library is managed, is developed by C and C Plus Plus, including image procossing storehouse and machine vision algorithm storehouse two parts composition, can be cross-platform Use, the function that OpenCV is provided can facilitate programmer's call function storehouse to realize image algorithm, and be the realization of image algorithm Multiple interfaces, image procossing interface and matrix class computing interface are provided, while also cover many advanced mathematical function such as Fu In leaf transformation, integral operation, calculus of differences etc., for the function of the function of camera modeling, it is possible to achieve to the school of camera Quasi- function.
Visual Studio IDEs (IDE), are one of first choices of software development under windows platform. Visual Studio 2010 provide new template, design tool and test and debugging acid for client, in Web exploitations Also upgrading is optimized, the api function that VS2010 is provided can make user easily use Windows system resources.
The present embodiment provides a kind of development machine based on machine vision and slagged tap monitoring method, such as Fig. 1, is gathered by camera Slag tap the view data of conveyer belt, the area of slagging tap, conveying bringing onto load and rocvk classification of processing region are entered by this method Row identification.
Fig. 2 is total flow chart of this method, be divided into country rock slag tap degree of crushing monitoring, conveyer belt load monitoring and enclose Rock classification three aspects of monitoring.Degree of crushing monitoring modular of slagging tap enters to the full-size, minimum dimension and average-size of rock ballast Row statistics, there is provided the breaking surrounding rock ability under current boring performance.Conveyer belt load monitoring module is slagged tap by detection to be accounted for entirely The depth of belt feeder assesses the load condition of current belt machine.Rocvk classification monitoring modular is judged by surface characteristics of slagging tap The situation of current driving country rock.Comprise the following steps that:
S1, it is high-speed cruising (speed is in 2m/s or so) due to belt feeder, the image of slagging tap of acquisition has certain mould Paste property, it is therefore desirable to strengthened image of slagging tap, the pretreatment of denoising.
Fig. 3 is the flow that fractional order image enhaucament is carried out to the video monitoring image of acquisition.
Greyscale transformation is carried out to the video image of slagging tap of camera collection first, obtains containing noisy gradient map of slagging tap Picture;Then, image is split using small probability strategy and Two-dimensional Maximum Ostu method, each area for the image that obtains slagging tap Domain:Interference noise region, texture region and smooth region, finally, using the fractional calculus mask of different orders to each Region is handled, and obtains self-adaptive solution and enhanced image of slagging tap.
Image of slagging tap inevitably is influenceed during collection and transmission by noise, causes image information not Certainty, difficulty is brought for follow-up image processing process.Conventional non-local mean filtering, Kalman filtering, Wavelet image Denoising, although and all there is a certain degree of denoising effect the methods of medium filtering, LPF, Wiener filtering, this A little Image denoising algorithms all directly or indirectly employ integer rank integration in the structure of denoising model, can so be made an uproar in removal The texture information of image is lost while sound.Denoising is carried out to image using fractional order integration not have to pre-estimate image Noise variance, and processing is directly filtered, thus compared to other Denoising Algorithms, fractional order integration algorithm is in image denoising side Face has higher efficiency, while also has more preferable effect in terms of image texture detailed information is retained.
In order to effectively carry out feature extraction to image of slagging tap, this method utilizes small probability strategy and Two-dimensional Maximum inter-class variance Method is split to image, and regional is handled using the fractional calculus mask of different orders, obtains adaptive Denoising and enhanced image of slagging tap.Defined according to fractional order G-L, when the v orders of fractional order are positive numbers to differentiate, when It is integral operation when the v orders of fractional order are negatives:
As v > 0, the fractional order differential that Griinwald-Letnikov (G-L) defines lower v orders is:
Wherein,Represent that G-L defines lower fractional order integration operator, subscript G-L represents G-L definition, subscript v expressions V rank differential is sought, subscript a and t represent the upper bound and the lower bound of integration type, and a is time t initial value.
As v < 0, the fractional order integration formula under G-L is defined is:
The image of the fractional order mask for having change order and camera collection is subjected to convolution algorithm and can be obtained by enhancing Image of slagging tap afterwards, enhanced picture noise are reduced, and texture becomes apparent from, and edge is more obvious.
The G-L of table 1 defines mask
v(v-1)/2 0 v(v-1)/2 0 v(v-1)/2
0 -v -v -v 0
v(v-1)/2 -v 8 -v v(v-1)/2
0 -v -v -v 0
v(v-1)/2 0 v(v-1)/2 0 v(v-1)/2
As shown in table 1, the 5*5 for the fractional order that G-L is defined mask is provided to realize fractional order image enhaucament.
S2, accompanying drawing 4 are the flow chart that degree of crushing monitors of slagging tap.Pass through pretreated image of slagging tap based on above-mentioned, such as Shown in accompanying drawing 5, binary conversion treatment is carried out to image first, then carries out inverse processing, finally, by marking white in image to connect Logical region, the maximum connected region of area is found out, removing border white block, (the maximum white UNICOM region of area is exactly border White block), and outline, middle white block are clinker, calculate maximum, the minimum of all middle white block areas Value, average value and standard deviation, the degree of crushing slagged tap is assessed by the size for counting clinker.
In view of the light lack of uniformity for video image of slagging tap, according to the distribution characteristics of image, the present embodiment is to figure of slagging tap The different piece of picture uses fragmentation threshold binaryzation, to mark the white connected region in video image of slagging tap to greatest extent.
This method extracts the profile of stone using the OpenCV operator findContours () provided, can select two The profile of image after value thresholding;The focus point of profile is determined by extracting the general profile of stone, as The seed point in the stone region, following needs are obtained after carrying out the burn into expansive working of appropriateness according to the actual conditions of image The sign image used, the region of needs is isolated using watershed algorithm.Finally, provided using OpenCV ContoursArea () function sorts to ask for the area that profile includes to area, finds out the stone in each frame input picture The minimum and maximum area of block, and calculate the standard deviation of average area and area.
S3, the flow chart that accompanying drawing 6 is load of machinery systems monitoring.After the degree of crushing analysis for completing to slag tap, regarded based on slagging tap Frequency image, by identifying position and the line of red roller bearing on conveyer belt, the outer peripheral tangent line of conveyer belt is obtained, is become using Hough (line detection algorithm) is changed to identify the tangent line slagged tap and had a common boundary with conveyer belt, and then calculates top width L2 and surface of slagging tap on belt Width L1, finally calculate the load of machinery systems slagged tap.
As shown in fig. 7, consider that the red color component value of roller bearing is bigger, enhanced red using step S1 from whole image The single channel image of colouring component carries out thresholding to image, obtains the profile in roller bearing region, try to achieve roller wheel as process object Two end points of the wide focus point as straight line, connect into belt sideline.Image profile structure as shown in Figure 8 is obtained, L2 is Top width degree on belt, L1 are face width of slagging tap, and H is that (H is varied less belt depth with load, and dynamic estimation is to can consider Immobilize)., can be according to equation below computational load according to profile:
Wherein, what d was represented is load percentage, for symbolizing the loading condition of slag crust band.It can be moved according to formula (3) The load percentage that state estimation is slagged tap.
S4, the flow chart that accompanying drawing 9 is rocvk classification.This method is classified using K-means clustering algorithms to country rock, pin The image pretreated to step S1, draws gradation of image value histogram, extract gradation of image value histogram average gray value, Input variable of the peak value of histogram as cluster, is obtained per a kind of center of mass point after training.
The training sample for needing to provide is { x(1),…,x(m), eachWithout corresponding group indication.In order to The sorting criterion that the center of mass point that known sample aggregates into K class and obtains class is inputted as test, image clustering of slagging tap Detailed process can be expressed as follows:
Step 1) randomly selects K cluster center of mass point (cluster centroids)
Step 2) repeats procedure below until convergence:
To sample i, the class belonging to it is judged by calculating distance:
c(i):=argminj||x(i)j||2
To classification j, center of mass point is calculated again according to all samples for belonging to such:
K is the number that given needs are classified, c(i)What is represented is sample i in the distance versus of the center of mass point with K class That class in small distance.Barycenter μjWhat is represented is the center of a sample's point for belonging to same class.
Therefore, when new images are inputted, first extracting the feature of the image, (including gradation of image value histogram is averaged The peak value of gray value, histogram), the then class according to belonging to below equation calculates the image:
Wherein, what Pi was represented is every a kind of center of mass point;What P was represented is the characteristic point of current input image;Dis is current The distance of characteristic point and class center of mass point, in this, as the standard of classification.
The clustering method provided based on above-mentioned steps, development machine is slagged tap and carries out Classification and Identification, mainly including granite, mud The types such as soil, igneous rock, sedimentary rock, metamorphic rock, realize the classification situation based on monitoring picture gray value analysis estimation country rock.
Above-described embodiment is the preferable embodiment of the present invention, but embodiments of the present invention are not by above-described embodiment Limitation, other any Spirit Essences without departing from the present invention with made under principle change, modification, replacement, combine, simplification, Equivalent substitute mode is should be, is included within protection scope of the present invention.

Claims (8)

  1. The monitoring method 1. a kind of development machine based on machine vision is slagged tap, it is characterised in that including:Country rock slag tap degree of crushing prison Survey step, conveyer belt load monitoring step and rocvk classification monitoring step;Slag tap degree of crushing monitoring step to rock ballast most Large scale, minimum dimension and average-size are counted, there is provided the breaking surrounding rock ability under current boring performance;Convey bringing onto load Monitoring step is slagged tap by detection accounts for the depth of whole belt feeder to assess the load condition of current belt machine;Rocvk classification monitors Step judges the current situation for tunneling country rock by surface characteristics of slagging tap.
  2. The monitoring method 2. development machine according to claim 1 based on machine vision is slagged tap, it is characterised in that concrete operations Step is as follows:
    S1, image of slagging tap is strengthened, the pretreatment of denoising;
    S2, country rock are slagged tap degree of crushing monitoring step:Pass through pretreated image of slagging tap based on above-mentioned, image is carried out first Binary conversion treatment, inverse processing is then carried out, finally by white connected region in mark image, find out the maximum connection of area Region, border white block is removed, and outline, middle white block are clinker, calculate all middle white block areas Maximum, minimum value, average value and standard deviation, the degree of crushing slagged tap is assessed by the size for counting clinker;
    S3, conveyer belt load monitoring step:After the degree of crushing analysis for completing to slag tap, slag tap based on step S1 is pretreated Image, by identifying position and the line of roller bearing on conveyer belt, belt sideline is obtained, is slagged tap using line detection algorithm to identify The tangent line having a common boundary with conveyer belt, and then the top width L2 and face width L1 that slags tap on belt is calculated, finally calculate the load slagged tap Degree:
    H is belt depth, can be according to equation below computational load:
    <mrow> <mi>d</mi> <mo>=</mo> <mrow> <mo>(</mo> <mfrac> <mrow> <mi>L</mi> <mn>1</mn> </mrow> <mrow> <mi>L</mi> <mn>2</mn> </mrow> </mfrac> <mo>-</mo> <mfrac> <mrow> <mi>L</mi> <mn>1</mn> <mrow> <mo>(</mo> <mi>L</mi> <mn>2</mn> <mo>-</mo> <mi>L</mi> <mn>1</mn> <mo>)</mo> </mrow> <mi>t</mi> <mi>a</mi> <mi>n</mi> <mi>&amp;alpha;</mi> </mrow> <mrow> <mi>L</mi> <mn>2</mn> <mi>H</mi> </mrow> </mfrac> <mo>)</mo> </mrow> <mo>*</mo> <mn>100</mn> <mi>%</mi> </mrow>
    Wherein, what d was represented is load percentage, for symbolizing the loading condition of slag crust band;
    S4, rocvk classification monitoring step:
    For training sample, country rock is classified using clustering algorithm, for the pretreated images of step S1, draws image Grey value histograms, the input variable of the average gray value of gradation of image value histogram, the peak value of histogram as cluster is extracted, Obtained after training per sorting criterion of a kind of center of mass point as test input;
    When new images are inputted, the feature of the image is first extracted, including it is the average gray value of gradation of image value histogram, straight The peak value of square figure, the then class according to belonging to below equation calculates the image:
    <mrow> <mi>d</mi> <mi>i</mi> <mi>s</mi> <mo>=</mo> <msqrt> <mrow> <msup> <mrow> <mo>(</mo> <mi>P</mi> <mi>i</mi> <mo>.</mo> <mi>x</mi> <mo>-</mo> <mi>P</mi> <mo>.</mo> <mi>x</mi> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mo>+</mo> <msup> <mrow> <mo>(</mo> <mi>P</mi> <mi>i</mi> <mo>.</mo> <mi>y</mi> <mo>-</mo> <mi>P</mi> <mo>.</mo> <mi>y</mi> <mo>)</mo> </mrow> <mn>2</mn> </msup> </mrow> </msqrt> </mrow>
    Wherein, what Pi was represented is every a kind of center of mass point;What P was represented is the characteristic point of current input image;Dis is current signature Point and the distance of class center of mass point, in this, as the standard of classification.
  3. The monitoring method 3. development machine according to claim 2 based on machine vision is slagged tap, it is characterised in that step S1 is pre- Processing step is specially:Greyscale transformation is carried out to the video image of slagging tap of camera collection first, obtains slagging tap containing noisy Gradient image;Then, image is split using small probability strategy and Two-dimensional Maximum Ostu method, obtains image of slagging tap Regional:Interference noise region, texture region and smooth region;Finally, using the fractional calculus mask of different orders Regional is handled, obtains self-adaptive solution and enhanced image of slagging tap.
  4. The monitoring method 4. development machine according to claim 3 based on machine vision is slagged tap, it is characterised in that using different The fractional calculus mask of order is handled regional, obtains self-adaptive solution and enhanced image of slagging tap:Root Defined according to fractional order G-L, be product when the v orders of fractional order are negatives when the v orders of fractional order are positive numbers to differentiate Partite transport is calculated:
    As v > 0, the fractional order differential that G-L defines lower v orders is:
    <mrow> <mmultiscripts> <mi>D</mi> <mi>t</mi> <mi>v</mi> <mi>a</mi> <mrow> <mi>G</mi> <mo>-</mo> <mi>L</mi> </mrow> </mmultiscripts> <mi>f</mi> <mrow> <mo>(</mo> <mi>t</mi> <mo>)</mo> </mrow> <mo>=</mo> <munder> <mi>lim</mi> <mrow> <mi>h</mi> <mo>&amp;RightArrow;</mo> <mn>0</mn> </mrow> </munder> <msup> <mi>h</mi> <mrow> <mo>-</mo> <mi>v</mi> </mrow> </msup> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>j</mi> <mo>=</mo> <mn>0</mn> </mrow> <mrow> <mo>&amp;lsqb;</mo> <mrow> <mo>(</mo> <mi>t</mi> <mo>-</mo> <mi>a</mi> <mo>)</mo> </mrow> <mo>/</mo> <mi>h</mi> <mo>&amp;rsqb;</mo> </mrow> </munderover> <msup> <mrow> <mo>(</mo> <mo>-</mo> <mn>1</mn> <mo>)</mo> </mrow> <mi>j</mi> </msup> <mfenced open = "(" close = ")"> <mtable> <mtr> <mtd> <mi>v</mi> </mtd> </mtr> <mtr> <mtd> <mi>j</mi> </mtd> </mtr> </mtable> </mfenced> <mi>f</mi> <mrow> <mo>(</mo> <mi>t</mi> <mo>-</mo> <mi>j</mi> <mi>h</mi> <mo>)</mo> </mrow> </mrow>
    Wherein,Represent that G-L defines lower fractional order integration operator, subscript G-L represents G-L definition, and v ranks are asked in subscript v expressions Differential, subscript a and t represent the upper bound and the lower bound of integration type, and a is time t initial value;
    As v < 0, the fractional order integration formula under G-L is defined is:
    <mrow> <mmultiscripts> <mi>I</mi> <mi>t</mi> <mi>v</mi> <mi>a</mi> <mrow> <mi>G</mi> <mo>-</mo> <mi>L</mi> </mrow> </mmultiscripts> <mi>f</mi> <mrow> <mo>(</mo> <mi>t</mi> <mo>)</mo> </mrow> <mo>=</mo> <mmultiscripts> <mi>D</mi> <mi>t</mi> <mrow> <mo>-</mo> <mi>v</mi> </mrow> <mi>a</mi> <mrow> <mi>G</mi> <mo>-</mo> <mi>L</mi> </mrow> </mmultiscripts> <mi>f</mi> <mrow> <mo>(</mo> <mi>t</mi> <mo>)</mo> </mrow> <mo>=</mo> <munder> <mi>lim</mi> <mrow> <mi>h</mi> <mo>&amp;RightArrow;</mo> <mn>0</mn> </mrow> </munder> <msup> <mi>h</mi> <mi>v</mi> </msup> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>j</mi> <mo>=</mo> <mn>0</mn> </mrow> <mrow> <mo>&amp;lsqb;</mo> <mrow> <mo>(</mo> <mi>t</mi> <mo>-</mo> <mi>a</mi> <mo>)</mo> </mrow> <mo>/</mo> <mi>h</mi> <mo>&amp;rsqb;</mo> </mrow> </munderover> <msup> <mrow> <mo>(</mo> <mo>-</mo> <mn>1</mn> <mo>)</mo> </mrow> <mi>j</mi> </msup> <mfenced open = "(" close = ")"> <mtable> <mtr> <mtd> <mrow> <mo>-</mo> <mi>v</mi> </mrow> </mtd> </mtr> <mtr> <mtd> <mi>j</mi> </mtd> </mtr> </mtable> </mfenced> <mi>f</mi> <mrow> <mo>(</mo> <mi>t</mi> <mo>-</mo> <mi>j</mi> <mi>h</mi> <mo>)</mo> </mrow> </mrow>
    To have the image progress convolution algorithm that fractional order mask and the camera of change order gather can be obtained by it is enhanced Slag tap image.
  5. The monitoring method 5. development machine according to claim 2 based on machine vision is slagged tap, it is characterised in that step S2 In, it is contemplated that the light lack of uniformity for video image of slagging tap, according to the distribution characteristics of image, the different piece for image of slagging tap is adopted With fragmentation threshold binaryzation, to mark the white connected region in video image of slagging tap to greatest extent.
  6. The monitoring method 6. development machine according to claim 2 based on machine vision is slagged tap, it is characterised in that step S2 In, the profile of stone is extracted using the OpenCV operator findContours provided, after binary-state threshold being selected Image profile;The focus point of profile is determined by extracting the general profile of stone, as the stone region Seed point, the marking pattern for next needing to use is obtained after the burn into expansive working of appropriateness is carried out according to the actual conditions of image Picture, the region of needs is isolated using watershed algorithm;Finally, asked for using the contoursArea functions of OpenCV offers The area that profile includes, and area is sorted, the minimum and maximum area of the stone in each frame input picture is found out, and count Calculate the standard deviation of average area and area.
  7. The monitoring method 7. development machine according to claim 2 based on machine vision is slagged tap, it is characterised in that step S3 In, consider that the color component value of roller bearing is bigger, uses the list of the color component of the enhanced roller bearings of step S1 from whole image Channel image carries out thresholding to image, obtains the profile in roller bearing region, try to achieve the focus point of roller bearing profile as process object As two end points of straight line, belt sideline is connected into.
  8. The monitoring method 8. development machine according to claim 2 based on machine vision is slagged tap, it is characterised in that in step S4 The training sample for needing to provide is { x(1),…,x(m), eachWithout corresponding group indication, in order to by known to Sample aggregates into K class and obtains sorting criterion of the center of mass point of class as test input, the specific mistake for image clustering of slagging tap Journey is as follows:
    Step 1) randomly selects K cluster center of mass point
    Step 2) repeats procedure below until convergence:
    To sample i, the class belonging to it is judged by calculating distance:
    c(i):=arg minj||x(i)j||2
    To classification j, center of mass point is calculated again according to all samples for belonging to such:
    <mrow> <msub> <mi>&amp;mu;</mi> <mi>j</mi> </msub> <mo>:</mo> <mo>=</mo> <mfrac> <mrow> <msubsup> <mi>&amp;Sigma;</mi> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>m</mi> </msubsup> <mn>1</mn> <mo>{</mo> <msup> <mi>c</mi> <mrow> <mo>(</mo> <mi>i</mi> <mo>)</mo> </mrow> </msup> <mo>=</mo> <mi>j</mi> <mo>}</mo> <msup> <mi>x</mi> <mrow> <mo>(</mo> <mi>i</mi> <mo>)</mo> </mrow> </msup> </mrow> <mrow> <msubsup> <mi>&amp;Sigma;</mi> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>m</mi> </msubsup> <mn>1</mn> <mo>{</mo> <msup> <mi>c</mi> <mrow> <mo>(</mo> <mi>i</mi> <mo>)</mo> </mrow> </msup> <mo>=</mo> <mi>j</mi> <mo>}</mo> </mrow> </mfrac> </mrow>
    K is the number that given needs are classified, c(i)What is represented is sample i distances in the distance versus of the center of mass point with K class That less class, barycenter μjWhat is represented is the center of a sample's point for belonging to same class.
CN201710743406.2A 2017-08-25 2017-08-25 Tunneling machine slag tapping monitoring method based on machine vision Active CN107516315B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710743406.2A CN107516315B (en) 2017-08-25 2017-08-25 Tunneling machine slag tapping monitoring method based on machine vision

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710743406.2A CN107516315B (en) 2017-08-25 2017-08-25 Tunneling machine slag tapping monitoring method based on machine vision

Publications (2)

Publication Number Publication Date
CN107516315A true CN107516315A (en) 2017-12-26
CN107516315B CN107516315B (en) 2020-04-28

Family

ID=60724315

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710743406.2A Active CN107516315B (en) 2017-08-25 2017-08-25 Tunneling machine slag tapping monitoring method based on machine vision

Country Status (1)

Country Link
CN (1) CN107516315B (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108734703A (en) * 2018-04-26 2018-11-02 广东道氏技术股份有限公司 Polished bricks printing pattern detection method, system and device based on machine vision
CN109002777A (en) * 2018-06-29 2018-12-14 电子科技大学 A kind of infrared small target detection method towards complex scene
CN109305534A (en) * 2018-10-25 2019-02-05 安徽理工大学 Coal wharf's belt conveyor self-adaptation control method based on computer vision
CN110188624A (en) * 2019-05-10 2019-08-30 国网福建省电力有限公司龙岩供电公司 A kind of substation's wind drift recognition methods and system based on deep learning

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104502990A (en) * 2015-01-06 2015-04-08 中国地质大学(武汉) Geological survey method of tunnel face based on digital image
CN206016814U (en) * 2016-09-18 2017-03-15 中铁工程装备集团有限公司 A kind of record and the device for analyzing TBM slag pieces

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104502990A (en) * 2015-01-06 2015-04-08 中国地质大学(武汉) Geological survey method of tunnel face based on digital image
CN206016814U (en) * 2016-09-18 2017-03-15 中铁工程装备集团有限公司 A kind of record and the device for analyzing TBM slag pieces

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
KUNUI HONG ET AL.: "Determination of geological strength index of jointed rock mass based on image processing", 《JOURNAL OF ROCK MECHANICS AND GEOTECHNICAL ENGINEERING》 *
杨柱中 等: "基于分数阶微分的图像增强", 《计算机辅助设计与图形学学报》 *
范朝东 等: "基于小概率策略的Otsu图像分割方法", 《电子与信息学报》 *
袭著晖 ET AL.: "双护盾TBM施工隧洞围岩分类方法及应用", 《土工基础》 *

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108734703A (en) * 2018-04-26 2018-11-02 广东道氏技术股份有限公司 Polished bricks printing pattern detection method, system and device based on machine vision
CN108734703B (en) * 2018-04-26 2020-12-01 广东道氏技术股份有限公司 Polished tile printing pattern detection method, system and device based on machine vision
CN109002777A (en) * 2018-06-29 2018-12-14 电子科技大学 A kind of infrared small target detection method towards complex scene
CN109002777B (en) * 2018-06-29 2021-03-30 电子科技大学 Infrared small target detection method for complex scene
CN109305534A (en) * 2018-10-25 2019-02-05 安徽理工大学 Coal wharf's belt conveyor self-adaptation control method based on computer vision
CN110188624A (en) * 2019-05-10 2019-08-30 国网福建省电力有限公司龙岩供电公司 A kind of substation's wind drift recognition methods and system based on deep learning

Also Published As

Publication number Publication date
CN107516315B (en) 2020-04-28

Similar Documents

Publication Publication Date Title
CN103996018B (en) Face identification method based on 4DLBP
CN102043950B (en) Vehicle outline recognition method based on canny operator and marginal point statistic
CN103366190B (en) A kind of method of identification traffic signss
CN103886330B (en) Sorting technique based on semi-supervised SVM integrated study
CN107516315A (en) A kind of development machine based on machine vision is slagged tap monitoring method
CN103048329B (en) A kind of road surface crack detection method based on active contour model
CN104408449B (en) Intelligent mobile terminal scene literal processing method
Liu et al. Recognition of the operational statuses of reagent addition using dynamic bubble size distribution in copper flotation process
CN109117703B (en) Hybrid cell type identification method based on fine-grained identification
CN105825233B (en) A kind of pedestrian detection method based on on-line study random fern classifier
CN110110780A (en) A kind of picture classification method based on confrontation neural network and magnanimity noise data
CN102024259A (en) Bacterial colony automatic detection method
CN109086772A (en) A kind of recognition methods and system distorting adhesion character picture validation code
CN103177266A (en) Intelligent stock pest identification system
CN108764251A (en) A kind of workpiece surface crack detection method and system
CN106447673A (en) Chip pin extraction method under non-uniform illumination condition
CN107886066A (en) A kind of pedestrian detection method based on improvement HOG SSLBP
CN106780376A (en) The background image dividing method of partitioning algorithm is detected and combined based on conspicuousness
CN105426924A (en) Scene classification method based on middle level features of images
CN105354547A (en) Pedestrian detection method in combination of texture and color features
CN107578414B (en) Method for processing pavement crack image
CN117330582A (en) Polymer PE film surface crystal point detecting system
CN107256545A (en) A kind of broken hole flaw detection method of large circle machine
Li et al. A multi-stage approach for automatic classification of environmental microorganisms
CN111723861B (en) Gangue percentage identification method for coal mine control baffle

Legal Events

Date Code Title Description
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