CN117079268A - Construction method and application method of three-dimensional data set of internal diseases of road - Google Patents
Construction method and application method of three-dimensional data set of internal diseases of road Download PDFInfo
- Publication number
- CN117079268A CN117079268A CN202311336935.2A CN202311336935A CN117079268A CN 117079268 A CN117079268 A CN 117079268A CN 202311336935 A CN202311336935 A CN 202311336935A CN 117079268 A CN117079268 A CN 117079268A
- Authority
- CN
- China
- Prior art keywords
- road
- image
- disease
- data set
- internal
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 201000010099 disease Diseases 0.000 title claims abstract description 221
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 title claims abstract description 221
- 238000000034 method Methods 0.000 title claims abstract description 79
- 238000010276 construction Methods 0.000 title claims abstract description 7
- 230000000149 penetrating effect Effects 0.000 claims abstract description 17
- 238000012545 processing Methods 0.000 claims abstract description 16
- 230000009467 reduction Effects 0.000 claims abstract description 13
- 239000011159 matrix material Substances 0.000 claims description 97
- 238000013527 convolutional neural network Methods 0.000 claims description 31
- 238000012549 training Methods 0.000 claims description 19
- 238000004364 calculation method Methods 0.000 claims description 18
- 238000012360 testing method Methods 0.000 claims description 18
- 230000006870 function Effects 0.000 claims description 15
- 238000002372 labelling Methods 0.000 claims description 15
- 238000012795 verification Methods 0.000 claims description 12
- 238000001514 detection method Methods 0.000 claims description 10
- 239000011800 void material Substances 0.000 claims description 10
- 230000007547 defect Effects 0.000 claims description 8
- 239000000463 material Substances 0.000 claims description 8
- 230000004913 activation Effects 0.000 claims description 6
- 210000002569 neuron Anatomy 0.000 claims description 6
- 208000037920 primary disease Diseases 0.000 claims description 6
- 238000005070 sampling Methods 0.000 claims description 5
- 238000006243 chemical reaction Methods 0.000 claims description 4
- 230000002708 enhancing effect Effects 0.000 claims description 4
- 241000287196 Asthenes Species 0.000 claims description 3
- 230000000694 effects Effects 0.000 claims description 3
- 238000003672 processing method Methods 0.000 claims description 3
- 230000009466 transformation Effects 0.000 claims description 3
- 238000011426 transformation method Methods 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 2
- 238000012423 maintenance Methods 0.000 description 8
- 230000008569 process Effects 0.000 description 6
- 230000009471 action Effects 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 238000009472 formulation Methods 0.000 description 2
- 239000000203 mixture Substances 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000013519 translation Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005553 drilling Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/64—Three-dimensional objects
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N22/00—Investigating or analysing materials by the use of microwaves or radio waves, i.e. electromagnetic waves with a wavelength of one millimetre or more
- G01N22/02—Investigating the presence of flaws
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
- G01S13/88—Radar or analogous systems specially adapted for specific applications
- G01S13/885—Radar or analogous systems specially adapted for specific applications for ground probing
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
- G01S13/88—Radar or analogous systems specially adapted for specific applications
- G01S13/89—Radar or analogous systems specially adapted for specific applications for mapping or imaging
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/28—Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns
-
- 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
- G06V10/30—Noise filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/42—Global feature extraction by analysis of the whole pattern, e.g. using frequency domain transformations or autocorrelation
- G06V10/422—Global feature extraction by analysis of the whole pattern, e.g. using frequency domain transformations or autocorrelation for representing the structure of the pattern or shape of an object therefor
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/56—Extraction of image or video features relating to colour
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/776—Validation; Performance evaluation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/176—Urban or other man-made structures
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Abstract
A construction method and a use method of a three-dimensional data set for internal diseases of a road belong to the technical field of identification of the internal diseases of the road. In order to solve the problem of improving the accuracy of road disease identification, the invention adopts the ground penetrating radar to collect the road internal longitudinal section image and the plane image corresponding to the road internal longitudinal section image, and the collected road internal longitudinal section image is subjected to noise reduction treatment according to the noise reduction treatment method of the road internal image; constructing a sub-data set 1 by adopting the longitudinal section image in the road after the noise reduction treatment, and constructing a sub-data set 2 by adopting the plane image corresponding to the longitudinal section image in the road; carrying out disease identification on the built road internal images in the sub-data set 1 and the sub-data set 2 by utilizing the disease identification method of the road internal image in the step S2; and carrying out file processing on the obtained road internal images in the sub-data set 1 and the sub-data set 2 after disease identification to obtain a constructed road internal disease three-dimensional data set. The invention improves the accuracy of identifying road diseases.
Description
Technical Field
The invention belongs to the technical field of road interior disease identification, and particularly relates to a method for constructing a three-dimensional data set of road interior disease and a use method thereof.
Background
The road is the basis for safe driving of the vehicle. In recent years, due to the influence of urban road underground pipe network leakage, rain wash and peripheral construction disturbance factors, materials are loose in the road, gradually evolve into void, and further, the events such as road collapse are caused, and the road collapse event is in an increasing trend year by year, so that trip safety of people is seriously threatened, and the important problem of social concern is formed.
The traditional drilling coring method is used for carrying out fixed-point sampling detection on road structure defects, belongs to a lossy detection method, can damage an underground structure and has the defects of high detection cost, long time consumption, low efficiency and the like. Meanwhile, because the fixed-point sampling has certain randomness and cannot contain a complete underground structure, the detection result cannot accurately reflect the whole disease condition.
The ground penetrating radar replaces the traditional lossy detection method by virtue of the characteristics of high efficiency, no damage, convenience and accuracy. The ground penetrating radar emits high-frequency electromagnetic waves to the underground, then receives echo signals to generate GPR images, and judges the specific condition of the detected target according to the image result, so that efficient nondestructive detection of the underground target is realized. The ground penetrating radar can generate a large amount of data when detecting road structure diseases, however, the identification of the disease data at present depends on the interpretation of radar images manually, and the ground penetrating radar has the advantages of high cost, long time consumption and low efficiency. In addition, due to the perceived difference, the interpretation standard is difficult to determine by a manual interpretation mode, and errors are inevitably caused to influence the judgment of the detection result. Therefore, the recognition speed of diseases in the road is improved in an intelligent recognition mode, and the influence of artificial subjective factors is eliminated.
The invention relates to a method and a system for enhancing the B-scan image characteristics of a ground penetrating radar based on deep learning, which are applied to the patent of 202111616020.8 and are named as the invention.
The invention patent with the application number of 202210905240.0 and the invention name of a clutter suppression method and system for a ground penetrating radar B-scan image constructs a training data set by acquiring a clutter-containing GPR B-scan image, a clutter-free GPR B-scan image and a clutter-containing background GPR B-scan image of an underground target area.
In the data set constructed by the method, the information of other sections in the road collected by the three-dimensional ground penetrating radar is ignored only according to the information of the longitudinal sections collected by the radar, namely the B-scan image. Meanwhile, when the data set is marked, the phase change of electromagnetic waves emitted by the radar is not considered, and the disease is marked only by hyperbolic form and gray information, so that the marking result in the data set is deviated from the real result due to the influence of artificial subjective factors.
Disclosure of Invention
The invention aims to solve the problem of improving the accuracy of road disease identification, and provides a method for constructing a three-dimensional data set of road internal diseases and a use method thereof.
In order to achieve the above purpose, the present invention is realized by the following technical scheme:
a method for constructing a three-dimensional data set of diseases in a road comprises the following steps:
s1, constructing a noise reduction processing method of an image in a road;
s2, constructing a disease identification method of an image in the road;
s3, acquiring road internal longitudinal section images and plane images corresponding to the road internal longitudinal section images by adopting a ground penetrating radar, and performing noise reduction treatment on the acquired road internal longitudinal section images according to a noise reduction treatment method of the road internal images in the step S1;
s4, constructing a sub-data set 1 by adopting the noise-reduced road internal longitudinal section image obtained in the step S3, and constructing a sub-data set 2 by adopting the plane image corresponding to the road internal longitudinal section image acquired in the step S3;
s5, carrying out disease identification on the road internal images in the sub-data set 1 and the sub-data set 2 constructed in the step S4 by using the disease identification method of the road internal image in the step S2;
s6, carrying out file processing on the road internal images in the sub-data set 1 and the sub-data set 2 obtained in the step S5 after disease identification to obtain a constructed road internal disease three-dimensional data set;
the specific implementation method of the step S6 comprises the following steps:
S6.1, marking the images in the sub-data set 1 after disease identification obtained in the step S5 by using LabelImg software, marking the diseases of the images in the sub-data set 1 after disease identification by using rectangular frames and marking the disease types, and storing the name of the marking file and the name of the image in the sub-data set 1 after disease identificationSuch that the information of the annotation file comprisesWherein->For marking the point coordinates of the upper left corner of the rectangular box of the image in the disease-identified sub-dataset 1,/->For marking the length of the rectangular frame of the image in the disease-identified sub-dataset 1 in the x-axis direction,/v>Obtaining a data set 1 of a three-dimensional data set of the internal road diseases for marking the length of a rectangular frame of the image in the sub data set 1 after disease identification along the y-axis direction;
s6.2, using LabelImg software to the sub-data set 2 after disease identification obtained in the step S5, marking the disease of the image in the sub-data set 2 after disease identification by using a rectangular frame, marking the disease type, and storing the naming of the marking file to be consistent with the naming of the image in the sub-data set 2 after disease identification, wherein the information of the marking file comprisesWherein->For marking the point coordinates of the upper left corner of the rectangular box of the image in the disease-identified sub-dataset 2,/-, for example >For marking the length of the rectangular frame of the image in the disease-identified sub-dataset 2 in the x-axis direction,/v>And (3) obtaining the data set 2 of the three-dimensional data set of the road internal diseases for marking the length of the rectangular frame of the image in the sub data set 2 after the disease identification along the y-axis direction.
Further, the specific implementation method of the step S1 includes the following steps:
s1.1, acquiring an internal image of a road by adopting a ground penetrating radar, and constructing acquired internal image data of the road into an internal image data matrix of the roadA，
；
Wherein,first of the acquired road interior imageiThe time domain data of the individual waves,mis the total number of channels;
；
wherein,first of the acquired road interior imageiFirst part of the wave>Amplitude data, n is the total number;
s1.2, constructing a background noise frequency matrix of an image in a road;
s1.2.1, establishing a rectangular coordinate system, wherein the vertex of the upper left corner of an internal image of a road is taken as a coordinate origin, the horizontal right is taken as an x-axis positive direction, the vertical downward is taken as a y-axis positive direction, the length of 1 pixel is defined as a unit length, the internal image size of the road is set as a x b, a is the image length along the x-axis direction, and b is the image length along the y-axis direction;
s1.2.2 manually selecting the position of a road area without diseases, randomly selecting a road area without diseases with the size of 10 x b, setting a waveform time domain data matrix of the road area without diseases as R, and comprising waveform time domain data of 10 single-channel waves, wherein the expression is as follows:
；
；
Wherein,is the firstiWaveform time domain data of single-channel wave, +.>Is the firstiIn a single wavegThe amplitude data of the plurality of amplitude data,Ttranspose the matrix;
s1.2.3, converting the single-channel waveform time domain data into single-channel waveform frequency domain data by adopting a Fourier transform mode, wherein the calculation expression is as follows:
；
wherein,is the firstiWaveform frequency domain data of single-channel wave, +.>Is imaginary unit, ++>For frequency +.>Time is; obtaining waveform frequency domain data matrix of road area without disease>；
S1.2.4 calculating frequency according to the waveform time domain data of the single-channel wave in the step S1.2.2, and constructing a background noise frequency matrixWherein->Is the firstiSingle channel waveThe frequency calculated from the waveform time domain data of (c),mis the total number of channels;
s1.3, converting the time domain data of the road internal image acquired in the step S1.1 into frequency domain data, wherein the conversion expression is as follows:
；
wherein,first of the acquired road interior imageiFrequency domain data of individual waves to obtain a waveform frequency domain data matrix of the acquired road internal image>；
Then calculating frequency according to the waveform time domain data of the road internal image acquired in the step S1.1, and constructing a frequency matrix of the road internal imageWherein->First of the acquired road interior image iFrequency calculated by the waveform time domain data;
traversing by taking background noise frequency matrix C as comparison objectLAll elements in (1), deleteLThe same elements as C in the list and then deleteFCorresponding elements in the road interior image are traversed to obtain a waveform frequency domain data matrix of the road interior imageWhereinGpIs the firstpWaveform frequency domain data of the internal images of the road after the traversal;
s1.4, converting frequency domain data in a waveform frequency domain data matrix of the traversed road internal image obtained in the step S1.3 into time domain data, and adopting a calculation expression of Fourier transformation to obtain the following formula:
；
wherein,is the firstiThe time domain data corresponding to the waveform frequency domain data of the road internal image after the traversal is obtained, and the time domain matrix H corresponding to the waveform frequency domain data matrix of the road internal image after the traversal is:
；
；
wherein,is the firstiThe +.f. of the time domain data corresponding to the waveform frequency domain data of the road internal image after the traversal>Amplitude data;
s1.5, reconstructing the road internal image by using a time domain matrix corresponding to the waveform frequency domain data matrix of the traversed road internal image obtained in the step S1.4;
s1.5.1 first, a time domain matrix corresponding to the waveform frequency domain data matrix of the road internal image after traversal HIn (a) and (b)Traversing to obtain maximum->And minimum->；
S1.5.2, pair ofHThe amplitude data in the gray scale image is subjected to standardization processing and mapped into a gray scale value range of 0-255, and the calculation expression is as follows:
；
wherein,is->Normalizing the processed amplitude data;
s1.5.3, constructing a single-channel matrix based on the amplitude data normalized in the step S1.5.2The expression isThen constructing all channel matrixes based on single channel matrixes；
S1.5.4 the entire channel matrix obtained in step S1.5.3KThe element value in the two-dimensional image is used as a gray value, the x-axis direction of the two-dimensional image is used as the motion direction of radar sampling data, the y-axis direction of the two-dimensional image is used as the road depth direction, and a gray map M is drawn to obtain a reconstructed road internal image.
Further, the specific implementation method for setting the gain coefficient in the step S2 includes the following steps:
s2.1, dividing the reconstructed road internal image obtained in the step S1 along the vertical direction of the image, wherein the dividing number is set asNhEach divided image has a height ofHhWidth of；
S2.2, setting the image matrix of the ith divided image asThen the average value +.>The expression is:
；
S2.3, constructing a mean matrix based on the average value of the image matrices of the divided images obtained in the step S2.2Then calculate +.>And->Difference matrix->The expression is:
；
s2.4, carrying out absolute value operation on all elements of the difference matrix obtained in the step S2.3 to construct an absolute value matrix of the difference matrixThe expression is:
；
then calculate the standard deviation of the absolute value matrix of the difference matrixThe expression is:
；
s2.5, constructing a standard deviation matrix according to the standard deviation of the absolute value matrix of the difference matrix obtained in the step S2.4；
S2.6, setting critical standard deviation asStandard deviation of standard deviation matrix +.>And critical standard deviation->Comparison is made when->At the time, the gain factor is set to +.>，/>The value of (2) is determined according to the actual geological condition, the road material type, the radar parameter and the disease type parameter; when->Setting the gain coefficient to be 1;
s2.7, constructing a standard deviation matrix according to the gain coefficient calculated in the step S2.6Corresponding gain coefficient matrix->The method is used for enhancing the characteristics of the echo image of the position of the disease area in the road to obtain the road internal image of the primary disease identification.
Further, the specific implementation method for extracting the disease target area in the step S2 includes the following steps:
S2.9, performing binarization processing on the road internal image identified by the primary disease obtained in the step S2.8 by adopting a maximum inter-class variance method;
s2.10, counting the total number of pixels in all the connected areas of the road internal image identified by the preliminary disease after the binarization processing in the step S2.9, deleting the connected areas with the total number of pixels smaller than 500, reserving the connected areas with the total number of pixels larger than or equal to 500, and numbering the connected areas of the road internal image identified by the preliminary disease after the pixel deleting processing as N1, N2 … Ni … Nc;
s2.11, establishing a coordinate system for the road internal image identified by the preliminary disease after the pixel deleting treatment in the step S2.10, taking the top point of the left upper corner of the image as a coordinate origin, taking the horizontal right as the positive direction of the x axis, taking the vertical downward as the positive direction of the y axis, extracting coordinate points of pixels in all connected areas of the road internal image identified by the preliminary disease after the pixel deleting treatment, and setting the coordinate of the leftmost pixel point of the connected area Ni as the coordinate of the left-most pixel pointThe coordinate of the rightmost pixel point is +.>The coordinate of the topmost pixel point is +.>The coordinate expression is defined as:
；
；
；
wherein,is the leftmost pixel point of the connected region NixAxis coordinates->Is the leftmost pixel point of the connected region Ni yAn axis coordinate;
s2.12, adopting a method for calculating dip angle to screenSelecting regions with hyperbolic characteristics, and respectively obtaining inclination angles from top pixel points to leftmost pixel points and rightmost pixel pointsAg 1 AndAg 2 the expression is:
；
；
then extracting the maximum value of the inclination angles from the top pixel point to the leftmost pixel point and the rightmost pixel point, and marking the maximum value asThe expression is:
；
setting critical angleExtracting +.>Is a hyperbolic region +.>；
S2.13, extracting hyperbolic areas of all the connected areas, and renumbering to beWherein->Is the total number of hyperbolic areas;
s2.14, sequentially extracting single-channel waveform time domain data of all hyperbolic areas passing through the vertex coordinates based on the vertex coordinates of all hyperbolas obtained in the step S2.13, and sequentially recording as:，
wherein,，/>is the firstiSingle-channel waveform time domain data of single hyperbolic area passing vertex coordinates,/and method for generating the same>Is the firstiSingle-channel waveform time domain data of single hyperbolic region over vertex coordinateslAmplitude data;
s2.15, converting single-channel waveform time domain data of the hyperbolic region over-vertex coordinates obtained in the step S2.14 into frequency domain data by adopting a wavelet transformation method, wherein the calculation expression is as follows:
；
；
Wherein,is the firstiSingle-channel waveform frequency domain data of peak coordinate crossing of hyperbola region, < ->For the scale of +>For translation amount->Is a basic wavelet;
then calculate the firstiPhase of single-track waveform of crossing vertex coordinates of hyperbolic regionThe computational expression is:
；
wherein,is->Imaginary part of->Is->The real part of (2);
s2.16, setting the electromagnetic wave at the transmitting position as based on the electromagnetic wave theoryElectromagnetic wave at the reflection position is +.>Constructing a direction function->The expression is:
；
calculating a direction function whenWhen the electromagnetic wave is transmitted from the high dielectric constant to the low dielectric constant medium, the dielectric constant of the material at the disease position is smaller than that at the emission position, and the first judgment is madeiThe hyperbolic areas are void areas in the disease road;
s2.17, calculating the phase of a single-channel waveform of the over-vertex coordinates of all hyperbolic areas, and judging the road internal void condition of all the hyperbolic areas to obtain image data of the void area in the damaged road.
Further, the images in the sub-data set 1 and the sub-data set 2 constructed in the step S4 are numbered 1,2, and 3 … Nu in sequence, wherein Nu is the total number of the images in the sub-data set 1 and the sub-data set 2.
The application method of the three-dimensional data set of the road internal diseases is realized by means of the construction method of the three-dimensional data set of the road internal diseases, and comprises the following steps:
Step one, randomly dividing the data set 1 of the three-dimensional data set of the road internal diseases obtained in the step 6 into a training set, a verification set and a test set of the data set 1 of the three-dimensional data set of the road internal diseases;
randomly dividing the data set 2 of the three-dimensional data set of the road internal diseases obtained in the step S6 into a training set, a verification set and a test set of the data set 2 of the three-dimensional data set of the road internal diseases;
inputting the training set, the verification set and the test set of the data set 1 of the three-dimensional data set of the road internal diseases obtained in the step one into a convolutional neural network for training, verifying and testing, and outputting model parameters of the convolutional neural network model 1, including the number of network layers, the number of neuron nodes of each layer, the learning rate, the weight, the bias, the activation function, the loss function and the convolutional kernel, so as to obtain the convolutional neural network model 1;
inputting a training set, a verification set and a test set of a data set 2 of the three-dimensional data set of the road internal diseases obtained in the step one into a convolutional neural network for training, verifying and testing, and outputting model parameters of the convolutional neural network model 2, including the number of network layers, the number of neuron nodes of each layer, the learning rate, the weight, the bias, the activation function, the loss function and the convolutional kernel, so as to obtain the convolutional neural network model 2;
Step three, detecting a longitudinal section view of the road internal disease image by adopting the convolutional neural network model 1 obtained in the step two; detecting a plan view of an image of the internal diseases of the road by adopting the convolutional neural network model 2 obtained in the second step;
setting the output of the convolutional neural network model 1 asWhen a disease is detected, record +.>1 is shown in the specification; when no disease is detected, record +.>Is 0;
setting the output of the convolutional neural network model 2 asWhen a disease is detected, record +.>1 is shown in the specification; when no disease is detected, record +.>Is 0;
then set the final output asThe calculation formula is as follows:
；
step four, obtaining the final output of the step threeMake a judgment when->When the value is 0, judging that no disease exists; when->If 1, it is further determined whether or not the disease area identified by the longitudinal section view of the road interior disease image and the plan view of the road interior disease image coincide.
Further, the specific implementation method of the fourth step comprises the following steps:
step four, extracting the rectangular frame of the longitudinal section map of the road internal disease image obtained in the step three, wherein the information of the labeling file of the longitudinal section map of the road internal disease image is obtained as follows；
Step four, extracting a rectangular frame of the plan view of the road interior defect image obtained in the step three, wherein the information of the labeling file of the plan view of the road interior defect image is obtained as follows ；
And step four, comparing and calculating the position overlapping area through the information of the labeling file of the longitudinal section image and the information of the labeling file of the plane image of the road internal disease image obtained in the step four and the step four, wherein the calculation formula is as follows:
；
wherein,a start position which is a position overlapping region;
；
wherein,the end position of the position overlapping area;
；
wherein,an overlap ratio that is a position overlap region;
the standard value of (2) is determined according to the labeling effect of the diseases.
The invention has the beneficial effects that:
according to the method for constructing the three-dimensional data set of the road internal diseases, firstly, noise reduction of the road internal ground penetrating radar image is carried out, the quality of pixels is improved, and a more effective image foundation is provided for disease identification; then, the phase information of radar single-channel wave data is obtained through a time-frequency domain conversion technology, and on the basis of original hyperbolic shape and brightness analysis, the analysis of electromagnetic wave phase information is integrated, so that the accuracy of disease identification and classification is improved; then, a ground penetrating radar image data set is established, the data set comprises data subsets of longitudinal section images and plane images of the ground penetrating radar, and through association of disease information in the two types of images and combination of a convolutional neural network, a use method of the data set is provided, and accuracy of intelligent identification of diseases is improved.
According to the method for constructing the three-dimensional data set of the road internal diseases, provided by the invention, the detection personnel can accurately and intelligently identify the road internal diseases. On the basis, the method can be used for assisting in guiding the evaluation of the internal conditions of the road and the formulation of the road maintenance decision scheme, so that the maintenance resource is optimized, the maintenance cost is reduced, and the maintenance efficiency is improved.
The method for constructing the three-dimensional data set of the internal diseases of the road can be used for constructing a high-quality data set, and can enable different research institutions, road management departments and experts in related fields to share data through data sharing so as to promote cooperation and communication. All the parties develop research work of classifying, evaluating and monitoring the diseases together by researching the characteristics and rules of the road diseases together, so that the development of road disease identification technology is accelerated, and the level of the whole industry is improved.
Drawings
Fig. 1 is a flowchart of a method for constructing a three-dimensional data set of an internal disease of a road according to the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be further described in detail below with reference to the accompanying drawings and detailed description. It should be understood that the embodiments described herein are for purposes of illustration only and are not intended to limit the invention, i.e., the embodiments described are merely some, but not all, of the embodiments of the invention. The components of the embodiments of the present invention generally described and illustrated in the figures herein can be arranged and designed in a wide variety of different configurations, and the present invention can have other embodiments as well.
Thus, the following detailed description of the embodiments of the invention, as presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, are intended to fall within the scope of the present invention.
For further understanding of the invention, the following detailed description is presented in conjunction with the accompanying drawings 1 to provide a further understanding of the invention in its aspects, features and efficacy:
the first embodiment is as follows:
a method for constructing a three-dimensional data set of diseases in a road comprises the following steps:
s1, constructing a noise reduction processing method of an image in a road;
further, the specific implementation method of the step S1 includes the following steps:
s1.1, acquiring an internal image of a road by adopting a ground penetrating radar, and constructing acquired internal image data of the road into an internal image data matrix of the roadA，
；
Wherein,first of the acquired road interior imageiThe time domain data of the individual waves,mis the total number of channels;
；
wherein,first of the acquired road interior imageiFirst part of the wave>Amplitude data, n is the total number;
S1.2, constructing a background noise frequency matrix of an image in a road;
s1.2.1, establishing a rectangular coordinate system, wherein the vertex of the upper left corner of an internal image of a road is taken as a coordinate origin, the horizontal right is taken as an x-axis positive direction, the vertical downward is taken as a y-axis positive direction, the length of 1 pixel is defined as a unit length, the internal image size of the road is set as a x b, a is the image length along the x-axis direction, and b is the image length along the y-axis direction;
s1.2.2 manually selecting the position of a road area without diseases, randomly selecting a road area without diseases with the size of 10 x b, setting a waveform time domain data matrix of the road area without diseases as R, and comprising waveform time domain data of 10 single-channel waves, wherein the expression is as follows:
；
；
wherein,is the firstiWaveform time domain data of single-channel wave, +.>Is the firstiIn a single wavegThe amplitude data of the plurality of amplitude data,Ttranspose the matrix;
s1.2.3, converting the single-channel waveform time domain data into single-channel waveform frequency domain data by adopting a Fourier transform mode, wherein the calculation expression is as follows:
；
wherein,is the firstiWaveform frequency domain data of single-channel wave, +.>Is imaginary unit, ++>For frequency +.>Time is; obtaining waveform frequency domain data matrix of road area without disease >；
S1.2.4 calculating frequency according to the waveform time domain data of the single-channel wave in the step S1.2.2, and constructing a background noise frequency matrixWherein->Is the firstiThe frequency calculated from the waveform time domain data of the single channel wave,mis the total number of channels;
s1.3, converting the time domain data of the road internal image acquired in the step S1.1 into frequency domain data, wherein the conversion expression is as follows:
；
wherein,first of the acquired road interior imageiFrequency domain data of individual waves to obtain a waveform frequency domain data matrix of the acquired road internal image>；
Then calculating frequency according to the waveform time domain data of the road internal image acquired in the step S1.1, and constructing a frequency matrix of the road internal imageWherein->First of the acquired road interior imageiFrequency calculated by the waveform time domain data;
traversing by taking background noise frequency matrix C as comparison objectLAll elements in (1), deleteLThe same elements as C in the list and then deleteFCorresponding elements in the road interior image are traversed to obtain a waveform frequency domain data matrix of the road interior imageWhereinGpIs the firstpWaveform frequency domain data of the internal images of the road after the traversal;
traversing by taking background noise frequency matrix C as comparison objectLAll elements in (1), deleteLThe same elements as C in the list and then delete FCorresponding elements in the road interior image are traversed to obtain a waveform frequency domain data matrix of the road interior imageWhereinGpIs the firstpWaveform frequency domain data of the internal images of the road after the traversal;
s1.4, converting frequency domain data in a waveform frequency domain data matrix of the traversed road internal image obtained in the step S1.3 into time domain data, and adopting a calculation expression of Fourier transformation to obtain the following formula:
；
wherein,is the firstiTime domain data corresponding to the waveform frequency domain data of the road internal images after the traversal are obtained, and a waveform frequency domain data matrix of the road internal images after the traversal is obtainedThe corresponding time domain matrix H is:
；
；
wherein,is the firstiThe +.f. of the time domain data corresponding to the waveform frequency domain data of the road internal image after the traversal>Amplitude data;
s1.5, reconstructing the road internal image by using a time domain matrix corresponding to the waveform frequency domain data matrix of the traversed road internal image obtained in the step S1.4;
s1.5.1 first, a time domain matrix corresponding to the waveform frequency domain data matrix of the road internal image after traversalHIn (a) and (b)Traversing to obtain maximum->And minimum->；
S1.5.2, pair ofHThe amplitude data in the gray scale image is subjected to standardization processing and mapped into a gray scale value range of 0-255, and the calculation expression is as follows:
；
Wherein,is->Normalizing the processed amplitude data; />
S1.5.3, constructing a single-channel matrix based on the amplitude data normalized in the step S1.5.2The expression isThen constructing all channel matrixes based on single channel matrixes；
S1.5.4 the entire channel matrix obtained in step S1.5.3KThe element value in the two-dimensional image is used as a gray value, the x-axis direction of the two-dimensional image is used as the motion direction of radar sampling data, the y-axis direction of the two-dimensional image is used as the road depth direction, and a gray map M is drawn to obtain a reconstructed road internal image;
s2, constructing a disease identification method of an image in the road;
further, the specific implementation method for setting the gain coefficient in the step S2 includes the following steps:
s2.1, dividing the reconstructed road internal image obtained in the step S1 along the vertical direction of the image, wherein the dividing number is set asNhEach divided image has a height ofHhWidth of；
S2.2, setting the image matrix of the ith divided image asThen the average value +.>The expression is:
；
s2.3, obtained on the basis of step S2.2Average value of image matrix of divided image to construct average value matrixThen calculate +.>And->Difference matrix- >The expression is:
；
s2.4, carrying out absolute value operation on all elements of the difference matrix obtained in the step S2.3 to construct an absolute value matrix of the difference matrixThe expression is:
；
then calculate the standard deviation of the absolute value matrix of the difference matrixThe expression is:
；
s2.5, constructing a standard deviation matrix according to the standard deviation of the absolute value matrix of the difference matrix obtained in the step S2.4；
S2.6, setting critical standard deviation asStandard deviation of standard deviation matrix +.>And critical standard deviation->Comparison is made when->At the time, the gain factor is set to +.>，/>The value of (2) is determined according to the actual geological condition, the road material type, the radar parameter and the disease type parameter; when->Setting the gain coefficient to be 1;
s2.7, constructing a standard deviation matrix according to the gain coefficient calculated in the step S2.6Corresponding gain coefficient matrix->The method is used for enhancing the characteristics of the echo image of the position of the disease area in the road to obtain an image of the road in which the primary disease is identified;
further, the specific implementation method for extracting the disease target area in the step S2 includes the following steps:
s2.9, performing binarization processing on the road internal image identified by the primary disease obtained in the step S2.8 by adopting a maximum inter-class variance method;
S2.10, counting the total number of pixels in all the connected areas of the road internal image identified by the preliminary disease after the binarization processing in the step S2.9, deleting the connected areas with the total number of pixels smaller than 500, reserving the connected areas with the total number of pixels larger than or equal to 500, and numbering the connected areas of the road internal image identified by the preliminary disease after the pixel deleting processing as N1, N2 … Ni … Nc;
s2.11, establishing a coordinate system for the road internal image identified by the preliminary disease after the pixel deleting treatment in the step S2.10, taking the top point of the left upper corner of the image as a coordinate origin, taking the horizontal right as the positive direction of the x axis, taking the vertical downward as the positive direction of the y axis, extracting coordinate points of pixels in all connected areas of the road internal image identified by the preliminary disease after the pixel deleting treatment, and setting the coordinate of the leftmost pixel point of the connected area Ni as the coordinate of the left-most pixel pointThe coordinate of the rightmost pixel point is +.>The coordinate of the topmost pixel point is +.>The coordinate expression is defined as:
；
；
；
wherein,is the leftmost pixel point of the connected region NixAxis coordinates->Is the leftmost pixel point of the connected region NiyAn axis coordinate;
s2.12, screening out a region with hyperbolic characteristics by adopting a method for calculating the inclination angle, and respectively obtaining the inclination angles from the top pixel point to the leftmost pixel point and the rightmost pixel point Ag 1 AndAg 2 the expression is:
；
；
then extracting the maximum value of the inclination angles from the top pixel point to the leftmost pixel point and the rightmost pixel point, and marking the maximum value asThe expression is:
；
setting critical angleExtracting +.>Is a hyperbolic region +.>；
S2.13, extracting hyperbolic areas of all the connected areas, and renumbering to beWherein->Is the total number of hyperbolic areas;
s2.14, sequentially extracting single-channel waveform time domain data of all hyperbolic areas passing through the vertex coordinates based on the vertex coordinates of all hyperbolas obtained in the step S2.13, and sequentially recording as:，
wherein,，/>is the firstiSingle-channel waveform time domain data of single hyperbolic area passing vertex coordinates,/and method for generating the same>Is the firstiSingle-channel waveform time domain data of single hyperbolic region over vertex coordinateslAmplitude data;
s2.15, converting single-channel waveform time domain data of the hyperbolic region over-vertex coordinates obtained in the step S2.14 into frequency domain data by adopting a wavelet transformation method, wherein the calculation expression is as follows:
；
；
wherein,is the firstiSingle-channel waveform frequency domain data of peak coordinate crossing of hyperbola region, < ->For the scale of +>For translation amount->Is a basic wavelet;
then calculate the firstiPhase of single-track waveform of crossing vertex coordinates of hyperbolic region The computational expression is:
；
wherein,is->Imaginary part of->Is->The real part of (2);
s2.16, setting the electromagnetic wave at the transmitting position as based on the electromagnetic wave theoryElectromagnetic wave at the reflection position is +.>Constructing a direction function->The expression is:
；
calculating a direction function whenWhen the electromagnetic wave is transmitted from the high dielectric constant to the low dielectric constant medium, the dielectric constant of the material at the disease position is smaller than that at the emission position, and the first judgment is madeiThe hyperbolic areas are void areas in the disease road;
when (when)When the electromagnetic wave is transmitted from a low dielectric constant to a high dielectric constant medium, the dielectric constant of the material at the disease position is larger than that of the emission position;
s2.17, calculating the phase of a single-channel waveform of the vertex passing coordinates of all hyperbolic areas, and judging the internal void condition of the road of all the hyperbolic areas to obtain image data of the void area inside the damaged road;
s3, acquiring road internal longitudinal section images and plane images corresponding to the road internal longitudinal section images by adopting a ground penetrating radar, and performing noise reduction treatment on the acquired road internal longitudinal section images according to a noise reduction treatment method of the road internal images in the step S1;
s4, constructing a sub-data set 1 by adopting the noise-reduced road internal longitudinal section image obtained in the step S3, and constructing a sub-data set 2 by adopting the plane image corresponding to the road internal longitudinal section image acquired in the step S3;
Further, the images in the sub-data set 1 and the sub-data set 2 constructed in the step S4 are numbered 1,2 and 3 … Nu in sequence, wherein Nu is the total number of the images in the sub-data set 1 and the sub-data set 2;
s5, carrying out disease identification on the road internal images in the sub-data set 1 and the sub-data set 2 constructed in the step S4 by using the disease identification method of the road internal image in the step S2;
s6, carrying out file processing on the road internal images in the sub-data set 1 and the sub-data set 2 obtained in the step S5 after disease identification to obtain a constructed road internal disease three-dimensional data set
Further, the specific implementation method of the step S6 includes the following steps:
s6.1, marking the images in the sub-data set 1 after disease identification obtained in the step S5 by using LabelImg software, marking the diseases of the images in the sub-data set 1 after disease identification by using rectangular frames and marking the disease types, and storing the names of marking files to be consistent with the names of the images in the sub-data set 1 after disease identification, wherein the information of the marking files comprisesWherein->For marking the point coordinates of the upper left corner of the rectangular box of the image in the disease-identified sub-dataset 1,/->Rectangular boxes marking images in the disease-identified sub-dataset 1 along the x-axis direction Length of->Obtaining a data set 1 of a three-dimensional data set of the internal road diseases for marking the length of a rectangular frame of the image in the sub data set 1 after disease identification along the y-axis direction;
s6.2, using LabelImg software to the sub-data set 2 after disease identification obtained in the step S5, marking the disease of the image in the sub-data set 2 after disease identification by using a rectangular frame, marking the disease type, and storing the naming of the marking file to be consistent with the naming of the image in the sub-data set 2 after disease identification, wherein the information of the marking file comprisesWherein->For marking the point coordinates of the upper left corner of the rectangular box of the image in the disease-identified sub-dataset 2,/-, for example>For marking the length of the rectangular frame of the image in the disease-identified sub-dataset 2 in the x-axis direction,/v>And (3) obtaining the data set 2 of the three-dimensional data set of the road internal diseases for marking the length of the rectangular frame of the image in the sub data set 2 after the disease identification along the y-axis direction.
The method for constructing the three-dimensional data set of the road internal diseases is realized, the road internal disease data set established based on the radar collected road image in the prior art is improved, only the longitudinal section information of the road internal structure is considered, the effective information of other dimensions is ignored, and the data set is difficult to accurately distinguish different types of diseases, so that the disease identification accuracy is low;
The method for constructing the three-dimensional data set of the road internal diseases is realized, the influence of artificial subjective factors in the data marking process is improved, the marking result in the data set is deviated from the real result because the electromagnetic wave propagation characteristics, such as phases, are not considered in the image marking process in the prior art, and the change rule of the propagation process of different media is not considered.
The method for constructing the three-dimensional data set of the road internal diseases is realized, and detection personnel can use the data set to realize accurate and intelligent identification of the road internal diseases. On the basis, the method can be used for assisting in guiding the evaluation of the internal conditions of the road and the formulation of a road maintenance decision scheme, so that the maintenance resource is optimized, the maintenance cost is reduced, and the maintenance efficiency is improved.
The second embodiment is as follows:
the application method of the three-dimensional data set of the road internal diseases is realized by the construction method of the three-dimensional data set of the road internal diseases according to the first specific embodiment, and comprises the following steps:
step one, randomly dividing the data set 1 of the three-dimensional data set of the road internal diseases obtained in the step 6 into a training set, a verification set and a test set of the data set 1 of the three-dimensional data set of the road internal diseases;
Randomly dividing the data set 2 of the three-dimensional data set of the road internal diseases obtained in the step S6 into a training set, a verification set and a test set of the data set 2 of the three-dimensional data set of the road internal diseases;
inputting the training set, the verification set and the test set of the data set 1 of the three-dimensional data set of the road internal diseases obtained in the step one into a convolutional neural network for training, verifying and testing, and outputting model parameters of the convolutional neural network model 1, including the number of network layers, the number of neuron nodes of each layer, the learning rate, the weight, the bias, the activation function, the loss function and the convolutional kernel, so as to obtain the convolutional neural network model 1;
inputting a training set, a verification set and a test set of a data set 2 of the three-dimensional data set of the road internal diseases obtained in the step one into a convolutional neural network for training, verifying and testing, and outputting model parameters of the convolutional neural network model 2, including the number of network layers, the number of neuron nodes of each layer, the learning rate, the weight, the bias, the activation function, the loss function and the convolutional kernel, so as to obtain the convolutional neural network model 2;
step three, detecting a longitudinal section view of the road internal disease image by adopting the convolutional neural network model 1 obtained in the step two; detecting a plan view of an image of the internal diseases of the road by adopting the convolutional neural network model 2 obtained in the second step;
Setting the output of the convolutional neural network model 1 asWhen a disease is detected, record +.>1 is shown in the specification; when no disease is detected, record +.>Is 0;
setting the output of the convolutional neural network model 2 asWhen a disease is detected, record +.>1 is shown in the specification; when no disease is detected, record +.>Is 0;
then set the final output asThe calculation formula is as follows:
；
step four, obtaining the final output of the step threeMake a judgment when->When the value is 0, judging that no disease exists; when->If 1, it is further determined whether or not the disease area identified by the longitudinal section view of the road interior disease image and the plan view of the road interior disease image coincide.
Further, the specific implementation method of the fourth step comprises the following steps:
step four, extracting the rectangular frame of the longitudinal section map of the road internal disease image obtained in the step three, wherein the information of the labeling file of the longitudinal section map of the road internal disease image is obtained as follows；
Step four, extracting a rectangular frame of the plan view of the road interior defect image obtained in the step three, wherein the information of the labeling file of the plan view of the road interior defect image is obtained as follows；
And step four, comparing and calculating the position overlapping area through the information of the labeling file of the longitudinal section image and the information of the labeling file of the plane image of the road internal disease image obtained in the step four and the step four, wherein the calculation formula is as follows:
；
Wherein,a start position which is a position overlapping region;
；
wherein,the end position of the position overlapping area;
；
wherein,an overlap ratio that is a position overlap region;
the standard value of (2) is determined according to the labeling effect of the diseases.
It is noted that relational terms such as "first" and "second", and the like, are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
Although the application has been described above with reference to specific embodiments, various modifications may be made and equivalents may be substituted for elements thereof without departing from the scope of the application. In particular, the features of the disclosed embodiments may be combined with each other in any manner so long as there is no structural conflict, and the exhaustive description of these combinations is not given in this specification solely for the sake of brevity and resource saving. Therefore, it is intended that the application not be limited to the particular embodiments disclosed herein, but that the application will include all embodiments falling within the scope of the appended claims.
Claims (7)
1. The method for constructing the three-dimensional data set of the internal diseases of the road is characterized by comprising the following steps:
s1, constructing a noise reduction processing method of an image in a road;
s2, constructing a disease identification method of an image in the road;
s3, acquiring road internal longitudinal section images and plane images corresponding to the road internal longitudinal section images by adopting a ground penetrating radar, and performing noise reduction treatment on the acquired road internal longitudinal section images according to a noise reduction treatment method of the road internal images in the step S1;
s4, constructing a sub-data set 1 by adopting the noise-reduced road internal longitudinal section image obtained in the step S3, and constructing a sub-data set 2 by adopting the plane image corresponding to the road internal longitudinal section image acquired in the step S3;
s5, carrying out disease identification on the road internal images in the sub-data set 1 and the sub-data set 2 constructed in the step S4 by using the disease identification method of the road internal image in the step S2;
s6, carrying out file processing on the road internal images in the sub-data set 1 and the sub-data set 2 obtained in the step S5 after disease identification to obtain a constructed road internal disease three-dimensional data set;
the specific implementation method of the step S6 comprises the following steps:
s6.1, marking the images in the sub-data set 1 after disease identification obtained in the step S5 by using LabelImg software, marking the diseases of the images in the sub-data set 1 after disease identification by using rectangular frames and marking the disease types, and storing the names of marking files to be consistent with the names of the images in the sub-data set 1 after disease identification, wherein the information of the marking files comprises Wherein->For marking the point coordinates of the upper left corner of the rectangular box of the image in the disease-identified sub-dataset 1,/->For marking the length of the rectangular frame of the image in the disease-identified sub-dataset 1 in the x-axis direction,/v>Obtaining a data set 1 of a three-dimensional data set of the internal road diseases for marking the length of a rectangular frame of the image in the sub data set 1 after disease identification along the y-axis direction;
s6.2, using LabelImg software to the sub-data set 2 after disease identification obtained in the step S5, marking the disease of the image in the sub-data set 2 after disease identification by using a rectangular frame, marking the disease type, and storing the naming of the marking file to be consistent with the naming of the image in the sub-data set 2 after disease identification, wherein the information of the marking file comprisesWherein->For marking the point coordinates of the upper left corner of the rectangular box of the image in the disease-identified sub-dataset 2,/-, for example>For marking the length of the rectangular frame of the image in the disease-identified sub-dataset 2 in the x-axis direction,/v>And (3) obtaining the data set 2 of the three-dimensional data set of the road internal diseases for marking the length of the rectangular frame of the image in the sub data set 2 after the disease identification along the y-axis direction.
2. The method for constructing a three-dimensional dataset for internal diseases of roads according to claim 1, wherein the specific implementation method of step S1 comprises the steps of:
S1.1, acquiring an internal image of a road by adopting a ground penetrating radar, and constructing acquired internal image data of the road into an internal image data matrix of the roadA，
；
Wherein,first of the acquired road interior imageiThe time domain data of the individual waves,mis the total number of channels;
；
wherein,first of the acquired road interior imageiFirst part of the wave>Amplitude data, n is the total number;
s1.2, constructing a background noise frequency matrix of an image in a road;
s1.2.1, establishing a rectangular coordinate system, wherein the vertex of the upper left corner of an internal image of a road is taken as a coordinate origin, the horizontal right is taken as an x-axis positive direction, the vertical downward is taken as a y-axis positive direction, the length of 1 pixel is defined as a unit length, the internal image size of the road is set as a x b, a is the image length along the x-axis direction, and b is the image length along the y-axis direction;
s1.2.2 manually selecting the position of a road area without diseases, randomly selecting a road area without diseases with the size of 10 x b, setting a waveform time domain data matrix of the road area without diseases as R, and comprising waveform time domain data of 10 single-channel waves, wherein the expression is as follows:
；
；
wherein,is the firstiWaveform time domain data of single-channel wave, +.>Is the firstiIn a single wavegThe amplitude data of the plurality of amplitude data, TTranspose the matrix;
s1.2.3, converting the single-channel waveform time domain data into single-channel waveform frequency domain data by adopting a Fourier transform mode, wherein the calculation expression is as follows:
；
wherein,is the firstiWaveform frequency domain data of single-channel wave, +.>Is imaginary unit, ++>For frequency +.>Time is; obtaining waveform frequency domain data matrix of road area without disease>；
S1.2.4 calculating frequency according to the waveform time domain data of the single-channel wave in the step S1.2.2, and constructing a background noise frequency matrixWherein->Is the firstiThe frequency calculated from the waveform time domain data of the single channel wave,mis the total number of channels;
s1.3, converting the time domain data of the road internal image acquired in the step S1.1 into frequency domain data, wherein the conversion expression is as follows:
；
wherein,first of the acquired road interior imageiFrequency domain data of individual waves to obtain a waveform frequency domain data matrix of the acquired road internal image>；
Then calculating frequency according to the waveform time domain data of the road internal image acquired in the step S1.1, and constructing a frequency matrix of the road internal imageWherein->First of the acquired road interior imageiFrequency calculated by the waveform time domain data;
traversing by taking background noise frequency matrix C as comparison object LAll elements in (1), deleteLThe same elements as C in the list and then deleteFCorresponding elements in the road interior image are traversed to obtain a waveform frequency domain data matrix of the road interior imageWhereinGpIs the firstpWaveform frequency domain data of the internal images of the road after the traversal;
s1.4, converting frequency domain data in a waveform frequency domain data matrix of the traversed road internal image obtained in the step S1.3 into time domain data, and adopting a calculation expression of Fourier transformation to obtain the following formula:
；
wherein,is the firstiThe time domain data corresponding to the waveform frequency domain data of the road internal image after the traversal is obtained, and the time domain matrix H corresponding to the waveform frequency domain data matrix of the road internal image after the traversal is:
；
；
wherein,is the firstiThe +.f. of the time domain data corresponding to the waveform frequency domain data of the road internal image after the traversal>Amplitude data;
s1.5, reconstructing the road internal image by using a time domain matrix corresponding to the waveform frequency domain data matrix of the traversed road internal image obtained in the step S1.4;
s1.5.1 first, a time domain matrix corresponding to the waveform frequency domain data matrix of the road internal image after traversalHIn (a) and (b)Traversing to obtain maximum->And minimum->；
S1.5.2, pair ofHThe amplitude data in the gray scale image is subjected to standardization processing and mapped into a gray scale value range of 0-255, and the calculation expression is as follows:
；
Wherein,is->Normalizing the processed amplitude data;
s1.5.3, constructing a single-channel matrix based on the amplitude data normalized in the step S1.5.2The expression isThen constructing all channel matrixes based on single channel matrixes；
S1.5.4 the entire channel matrix obtained in step S1.5.3KThe element value in the two-dimensional image is used as a gray value, the x-axis direction of the two-dimensional image is used as the motion direction of radar sampling data, the y-axis direction of the two-dimensional image is used as the road depth direction, and a gray map M is drawn to obtain a reconstructed road internal image.
3. The method for constructing a three-dimensional dataset for internal diseases of roads according to claim 2, wherein the specific implementation method for setting the gain factor in step S2 comprises the steps of:
s2.1, dividing the reconstructed road internal image obtained in the step S1 along the vertical direction of the image, wherein the dividing number is set asNhEach division ofThe height of the image isHhWidth of；
S2.2, setting the image matrix of the ith divided image asThen the average value +.>The expression is:
；
s2.3, constructing a mean matrix based on the average value of the image matrices of the divided images obtained in the step S2.2Then calculate +. >And->Difference matrix->The expression is:
；
s2.4, carrying out absolute value operation on all elements of the difference matrix obtained in the step S2.3 to construct an absolute value matrix of the difference matrixThe expression is:
；
then calculate the standard deviation of the absolute value matrix of the difference matrixThe expression is:
；
s2.5, constructing a standard deviation matrix according to the standard deviation of the absolute value matrix of the difference matrix obtained in the step S2.4；
S2.6, setting critical standard deviation asStandard deviation of standard deviation matrix +.>And critical standard deviation->Comparison is made when->At the time, the gain factor is set to +.>，/>The value of (2) is determined according to the actual geological condition, the road material type, the radar parameter and the disease type parameter; when->Setting the gain coefficient to be 1;
s2.7 according toS2.6, constructing a standard deviation matrix by calculating gain coefficientsCorresponding gain coefficient matrixThe method is used for enhancing the characteristics of the echo image of the position of the disease area in the road to obtain the road internal image of the primary disease identification.
4. The method for constructing a three-dimensional dataset for internal diseases of roads according to claim 3, wherein the specific implementation method for extracting the disease target area in step S2 comprises the following steps:
S2.9, performing binarization processing on the road internal image identified by the primary disease obtained in the step S2.8 by adopting a maximum inter-class variance method;
s2.10, counting the total number of pixels in all the connected areas of the road internal image identified by the preliminary disease after the binarization processing in the step S2.9, deleting the connected areas with the total number of pixels smaller than 500, reserving the connected areas with the total number of pixels larger than or equal to 500, and numbering the connected areas of the road internal image identified by the preliminary disease after the pixel deleting processing as N1, N2 … Ni … Nc;
s2.11, establishing a coordinate system for the road internal image identified by the preliminary disease after the pixel deleting treatment in the step S2.10, taking the top point of the left upper corner of the image as a coordinate origin, taking the horizontal right as the positive direction of the x axis, taking the vertical downward as the positive direction of the y axis, extracting coordinate points of pixels in all connected areas of the road internal image identified by the preliminary disease after the pixel deleting treatment, and setting the coordinate of the leftmost pixel point of the connected area Ni as the coordinate of the left-most pixel pointThe coordinate of the rightmost pixel point is +.>The coordinate of the topmost pixel point is +.>The coordinate expression is defined as:
；
；
；
wherein,is the leftmost pixel point of the connected region NixAxis coordinates->Is the leftmost pixel point of the connected region Ni yAn axis coordinate;
s2.12, screening out a region with hyperbolic characteristics by adopting a method for calculating the inclination angle, and respectively obtaining the inclination angles from the top pixel point to the leftmost pixel point and the rightmost pixel pointAg 1 AndAg 2 the expression is:
；
；
then extracting the maximum value of the inclination angles from the top pixel point to the leftmost pixel point and the rightmost pixel point, and marking the maximum value asThe expression is:
；
setting critical angleExtracting +.>Is a hyperbolic region +.>；
S2.13, extracting hyperbolic areas of all the connected areas, and renumbering to beWherein->Is the total number of hyperbolic areas;
s2.14, sequentially extracting single-channel waveform time domain data of all hyperbolic areas passing through the vertex coordinates based on the vertex coordinates of all hyperbolas obtained in the step S2.13, and sequentially recording as:，
wherein,，/>is the firstiSingle-channel waveform time domain data of single hyperbolic area passing vertex coordinates,/and method for generating the same>Is the firstiSingle-channel waveform time domain data of single hyperbolic region over vertex coordinateslAmplitude data;
s2.15, converting single-channel waveform time domain data of the hyperbolic region over-vertex coordinates obtained in the step S2.14 into frequency domain data by adopting a wavelet transformation method, wherein the calculation expression is as follows:
；
；
Wherein,is the firstiSingle-channel waveform frequency domain data of peak coordinate crossing of hyperbola region, < ->For the scale of +>In order to be able to translate the quantity,is a basic wavelet;
then calculate the firstiPhase of single-track waveform of crossing vertex coordinates of hyperbolic regionThe computational expression is:
；
wherein,is->Imaginary part of->Is->The real part of (2);
s2.16, setting the electromagnetic wave at the transmitting position as based on the electromagnetic wave theoryElectromagnetic wave at the reflection position is +.>Constructing a direction function->The expression is:
；
calculating a direction function whenWhen the electromagnetic wave is transmitted from the high dielectric constant to the low dielectric constant medium, the dielectric constant of the material at the disease position is smaller than that at the emission position, and the first judgment is madeiThe hyperbolic areas are void areas in the disease road;
s2.17, calculating the phase of a single-channel waveform of the over-vertex coordinates of all hyperbolic areas, and judging the road internal void condition of all the hyperbolic areas to obtain image data of the void area in the damaged road.
5. The method for constructing three-dimensional data sets of internal diseases of roads according to claim 4, wherein the images in the sub-data set 1 and the sub-data set 2 constructed in the step S4 are numbered 1,2,3 and … Nu, respectively, and Nu is the total number of the images in the sub-data set 1 and the sub-data set 2.
6. A method for using a three-dimensional data set of internal diseases of a road, which is realized by the three-dimensional data set construction method of internal diseases of a road according to one of claims 1 to 5, and is characterized by comprising the following steps:
step one, randomly dividing the data set 1 of the three-dimensional data set of the road internal diseases obtained in the step 6 into a training set, a verification set and a test set of the data set 1 of the three-dimensional data set of the road internal diseases;
randomly dividing the data set 2 of the three-dimensional data set of the road internal diseases obtained in the step S6 into a training set, a verification set and a test set of the data set 2 of the three-dimensional data set of the road internal diseases;
inputting the training set, the verification set and the test set of the data set 1 of the three-dimensional data set of the road internal diseases obtained in the step one into a convolutional neural network for training, verifying and testing, and outputting model parameters of the convolutional neural network model 1, including the number of network layers, the number of neuron nodes of each layer, the learning rate, the weight, the bias, the activation function, the loss function and the convolutional kernel, so as to obtain the convolutional neural network model 1;
inputting a training set, a verification set and a test set of a data set 2 of the three-dimensional data set of the road internal diseases obtained in the step one into a convolutional neural network for training, verifying and testing, and outputting model parameters of the convolutional neural network model 2, including the number of network layers, the number of neuron nodes of each layer, the learning rate, the weight, the bias, the activation function, the loss function and the convolutional kernel, so as to obtain the convolutional neural network model 2;
Step three, detecting a longitudinal section view of the road internal disease image by adopting the convolutional neural network model 1 obtained in the step two; detecting a plan view of an image of the internal diseases of the road by adopting the convolutional neural network model 2 obtained in the second step;
setting the output of the convolutional neural network model 1 asWhen a disease is detected, record +.>1 is shown in the specification; no detection ofIn case of disease, record->Is 0;
setting the output of the convolutional neural network model 2 asWhen a disease is detected, record +.>1 is shown in the specification; when no disease is detected, record +.>Is 0;
then set the final output asThe calculation formula is as follows:
；
step four, obtaining the final output of the step threeMake a judgment when->When the value is 0, judging that no disease exists; when->If 1, it is further determined whether or not the disease area identified by the longitudinal section view of the road interior disease image and the plan view of the road interior disease image coincide.
7. The method of using a three-dimensional dataset for internal diseases of roads according to claim 6, wherein the specific implementation method of step four comprises the steps of:
step (a)41. Extracting a rectangular frame of a longitudinal section view of the road internal disease image obtained in the third step, and obtaining the information of a labeling file of the longitudinal section view of the road internal disease image as follows ；
Step four, extracting a rectangular frame of the plan view of the road interior defect image obtained in the step three, wherein the information of the labeling file of the plan view of the road interior defect image is obtained as follows；
And step four, comparing and calculating the position overlapping area through the information of the labeling file of the longitudinal section image and the information of the labeling file of the plane image of the road internal disease image obtained in the step four and the step four, wherein the calculation formula is as follows:
；
wherein,a start position which is a position overlapping region;
；
wherein,the end position of the position overlapping area;
；
wherein,overlap ratio as a positional overlap region；
The standard value of (2) is determined according to the labeling effect of the diseases.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311336935.2A CN117079268B (en) | 2023-10-17 | 2023-10-17 | Construction method and application method of three-dimensional data set of internal diseases of road |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311336935.2A CN117079268B (en) | 2023-10-17 | 2023-10-17 | Construction method and application method of three-dimensional data set of internal diseases of road |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117079268A true CN117079268A (en) | 2023-11-17 |
| CN117079268B CN117079268B (en) | 2023-12-26 |
Family
ID=88713754
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311336935.2A Active CN117079268B (en) | 2023-10-17 | 2023-10-17 | Construction method and application method of three-dimensional data set of internal diseases of road |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117079268B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110458028A (en) * | 2019-07-12 | 2019-11-15 | 北京交通大学 | A kind of tunnel-liner typical disease automatic identification method based on geological radar |
| CN112132946A (en) * | 2020-09-29 | 2020-12-25 | 深圳安德空间技术有限公司 | Data extraction and display method for three-dimensional ground penetrating radar |
| CN112232392A (en) * | 2020-09-29 | 2021-01-15 | 深圳安德空间技术有限公司 | Data interpretation and identification method for three-dimensional ground penetrating radar |
| CN113688692A (en) * | 2021-07-29 | 2021-11-23 | 浙江大学 | Water supply pipeline leakage detection method based on time-frequency scale characteristics of ground penetrating radar |
| CN114331890A (en) * | 2021-12-27 | 2022-04-12 | 中南大学 | Ground penetrating radar B-scan image feature enhancement method and system based on deep learning |
-
2023
- 2023-10-17 CN CN202311336935.2A patent/CN117079268B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110458028A (en) * | 2019-07-12 | 2019-11-15 | 北京交通大学 | A kind of tunnel-liner typical disease automatic identification method based on geological radar |
| CN112132946A (en) * | 2020-09-29 | 2020-12-25 | 深圳安德空间技术有限公司 | Data extraction and display method for three-dimensional ground penetrating radar |
| CN112232392A (en) * | 2020-09-29 | 2021-01-15 | 深圳安德空间技术有限公司 | Data interpretation and identification method for three-dimensional ground penetrating radar |
| CN113688692A (en) * | 2021-07-29 | 2021-11-23 | 浙江大学 | Water supply pipeline leakage detection method based on time-frequency scale characteristics of ground penetrating radar |
| CN114331890A (en) * | 2021-12-27 | 2022-04-12 | 中南大学 | Ground penetrating radar B-scan image feature enhancement method and system based on deep learning |
Non-Patent Citations (1)
| Title |
|---|
| 李曼如: "基于探地雷达和深度学习的道路地下病害智能检测研究", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》, no. 01, pages 1 - 113 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117079268B (en) | 2023-12-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhu et al. | Intelligent logging lithological interpretation with convolution neural networks | |
| Li et al. | Detection of concealed cracks from ground penetrating radar images based on deep learning algorithm | |
| Liu et al. | Detection and localization of rebar in concrete by deep learning using ground penetrating radar | |
| Dorafshan et al. | Deep learning models for bridge deck evaluation using impact echo | |
| Liu et al. | Novel YOLOv3 model with structure and hyperparameter optimization for detection of pavement concealed cracks in GPR images | |
| CN112462346A (en) | Ground penetrating radar roadbed defect target detection method based on convolutional neural network | |
| CN111476088B (en) | Asphalt pavement water damage identification model construction method, identification method and system | |
| CN114296075B (en) | Ground penetrating radar image artificial intelligence identification method and device | |
| Wang et al. | Automatic detection of asphalt pavement thickness: A method combining GPR images and improved Canny algorithm | |
| CN113009447B (en) | Road underground cavity detection and early warning method based on deep learning and ground penetrating radar | |
| CN105137498A (en) | Underground target detection and recognition system and method based on feature fusion | |
| CN111798411B (en) | Concrete inner steel bar intelligent positioning method based on ground penetrating radar and deep learning | |
| CN112989481B (en) | Method for processing stable visual image data of complex geological tunnel construction surrounding rock | |
| Barkataki et al. | Classification of soil types from GPR B scans using deep learning techniques | |
| Gui et al. | Object-based crack detection and attribute extraction from laser-scanning 3D profile data | |
| Zhang et al. | Experimental research on evaluation of soil water content using ground penetrating radar and wavelet packet-based energy analysis | |
| CN113468804B (en) | Underground pipeline identification method based on matrix bundles and deep neural network | |
| Zhao et al. | A Comprehensive Horizon‐Picking Method on Subbottom Profiles by Combining Envelope, Phase Attributes, and Texture Analysis | |
| CN107657615A (en) | High Resolution SAR image change detection method based on increment CAE | |
| CN117079268B (en) | Construction method and application method of three-dimensional data set of internal diseases of road | |
| Yue et al. | Road subsurface distress recognition method using multiattribute feature fusion with ground penetrating radar | |
| CN114926431A (en) | Shale fracture development mode identification method based on electric imaging logging image | |
| CN115542278A (en) | Ground penetrating radar original data-based pavement damage area judgment method | |
| CN117077448B (en) | Road void area evolution prediction method, electronic equipment and storage medium | |
| Liu et al. | Advances in automatic identification of road subsurface distress using ground penetrating radar: State of the art and future trends |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |