CN109299681A - A kind of coal and rock face crack automatic identifying method based on support vector machines - Google Patents

A kind of coal and rock face crack automatic identifying method based on support vector machines Download PDF

Info

Publication number
CN109299681A
CN109299681A CN201811062930.4A CN201811062930A CN109299681A CN 109299681 A CN109299681 A CN 109299681A CN 201811062930 A CN201811062930 A CN 201811062930A CN 109299681 A CN109299681 A CN 109299681A
Authority
CN
China
Prior art keywords
coal
crackle
image
face crack
rock face
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
Application number
CN201811062930.4A
Other languages
Chinese (zh)
Inventor
李成武
艾迪昊
王启飞
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.)
China University of Mining and Technology Beijing CUMTB
Original Assignee
China University of Mining and Technology Beijing CUMTB
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 China University of Mining and Technology Beijing CUMTB filed Critical China University of Mining and Technology Beijing CUMTB
Priority to CN201811062930.4A priority Critical patent/CN109299681A/en
Publication of CN109299681A publication Critical patent/CN109299681A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • 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
    • G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
    • G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Multimedia (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses a kind of coal petrography face crack automatic identifying method based on support vector machines, this method include that the training method of crackle classifier and application class device carry out automatic identifying method to crackle.Wherein the training method of classifier includes: image acquisition, pretreatment, segmentation, connection field mark, connected component extract and input classifier training.This method mainly proposes using High frequency filter, histogram equalization, hysteresis threshold mode and handles coal petrography surface image, proposes 8 kinds of features for distinguishing coal and rock face crack and non-crackle: the ratio between area, maximum gauge, the deviation of zone boundary to centre distance, the average distance of zone boundary to centre distance, circularity, transverse and short axle, fluffy degree, compactness.It may be implemented to carry out automatic identification and label to the crackle of coal petrography image automatically by this method.This method can be efficiently modified the recognition methods of existing coal and rock face crack, improve the accuracy rate and operability of its identification.

Description

A kind of coal and rock face crack automatic identifying method based on support vector machines
Technical field:
The present invention relates to a kind of coal and rock face crack automatic identifying method based on support vector machines, specifically from Simultaneously display is marked in dynamic identification image, video or the crackle on the coal and rock surface in industrial camera signal.
Background technique:
China is important in the world coal production and consumption big country, coal as important strategic object in Chinese society and Consequence is seized of in economic development, however, the situation of coal in China exploitation is very severe, current China's coal-mine is averaged Mining depth has reached 500m, causes in recovery process that various accidents cause a tremendous loss of lives frequent occurrence, wherein coal petrography Dynamic disaster accident is to lead to one of the accident of personnel and property injures and deaths most serious.And the premise that coal rock dynamic disaster accident occurs It is the destruction unstability of coal and rock, and the direct embodiment of coal and rock generation destruction unstability is the crackle on coal and rock surface.Therefore, to coal The identification of rock mass face crack can effectively prevent the generation of coal rock dynamic disaster accident, reduce the generation of coal mining accident.
Currently, fractal technology or simple Threshold segmentation usually used to the Identification of Cracks on coal and rock surface, however by In surface there are a large amount of Original Cracks and joint, real-time or accuracy rate etc. can not be met the requirements by using conventional methods Index.In recent years, with the rapid development of computer technology and Digital image technology, to the processing method and reality of image and video Huge promotion has all been obtained in when property.The sample graph destroyed herein by nearly 1000 coal and rocks are obtained in experimentation Picture proposes a kind of coal and rock face crack automatic identifying method based on support vector machines.
Summary of the invention:
In view of the deficiencies of the prior art, it is high, real-time that the technical problem to be solved in the present invention is to provide a kind of recognition accuracies Property strong, the coal and rock face crack automatic identifying method with versatility, after identifying crackle, and be marked and Display.
In order to realize that appeal purpose, specific technical solution are as follows:
The characteristics of the method for the present invention is first against coal and rock face crack proposes to use high-pass filtering and histogram equalization Coal and rock surface image is pre-processed.Since the crackle on coal and rock surface and " edge " of image are similar, frequency range belongs to The high-frequency region of image, therefore several filtering are being compared, the present invention selects high-pass filtering method；On the other hand, body surface Crackle is often the minimum of its grey scale pixel value, but coal and rock is more special, itself has been the minimum of grey scale pixel value, Therefore the contrast for causing the two is low, is unfavorable for subsequent identification, therefore the present invention uses histogram to the image after filtering Figure equalization operation.After pre-processing, present invention proposition is split coal and rock surface using hysteresis threshold dividing method. Later, division and the label of connected domain are carried out to the image after segmentation, according to the definition of 8 connection, from top to bottom, from left to right Direction carries out label (difference) from small to large to each connected domain.Feature is carried out to each individual connected region later Extract, the invention proposes 8 kinds distinguish coal and rock face crack and other non-crackles feature: area, circularity, maximum gauge, The ratio between elliptical long axis and short axle, the deviation of zone boundary to centre distance, zone boundary to centre distance average distance, fluffy Looseness, compactness.Support vector machines is entered data into later, model is trained, and obtains Identification of Cracks by adjusting parameter Model.The image data of test is finally carried out to pretreatment and dividing method after the same method, feature is carried out to it and is mentioned Classifier is inputted after taking, and is judged.In order to which the crackle identified is effectively marked and shown, present invention proposition makes The crackle identified is marked and is shown with the method expanded in Digital Image Processing.
Detailed description of the invention:
Fig. 1 is the flow chart that the present invention carries out sample acquisition and classifier training.
Fig. 2 is the overview flow chart that the present invention identifies coal and rock crackle.
Fig. 3 is the effect picture identified using the present invention to a secondary coal and rock face crack.
Table 1 is the classifying quality using the present invention 600 images of test.
Specific embodiment:
Technical solution of the present invention is illustrated below in conjunction with attached drawing 1 and attached drawing 2.
As shown in Figure 1, the present invention describes the sample data training one for how passing through coal petrography Failure Instability first It can be in the method for its face crack of automatic identification.This method mainly includes 6 steps, is respectively as follows: the acquisition of (1) image；(2) image Pretreatment；(3) image segmentation；(4) connection field mark；(5) connected component extracts；(6) input SVM classifier training.
The acquisition of the image can be the coal and rock surface image of test or scene, can also pass through industrial phase It is intercepted in the video of machine shooting, should not be very little for the quantity of image pattern, preferably at 1000 or so, while by institute There is image to be converted to gray level image.
The image preprocessing mainly includes 2 steps: (1) image High frequency filter；(2) image histogram equalizes. For image High frequency filter, a high frequency filter is designed first, and image is converted into frequency by fast Fourier variation later Image is multiplied by domain with filter in a frequency domain, and image is finally converted to spatial domain from frequency domain.After High frequency filter, The crackle and edge detail information of image save, and other noises are filtered out, but due to coal and rock surface and its Contrast between crackle is too low, can affect to operations such as subsequent segmentations, therefore the present invention is pre-processing this One stage used histogram equalization method, continued with to the image after filtering.Specific operating method is by original graph The grey level histogram of picture becomes being uniformly distributed in whole tonal ranges from some gray scale interval for comparing concentration.
The image partition method refers to the method by hysteresis threshold, carries out two-value to the image on coal and rock surface Change, by image other noises and region further exclude.Specifically, hysteresis threshold method needs to set 3 parameters, point Not are as follows: minimum gradation value, maximum gradation value, maximum length value.For all pixels value in image, if its gray value is less than Minimal gray value parameter, then immediately by " receiving ", and gray value is greater than the pixel of maximum gradation value parameter then immediately by " refusal ", Direct between minimum gradation value and maximum gradation value for grey scale pixel value, then according to maximum length value, this parameter is sentenced It is disconnected, if the length of the connection of these pixel values is less than maximum length value, receives these pixel values, otherwise refuse.The present invention By comparison different parameters to the impact effect of segmentation, for the crackle on coal and rock surface, minimum gradation value, maximum gray scale Value, maximum length value select 10,60,10 respectively.
The connection field mark is mainly the label that region serial number is carried out to the image after segmentation, the subsequent identification of aspect And operation.Specifically, the method for the present invention according to 8 connection definition, from top to bottom, direction from left to right, to each connection Region carries out label from small to large, Region-1, Region-2 ..., Region-n is denoted as respectively, wherein to different figures Picture, since the image after segmentation is different, the value of n is also different.In addition, needing in this step by artificial It selects, the UNICOM domain of connected domain and non-crackle agency that crackle represents is respectively labeled as ' 1 ' and ' -1 ' and is recorded.
The connected component is extracted primarily directed to pretreatment and still there are many non-slit region quilts after dividing It mistakenly saves, it is therefore desirable to further distinguish using the feature of coal body face crack.Specifically, this hair The characteristics of bright face crack according to coal and rock, proposes 8 kinds of features for distinguishing coal and rock face crack and other objects, respectively Are as follows: area, maximum gauge, the deviation of zone boundary to centre distance, the average distance of zone boundary to centre distance, circularity, The ratio between elliptical long axis and short axle, fluffy degree, compactness.Wherein area and maximum gauge are more intuitive, because crackle It is formed and needs certain length and area, this 2 parameters can be used as primary characteristic parameter.Other parameters are respectively defined as:
Deviation of the zone boundary to centre distance:
Average distance of the zone boundary to centre distance:
Circularity:
The ratio between elliptical long axis and short axle:
Fluffy degree:
Compactness:
The input SVM classifier training is primarily referred to as after the feature for having extracted each connected domain, by what is extracted Value is stored in two-dimensional array, wherein every a line represents the serial number in each connection region, each column represent the 8 of each connection region Kind feature, the sample marked according to previous step, the region for extracting region and 500 non-crackle of 500 crackles make For the training data of SVM.The present invention is used as test set as training set, 20% to the data of data set 80%, is defined as follows:
Dregion={ (x1, y1), (x2, y2) ..., (xm, ym), xi=(xi1；xi2；...；xid), y=(- 1,1) m= 1000, d=8
Dregion_train={ (x1, y1), (x2, y2) ..., (xm, ym), xi=(xi1；xi2；...；xid), y=(- 1,1) m =800, d=8
Dregion_test={ (x1, y1), (x2, y2) ..., (xm, ym), xi=(xi1；xi2；...；xid), y=(- 1,1) m =200, d=8
SVM classifier is entered data into, different parameter and SVM kernel function is adjusted, can finally obtain for coal petrography Body face crack carries out the classifier of identification classification.
As shown in Fig. 2, the overall procedure that the present invention identifies coal and rock crackle is described below.It is image is first several According to input, can be used for the input of video data, the selection of image preprocessing later, image segmentation and the foregoing description is same Sample parameter also carries out same connected component labeling and feature extraction after the same method, result is stored in two-dimensional array.
After the two-dimensional array for obtaining indicating image connectivity domain serial number and feature, before each serial number is sequentially input In trained crackle sorter model, if it is judged that current connection Regional Representative's crackle, then carry out subsequent operation, it is otherwise defeated Enter next connected region to continue to judge.
It is that the region of crackle will be whole according to the serial number in the region recorded for being determined after classifier classification A connected region carries out expansive working, completes label and the identification of crackle, while the deposit of the serial number in crackle connection region is another It is recorded in a document.
For a large amount of image and video data, also it is referred to the method and is repeatedly scanned with, it should be noted that recording The serial number of every frame image in every image or video facilitates the subsequent analysis for carrying out crackle.
Wherein, table 1 and Fig. 3 are respectively using the classifying quality of the present invention 600 images of test and the present invention to a secondary coal petrography The effect picture that body face crack is identified.
Table 1

Claims (5)

1. a kind of coal and rock face crack automatic identifying method based on support vector machines, it is characterised in that this method includes crackle The recognition methods of the training method of classifier and overall crackle.Wherein the training method of crackle classifier specifically includes that (1) image It obtains；
(2) image preprocessing；(3) image segmentation；(4) connection field mark；(5) connected component extracts；(6) svm classifier is inputted Device training.The recognition methods of overall crackle is primarily referred to as completing using trained crackle sorter model to image or view The automatic identification of the crackle on coal and rock surface in frequency.
2. a kind of coal and rock face crack automatic identifying method based on support vector machines, feature according to right 1 exist In for the image on coal and rock surface, preprocess method uses High frequency filter and histogram equalization mode.
3. a kind of coal and rock face crack automatic identifying method based on support vector machines, feature according to right 1 exist In in training crackle classifier, dividing method selects hysteresis threshold dividing method, and the most optimized parameter is illustrated.
4. a kind of coal and rock face crack automatic identifying method based on support vector machines, feature according to right 1 exist In, feature extraction for coal and rock face crack, the invention proposes 8 kinds to distinguish coal and rock face cracks and other objects Characteristic parameter, be respectively as follows: area, maximum gauge, the deviation of zone boundary to centre distance, zone boundary to centre distance it is flat The ratio between equal distance, circularity, elliptical long axis and short axle, fluffy degree, compactness.Wherein, in the description, to each parameter Calculating all gives corresponding formula and calculation specifications.
5. a kind of coal and rock face crack automatic identifying method based on support vector machines, feature according to right 1 exist In the good sorter model of application training, the present invention operates by expansion form to mark to it crackle identified Note, and its serial number is saved in document, facilitate subsequent analysis.
CN201811062930.4A 2018-09-12 2018-09-12 A kind of coal and rock face crack automatic identifying method based on support vector machines Pending CN109299681A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201811062930.4A CN109299681A (en) 2018-09-12 2018-09-12 A kind of coal and rock face crack automatic identifying method based on support vector machines

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201811062930.4A CN109299681A (en) 2018-09-12 2018-09-12 A kind of coal and rock face crack automatic identifying method based on support vector machines

Publications (1)

Publication Number Publication Date
CN109299681A true CN109299681A (en) 2019-02-01

Family

ID=65166675

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201811062930.4A Pending CN109299681A (en) 2018-09-12 2018-09-12 A kind of coal and rock face crack automatic identifying method based on support vector machines

Country Status (1)

Country Link
CN (1) CN109299681A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109934409A (en) * 2019-03-18 2019-06-25 西安科技大学 A kind of high working face coal wall working face wall caving prediction network and its prediction technique
CN111179243A (en) * 2019-12-25 2020-05-19 武汉昕竺科技服务有限公司 Small-size chip crack detection method and system based on computer vision
CN111414658A (en) * 2020-03-17 2020-07-14 宜春学院 Rock mass mechanics parameter inverse analysis method

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2013020142A2 (en) * 2011-08-04 2013-02-07 University Of Southern California Image-based crack detection
CN103207999A (en) * 2012-11-07 2013-07-17 中国矿业大学(北京) Method and system for coal and rock boundary dividing based on coal and rock image feature extraction and classification and recognition
CN107369136A (en) * 2017-06-22 2017-11-21 福建省万龙新材料科技有限公司 Composite polycrystal-diamond face crack visible detection method

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2013020142A2 (en) * 2011-08-04 2013-02-07 University Of Southern California Image-based crack detection
CN103207999A (en) * 2012-11-07 2013-07-17 中国矿业大学(北京) Method and system for coal and rock boundary dividing based on coal and rock image feature extraction and classification and recognition
CN107369136A (en) * 2017-06-22 2017-11-21 福建省万龙新材料科技有限公司 Composite polycrystal-diamond face crack visible detection method

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
韩锟等: "基于区域级和像素级特征的路面裂缝检测方法", 《铁道科学与工程学报》 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109934409A (en) * 2019-03-18 2019-06-25 西安科技大学 A kind of high working face coal wall working face wall caving prediction network and its prediction technique
CN111179243A (en) * 2019-12-25 2020-05-19 武汉昕竺科技服务有限公司 Small-size chip crack detection method and system based on computer vision
CN111414658A (en) * 2020-03-17 2020-07-14 宜春学院 Rock mass mechanics parameter inverse analysis method
CN111414658B (en) * 2020-03-17 2023-06-30 宜春学院 Rock mass mechanical parameter inverse analysis method

Similar Documents

Publication Publication Date Title
CN109961049B (en) Cigarette brand identification method under complex scene
Gllavata et al. Text detection in images based on unsupervised classification of high-frequency wavelet coefficients
CN104751142B (en) A kind of natural scene Method for text detection based on stroke feature
CN101453575B (en) Video subtitle information extracting method
CN102081731B (en) Method and device for extracting text from image
CN105205488B (en) Word area detection method based on Harris angle points and stroke width
CN105069447B (en) A kind of recognition methods of human face expression
CN103729856B (en) A kind of Fabric Defects Inspection detection method utilizing S-transformation signal extraction
CN107122777A (en) A kind of vehicle analysis system and analysis method based on video file
CN109299681A (en) A kind of coal and rock face crack automatic identifying method based on support vector machines
CN104050471A (en) Natural scene character detection method and system
CN110060273B (en) Remote sensing image landslide mapping method based on deep neural network
CN104408449A (en) Intelligent mobile terminal scene character processing method
CN101551853A (en) Human ear detection method under complex static color background
CN105718552A (en) Clothing freehand sketch based clothing image retrieval method
CN111027544B (en) MSER license plate positioning method and system based on visual saliency detection
CN104463134A (en) License plate detection method and system
CN109190456A (en) Pedestrian detection method is overlooked based on the multiple features fusion of converging channels feature and gray level co-occurrence matrixes
CN102136074B (en) Man-machine interface (MMI) based wood image texture analyzing and identifying method
CN116452506A (en) Underground gangue intelligent visual identification and separation method based on machine learning
CN110335233A (en) Express-way guard-rail plates defect detecting system and method based on image processing techniques
CN105354547A (en) Pedestrian detection method in combination of texture and color features
CN112085723B (en) Automatic detection method for spring jumping fault of truck bolster
Mehrara et al. Novel edge detection using BP neural network based on threshold binarization
Zhao et al. An effective binarization method for disturbed camera-captured document 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: 20190201