CN114565793B - Road traffic crack monitoring method and system - Google Patents
Road traffic crack monitoring method and system Download PDFInfo
- Publication number
- CN114565793B CN114565793B CN202210192638.4A CN202210192638A CN114565793B CN 114565793 B CN114565793 B CN 114565793B CN 202210192638 A CN202210192638 A CN 202210192638A CN 114565793 B CN114565793 B CN 114565793B
- Authority
- CN
- China
- Prior art keywords
- road
- crack
- monitoring
- monitoring model
- training
- 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.)
- Active
Links
- 238000012544 monitoring process Methods 0.000 title claims abstract description 146
- 238000000034 method Methods 0.000 title claims abstract description 47
- 238000012549 training Methods 0.000 claims description 71
- 238000013528 artificial neural network Methods 0.000 claims description 70
- 230000002159 abnormal effect Effects 0.000 claims description 19
- 230000006870 function Effects 0.000 claims description 13
- 238000012706 support-vector machine Methods 0.000 claims description 8
- 238000006073 displacement reaction Methods 0.000 claims description 5
- 230000008569 process Effects 0.000 claims description 5
- 238000004891 communication Methods 0.000 claims description 3
- 238000012806 monitoring device Methods 0.000 claims description 2
- 238000001514 detection method Methods 0.000 abstract description 10
- 239000000463 material Substances 0.000 description 6
- 238000010801 machine learning Methods 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 238000002790 cross-validation Methods 0.000 description 2
- 238000000354 decomposition reaction Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 230000008439 repair process Effects 0.000 description 2
- 238000001228 spectrum Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 239000002699 waste material Substances 0.000 description 2
- 239000010426 asphalt Substances 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 239000004035 construction material Substances 0.000 description 1
- 230000008602 contraction Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000017525 heat dissipation Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—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
- 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
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
- G06F18/24155—Bayesian classification
-
- 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
- G06N3/045—Combinations of networks
-
- 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
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
- G08B21/18—Status alarms
-
- 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
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Molecular Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Probability & Statistics with Applications (AREA)
- Business, Economics & Management (AREA)
- Emergency Management (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a road traffic crack monitoring method and a system, wherein the characteristic that vibration parameters at a road crack are different from those of a normal road is utilized, firstly, vibration data of a target road are collected, and whether the target road is cracked or not is judged from the vibration data by using a first monitoring model; then collecting image data of a target road, and judging the specific type of the crack from the image data by using a second monitoring model; and finally, executing corresponding alarm operation according to the judged crack type. The invention is mainly divided into two parts, wherein the first part utilizes the vibration data of the target road to accurately judge whether the road is cracked, and the second part further judges the type of the crack on the basis of the first part, so that the accuracy of monitoring the road crack can be improved, and false detection caused by missing detection is avoided.
Description
Technical Field
The invention relates to the technical field of traffic infrastructure safety monitoring, in particular to a road traffic crack monitoring method and system.
Background
With the development of society, traffic infrastructure is rapidly developed, and large-scale road traffic networks are built in China. As an important part of modern transportation junction, the road bears the heavy duty of transportation and also relates to the safety of people involved in transportation.
The most common problem of roads is the occurrence of cracks, which lead to the attenuation of the resistance of the road structure and the potential safety hazard, which is mainly caused by the following reasons: 1. the subgrade is non-uniform in settlement; 2. when materials are stirred, chemical reaction occurs, heat is generated, the operation is improper in later construction, the internal temperature of the pavement is too high, the heat dissipation is slow, and along with the change of climate, the thermal expansion and cold contraction reaction easily occurs, so that cracks are caused; 3. cracks caused by too high pressure born by the pavement; 4. the service life of the road is too long, the material ages, and the hardness of the road surface is reduced. When a road is cracked, related maintenance personnel are required to be timely informed to repair the cracked road so as to ensure the road safety.
At present, whether a crack is generated on a road is mainly monitored by using a neural network, but in reality, due to various obstacles (such as vehicles, broken stones, garbage and the like) existing on the road, the input of the images containing the obstacles to the neural network easily causes the false detection of the neural network, for example, the road without the crack is judged as the crack, which leads to the waste of manpower and material resources.
Disclosure of Invention
The present invention aims to at least solve the technical problems existing in the prior art. Therefore, the invention provides the road traffic crack monitoring method and the road traffic crack monitoring system, which can improve the accuracy of road crack monitoring.
In a first aspect of the present invention, a road traffic crack monitoring method is provided, comprising the steps of:
obtaining vibration data of a target road;
inputting the vibration data into a first monitoring model to obtain a first monitoring result output by the first monitoring model;
if the first monitoring result shows that the target road has a crack, acquiring image data of the target road;
inputting the image data into a second monitoring model to obtain a second monitoring result output by the second monitoring model;
judging the crack type of the target road according to the second monitoring result, and generating an alarm signal corresponding to the crack type.
According to the first aspect of the invention, at least the following advantages are achieved:
the method utilizes the characteristic that vibration parameters of a road crack are different from those of a normal road, firstly, vibration data of a target road are collected, and whether the target road is cracked or not is judged from the vibration data by using a first monitoring model; then collecting image data of a target road, and judging the specific type of the crack from the image data by using a second monitoring model; and finally, executing corresponding alarm operation according to the judged crack type. The method is mainly divided into two parts, wherein the first part accurately judges whether the road is cracked or not by utilizing vibration data of the target road, and the second part further judges the type of the crack on the basis of the first part, so that the accuracy of monitoring the road crack can be improved, and false detection caused by missing detection is avoided.
According to some embodiments of the invention, before the inputting the vibration data into the first monitoring model, the method further comprises the steps of:
constructing a first monitoring model, wherein the first monitoring model comprises two different classes of classifiers; wherein the classifiers are both classifier;
selecting a first training sample, wherein the first training sample comprises a first classification sample of an abnormal road and a second classification sample of a normal road; the abnormal road is a road with a crack, the normal road is a road without a crack, the first classification sample and the second classification sample both comprise vibration data of the corresponding road, and the vibration data comprise vibration displacement, vibration amplitude and vibration frequency;
respectively inputting the first classification sample and the second classification sample into two classifiers for training; and taking classification results of the two classifiers as prediction results of the first monitoring model.
According to some embodiments of the invention, the two classifiers include a support vector machine classifier and a bayesian classifier;
the support vector machine classifier selects a Gaussian radial basis function as a kernel function;
the judging accuracy of the Bayesian classifier comprises the following steps:
wherein ,
the P (a) represents the prior probability of the first training sample, the P (y) i I a) represents the conditional probability of the first training sample, y i Representing the ith eigenvalue in training sample y, said a representing the type of training sample, said σ a，i Representing the variance of the ith feature in the class a training sample, said μ a，i Representing the mean of the ith feature in a class a training sample, said M representing the total number of features, said a 1 Representing the first classified sample, said a 2 Representing the second classification sample, the X a Representing a set of a class a training samples in the first training samples, wherein l represents the number of samples.
According to some embodiments of the invention, before the inputting the image data into the second monitoring model, the method further comprises the steps of:
constructing a second monitoring model, wherein the second monitoring model comprises a plurality of neural networks;
selecting a second training sample, wherein the second training sample comprises image data of an abnormal road; the abnormal road is a road with cracks;
and inputting a second training sample into the second monitoring model for training to obtain a prediction result of the second monitoring model.
According to some embodiments of the invention, the plurality of neural networks includes a VGG16 neural network and a U-Net neural network; the second monitoring model takes the U-Net neural network as a main body of the second monitoring model, replaces an encoder of the U-Net neural network with the VGG16 neural network, and initializes the weight of the VGG16 neural network through imageNet.
According to some embodiments of the invention, the step of inputting the second training samples into the second monitoring model for training includes:
inputting the image in the second training sample into a plurality of convolution layers in the U-Net neural network, wherein the plurality of convolution layers is 13 layers, and each layer is a convolution kernel of 3*3;
and upsampling the feature map of the convolution layer through the VGG16 neural network and respectively cascading the feature map with the feature map of the convolution layer.
According to some embodiments of the invention, the crack types include a lateral road crack, a longitudinal road crack, a block road crack, and a net road crack.
In a second aspect of the present invention, there is provided a road traffic crack monitoring system comprising:
a data acquisition unit for acquiring vibration data and image data of a target road;
the crack prediction unit is used for inputting the vibration data into a first monitoring model to obtain a first monitoring result output by the first monitoring model;
the crack classification unit is used for inputting the image data into a second monitoring model to obtain a second monitoring result output by the second monitoring model;
and the alarm unit is used for judging the crack type of the target road according to the second monitoring result and generating an alarm signal corresponding to the crack type.
According to the second aspect of the invention, at least the following advantages are achieved:
the system utilizes the characteristic that vibration parameters of a road crack can be different from those of a normal road, firstly, vibration data of a target road are collected, and whether the target road is cracked or not is judged from the vibration data by utilizing a first monitoring model; then collecting image data of a target road, and judging the specific type of the crack from the image data by using a second monitoring model; and finally, executing corresponding alarm operation according to the judged crack type. The system is mainly divided into two parts, the first part utilizes vibration data of a target road to accurately judge whether the road is cracked, and the second part further judges the type of the crack on the basis of the first part, so that the accuracy of monitoring the road crack can be improved, and false detection caused by missing detection is avoided.
In a third aspect of the present invention, there is provided a road traffic crack monitoring apparatus comprising: at least one control processor and a memory for communication connection with the at least one control processor; the memory stores instructions executable by the at least one control processor to enable the at least one control processor to perform the road traffic crack monitoring method described above.
In a fourth aspect of the present invention, a computer-readable storage medium stores computer-executable instructions for causing a computer to perform the road traffic crack monitoring method as described above.
Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention.
Drawings
The foregoing and/or additional aspects and advantages of the invention will become apparent and may be better understood from the following description of embodiments taken in conjunction with the accompanying drawings in which:
fig. 1 is a schematic structural diagram of a road traffic crack monitoring system according to an embodiment of the present invention;
fig. 2 is a flow chart of a road traffic crack monitoring method according to an embodiment of the present invention;
fig. 3 is a flow chart of a road traffic crack monitoring method according to another embodiment of the present invention;
fig. 4 is a flow chart of a road traffic crack monitoring method according to another embodiment of the present invention.
Detailed Description
Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to like or similar elements or elements having like or similar functions throughout. The embodiments described below by referring to the drawings are illustrative only and are not to be construed as limiting the invention.
In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "axial", "radial", "circumferential", etc. indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings are merely for convenience in describing the present invention and to simplify the description, and do not indicate or imply that the device or element being referred to must have a specific orientation, be configured and operated in a specific orientation, and therefore should not be construed as limiting the present invention. Furthermore, features defining "first", "second" may include one or more such features, either explicitly or implicitly. In the description of the present invention, unless otherwise indicated, the meaning of "a plurality" is two or more.
In the description of the present invention, it should be noted that, unless explicitly specified and limited otherwise, the terms "mounted," "connected," and "connected" are to be construed broadly, and may be either fixedly connected, detachably connected, or integrally connected, for example; can be mechanically or electrically connected; can be directly connected or indirectly connected through an intermediate medium, and can be communication between two elements. The specific meaning of the above terms in the present invention will be understood in specific cases by those of ordinary skill in the art.
At present, whether a crack is generated on a road is mainly monitored by using a neural network, but in reality, because various obstacles (such as vehicles, broken stones, garbage and the like) exist on the road, the image input to the neural network containing the obstacles easily causes false detection of the neural network, for example, the neural network judges that the crack exists on the road due to the obstacles existing in a normal road, and the neural network informs relevant manual investigation to go to the field, so that the waste of manpower and material resources is caused.
Referring to fig. 1, an embodiment of the present invention provides a road traffic crack monitoring system, including a data acquisition unit 1100, a crack prediction unit 1200, a crack classification unit 1300, and an alarm unit 1400, wherein:
the data acquisition unit 1100 is used to acquire vibration data and image data of a target road.
The crack prediction unit 1200 is configured to input vibration data to the first monitoring model, and obtain a first monitoring result output by the first monitoring model.
The crack classification unit 1300 is configured to input the image data to the second monitoring model, and obtain a second monitoring result output by the second monitoring model.
The alarm unit 1400 determines the type of the crack of the target road according to the second monitoring result, and generates an alarm signal corresponding to the type of the crack.
The system utilizes the characteristic that vibration parameters of an abnormal road crack are different from those of a normal road, firstly, vibration data of a target road are collected, and whether the target road is cracked or not is judged from the vibration data by utilizing a first monitoring model; then collecting image data of a target road, and judging the specific type of the crack from the image data by using a second monitoring model; and finally, executing corresponding alarm operation according to the judged crack type. The system is mainly divided into two parts, wherein the first part can preliminarily judge whether the target road has cracks by utilizing vibration data of the target road, and the second part is processed only when the target road has cracks; the second part further judges the type of the crack on the basis of the first part, and the system combines the vibration data and the image data of the road to judge twice, so that the accuracy of monitoring the crack of the road can be improved, and misjudgment is avoided.
Based on the above system embodiments, referring to fig. 2 to 4, a method embodiment of the present invention provides a road traffic crack monitoring method, which includes the following steps:
step S101, obtaining vibration data of a target road.
In this step, the target road refers to a road to be monitored. Road vibration data including vibration displacement, vibration amplitude, vibration frequency, and the like can be acquired by a sensor (e.g., a multi-axis gyroscope) provided at the road edge. Collecting vibration data generated when a vehicle passes through a plurality of collecting time points, collecting instantaneous values of the vibration data, judging whether cracks appear according to the change of the instantaneous values, namely performing Hilbert yellow transformation on the collected vibration data through MATLAB analysis software (empirical mode decomposition (Empirical mode decomposition) is performed on vibration data signals to obtain an intrinsic mode function (intrinsic mode function), performing Hilbert transformation on the intrinsic mode function, further obtaining Hilbert spectrums, time-frequency energy spectrums and the like of the signals), and then proving that the cracks appear on the road when the deviation between the Hilbert yellow transformed result of the vibration data and the result of normal road vibration data is larger than a threshold value. The threshold value and the vibration data of the normal road are statistically derived here, and are not particularly limited here because of the difference between different roads.
Because the vibration data generated before and after the occurrence of the road crack on the same road (for example, different road sections of a road) are different, whether the road is cracked or not can be judged according to the vibration data of the road, and compared with a monitoring mode of combining a neural network with an image, the accuracy of judging whether the crack is generated or not according to the vibration data of the road is higher. If the target road is judged to be cracked through the vibration data, the type of the crack can be further judged, so that maintenance personnel can process the crack, and traffic safety accidents are avoided. When the type of the crack is further judged, since vibration data of roads where different types of cracks occur are different, if the specific type of the crack is then judged by using the vibration data, an error is large, so that it is necessary to judge the specific type again by combining images with a neural network.
Step S103, vibration data are input into the first monitoring model, and a first monitoring result output by the first monitoring model is obtained.
Before step S103, the method further includes the steps of:
s1021, constructing a first monitoring model, wherein the first monitoring model comprises two different classes of classifiers; wherein, the classifier is a classifier.
Step S1022, selecting a first training sample, wherein the first training sample comprises a first classification sample of an abnormal road and a second classification sample of a normal road; the abnormal road is a road with a crack, the normal road is a road without a crack, the first classification sample and the second classification sample both comprise vibration data of the corresponding road, and the vibration data comprise vibration displacement, vibration amplitude and vibration frequency.
Step S1023, respectively inputting the first classification sample and the second classification sample into two classifiers for training, and taking the classification results of the two classifiers as the prediction results of the first monitoring model.
In step S1022, the first training samples include a first classification sample (i.e., positive sample) of the abnormal road and a second classification sample (i.e., negative sample) of the normal road. The first classification sample and the second classification sample are a plurality of monitoring points selected from the same construction road (avoiding the influence of construction materials of different roads on different vibration data), and corresponding sensors are installed on the monitoring points so as to obtain vibration data of a plurality of points (without cracks) on a normal road and vibration data of a plurality of points (crack points) on an abnormal road.
Because the uncertainty of the road (such as rolling the road by different vehicles), even if the same crack on the road or the same type and size of crack on the road, the vibration data collected by the collecting device (such as a multi-axis gyroscope) will not be different, if a single machine learning method is adopted, the reliability of the accuracy is lower, and the road crack personnel is safe, in order to further improve the accuracy of the discrimination, in step S1021 of the present embodiment, the first monitoring model includes two different types of classifiers, and in some embodiments, the two machine-learned classifiers are a support vector machine classifier and a bayesian classifier. The method is specifically as follows:
first, for a support vector machine classifier;
here, a gaussian radial basis function is selected as a kernel function of the classifier, a first training sample is input into the classifier, the classifier optimizes the superparameter using a grid search (accurate search of optimal superparameter) and a cross-validation method (cross-validation), and then the classifier is built using the searched superparameter. The support vector machine classifier firstly searches an optimal hyperplane, so that the distance between a first classified sample and a second classified sample in a first training sample and the optimal hyperplane is the largest, and classifies the first training sample.
Second, for bayesian classifiers;
the determination of the Bayesian classifier accurately comprises the following steps:
wherein ,
p (a) represents the prior probability of the first training sample, P (y) i I a) represents the conditional probability of the first training sample, y i Represents the ith eigenvalue in training sample y, a represents the type of training sample, σ a，i Representing the variance, μ of the ith feature in the class a training sample a，i Represents the mean of the ith feature in the class a training sample, M represents the total number of features, a 1 Representing a first classified sample, a 2 Representing a second classification sample, X a Representing the aggregate set of class a training samples in the first training sample, l representing the number of samples.
In this embodiment, if and only if the split results of both classifiers determine that a crack exists in the road, the final classification of the first monitoring model is that the crack exists in the road. The first monitoring model of the method uses two machine learning classifiers with different principles to judge whether the target road has cracks or not, and the judgment result is more accurate.
After the training of the first monitoring model is completed, the vibration data of the target road is input into the first monitoring model, and a first monitoring result output by the first monitoring model is obtained.
Step 105, if the first monitoring result shows that the target road has a crack, acquiring image data of the target road.
When the first monitoring model has accurately judged that the target road has a crack, the type of the crack needs to be further confirmed. In this embodiment, shooting points may be set on two sides of a road or in a road (for example, on a traffic light), or shooting may be performed by an unmanned aerial vehicle, which is not particularly limited herein.
And step S107, inputting the image data into a second monitoring model to obtain a second monitoring result output by the second monitoring model.
Before step S107, the method further includes the steps of:
step S1061, constructing a second monitoring model, where the second monitoring model includes a plurality of neural networks.
Step S1062, selecting a second training sample, wherein the second training sample comprises image data of an abnormal road; the abnormal road is a road where a crack exists.
And step S1063, inputting the second training sample into the second monitoring model for training to obtain a prediction result of the second monitoring model.
In the prior art, a neural network is used for identifying road cracks, and a U-Net neural network is often adopted, so that the requirements on the image quality of the image are high, but the image quality of the shot image is often low due to the influence of the actual situation, and the obtained identification effect is not ideal.
In this embodiment, the images are first acquired as training samples, preprocessed and labeled (labeled using Labelimg tools), as is well known in the art, and not described in detail herein. In some embodiments, the fracture types include road transverse fractures, road longitudinal fractures, road block fractures, and road mesh fractures.
Then, the plurality of neural networks includes a VGG16 neural network and a U-Net neural network; the second monitoring model takes the U-Net neural network as a main body of the model, replaces an encoder of the U-Net neural network with the VGG16 neural network, and initializes the weight of the VGG16 neural network through the ImageNet. Because of the problems of low image quality (high noise) and small number of training samples in the actual situation, the method combines the existing VGG16 neural network and the U-Net neural network, takes the U-Net neural network as a main body, removes an encoder in the U-Net neural network, improves classification accuracy by using the pretrained VGG16 neural network as the encoder of the U-Net neural network, and prevents overfitting.
Wherein: the U-Net neural network includes thirteen convolution layers, each being a 3*3 convolution kernel, with an activation function of RELU. The first two convolutional layers have a channel number of 64, the third and fourth convolutional layers have 128, the fifth and fourth convolutional layers have 256, and the seventh through thirteenth convolutional layers have 512. The first through tenth layers of the U-Net neural network are initialized using weights of the ImageNet pre-training. The VGG16 neural network as a decoder upsamples (2 x 2 up-convolves) the feature map of the convolutional layer and concatenates the feature map with the feature map of the convolutional layer, respectively, and then convolves the feature map with 3*3 to reduce the dimension, using ReLU as an activation function.
It should be noted that image training by a U-Net neural network is well known in the art, and the principle and internal structure thereof will not be described here. The improvement of the embodiment is that: the method combines the VGG16 neural network and the U-Net neural network which are used in the prior art, takes the U-Net neural network as a main body, removes the encoder inside the U-Net neural network, improves the classification precision by using the pretrained VGG16 neural network as the encoder of the U-Net neural network, and prevents the overfitting.
And step 109, judging the crack type of the target road according to the second monitoring result, and generating an alarm signal corresponding to the crack type. After the corresponding alarm signal is sent out in the step, a maintainer goes to the corresponding road area to repair the existing cracks, such as asphalt filling the cracks, so as to ensure the road safety.
The method has the following effects:
(1) The method utilizes the characteristic that vibration parameters of a road crack are different from those of a normal road, firstly, vibration data of a target road are collected, and whether the target road is cracked or not is judged from the vibration data by using a first monitoring model; then collecting image data of a target road, and judging the specific type of the crack from the image data by using a second monitoring model; and finally, executing corresponding alarm operation according to the judged crack type. The method is mainly divided into two parts, wherein the first part accurately judges whether the road is cracked or not by utilizing vibration data of the target road, and the second part further judges the type of the crack on the basis of the first part, so that the accuracy of monitoring the road crack can be improved, and false detection caused by missing detection is avoided.
(2) In the method, in the preliminary judgment of the first monitoring model, the two machine learning classifiers with different principles are utilized to judge whether the crack exists on the target road or not, and the judgment result is more accurate.
(3) In the method, in the classification monitoring of the second monitoring model, the existing VGG16 neural network and the U-Net neural network are combined, the U-Net neural network is taken as a main body, the encoder in the U-Net neural network is removed, the classification precision is improved by using the pretrained VGG16 neural network as the encoder of the U-Net neural network, and the overfitting is prevented.
In one embodiment of the present invention, a road traffic crack monitoring device is provided, which may be any type of intelligent terminal, such as a mobile phone, tablet computer, personal computer, etc. Specifically, the apparatus includes: one or more control processors and memory. The control processor and the memory may be connected by a bus or other means.
The memory, as a non-transitory computer readable storage medium, may be used to store non-transitory software programs, non-transitory computer-executable programs, and modules, such as program instructions/modules corresponding to the apparatus in embodiments of the present invention. The control processor executes various functional applications and data processing of the road traffic crack monitoring system by running non-transitory software programs, instructions and modules stored in the memory, namely, the road traffic crack monitoring method of the method embodiment is realized.
The memory may include a memory program area and a memory data area, wherein the memory program area may store an operating system, at least one application program required for a function; the storage data area may store data created by the use of the road traffic crack monitoring system, etc. In addition, the memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one magnetic disk storage device, flash memory device, or other non-transitory solid state storage device. In some embodiments, the memory optionally includes memory remotely located relative to the control processor, the remote memory being connectable to the device via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
The one or more modules are stored in the memory that, when executed by the one or more control processors, perform a road traffic crack monitoring method in the above-described method embodiments.
Embodiments of the present invention also provide a computer-readable storage medium storing computer-executable instructions that are executed by one or more control processors to cause the one or more control processors to perform a road traffic crack monitoring method according to the above-described method embodiments.
The above described embodiments of the apparatus are only illustrative, wherein the units described as separate components may or may not be physically separate, i.e. may be located in one place, or may be distributed over a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
From the above description of embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented in software plus a general purpose hardware platform. Those skilled in the art will appreciate that all or part of the processes implementing the methods of the above embodiments may be implemented by a computer program for instructing relevant hardware, where the program may be stored in a computer readable storage medium, and where the program, when executed, may include the processes of the embodiments of the methods described above. The storage medium may be a magnetic disk, an optical disk, a Read Only Memory (ROM), a random access Memory (Random Access Memory, RAM), or the like.
In the description of the present specification, reference to the terms "one embodiment," "some embodiments," "illustrative embodiments," "examples," "specific examples," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiments or examples. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
While embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that: many changes, modifications, substitutions and variations may be made to the embodiments without departing from the spirit and principles of the invention, the scope of which is defined by the claims and their equivalents.
Claims (6)
1. The road traffic crack monitoring method is characterized by comprising the following steps of:
obtaining vibration data of a target road;
constructing a first monitoring model, wherein the first monitoring model comprises two different classes of classifiers, and the two classifiers comprise a support vector machine classifier and a Bayesian classifier; the training of the first monitoring model comprises: selecting a first training sample, wherein the first training sample comprises a first classification sample of an abnormal road and a second classification sample of a normal road; the abnormal road is a road with a crack, the normal road is a road without a crack, the first classification sample and the second classification sample both comprise vibration data of the corresponding road, and the vibration data comprise vibration displacement, vibration amplitude and vibration frequency; respectively inputting the first classification sample and the second classification sample into two classifiers for training; taking classification results of the two classifiers as prediction results of the first monitoring model; inputting the vibration data into a first monitoring model to obtain a first monitoring result output by the first monitoring model;
if the first monitoring result shows that the target road has a crack, acquiring image data of the target road;
constructing a second monitoring model, wherein the second monitoring model comprises a plurality of neural networks; the plurality of neural networks includes a VGG16 neural network and a U-Net neural network; the second monitoring model takes the U-Net neural network as a main body of the second monitoring model, replaces an encoder of the U-Net neural network with the VGG16 neural network, and initializes the weight of the VGG16 neural network through imageNet; the process of training the second monitoring model comprises the following steps: selecting a second training sample, wherein the second training sample comprises image data of an abnormal road; the abnormal road is a road with cracks; inputting a second training sample into the second monitoring model for training to obtain a prediction result of the second monitoring model; inputting the image data into a second monitoring model to obtain a second monitoring result output by the second monitoring model;
judging the crack type of the target road according to the second monitoring result, and generating an alarm signal corresponding to the crack type; the crack types include a road transverse crack, a road longitudinal crack, a road block crack, and a road mesh crack.
2. The method of claim 1, wherein the support vector machine classifier selects a gaussian radial basis function as a kernel function;
the judging accuracy of the Bayesian classifier comprises the following steps:
wherein ,
the P (a) represents the prior probability of the first training sample, the P (y) i I a) represents the conditional probability of the first training sample, y i Representing the ith eigenvalue in training sample y, said a representing the type of training sample, said σ a,i Representing the variance of the ith feature in the class a training sample, said μ a,i Representing the mean of the ith feature in a class a training sample, said M representing the total number of features, said a 1 Representing the first classified sample, said a 2 Representing the second classification sample, the X a Representing a set of a class a training samples in the first training samples, wherein l represents the number of samples.
3. The method for monitoring the road traffic crack according to claim 1, wherein the step of inputting the second training samples into the second monitoring model for training comprises the steps of:
inputting the image in the second training sample into a plurality of convolution layers in the U-Net neural network, wherein the plurality of convolution layers is 13 layers, and each layer is a convolution kernel of 3*3;
and upsampling the feature map of the convolution layer through the VGG16 neural network and respectively cascading the feature map with the feature map of the convolution layer.
4. A road traffic crack monitoring system, comprising:
a data acquisition unit for acquiring vibration data and image data of a target road;
the crack prediction unit is used for constructing a first monitoring model, wherein the first monitoring model comprises two different classes of classifiers, and the two classifiers comprise a support vector machine classifier and a Bayesian classifier; the training of the first monitoring model comprises: selecting a first training sample, wherein the first training sample comprises a first classification sample of an abnormal road and a second classification sample of a normal road; the abnormal road is a road with a crack, the normal road is a road without a crack, the first classification sample and the second classification sample both comprise vibration data of the corresponding road, and the vibration data comprise vibration displacement, vibration amplitude and vibration frequency; respectively inputting the first classification sample and the second classification sample into two classifiers for training; taking classification results of the two classifiers as prediction results of the first monitoring model; inputting the vibration data into a first monitoring model to obtain a first monitoring result output by the first monitoring model;
the crack classification unit is used for constructing a second monitoring model, and the second monitoring model comprises a plurality of neural networks; the plurality of neural networks includes a VGG16 neural network and a U-Net neural network; the second monitoring model takes the U-Net neural network as a main body of the second monitoring model, replaces an encoder of the U-Net neural network with the VGG16 neural network, and initializes the weight of the VGG16 neural network through imageNet; the process of training the second monitoring model comprises the following steps: selecting a second training sample, wherein the second training sample comprises image data of an abnormal road; the abnormal road is a road with cracks; inputting a second training sample into the second monitoring model for training to obtain a prediction result of the second monitoring model; inputting the image data of the target road to a second monitoring model to obtain a second monitoring result output by the second monitoring model;
the alarm unit judges the crack type of the target road according to the second monitoring result and generates an alarm signal corresponding to the crack type; the crack types include a road transverse crack, a road longitudinal crack, a road block crack, and a road mesh crack.
5. A road traffic crack monitoring device, comprising: at least one control processor and a memory for communication connection with the at least one control processor; the memory stores instructions executable by the at least one control processor to enable the at least one control processor to perform the road traffic crack monitoring method of any one of claims 1 to 3.
6. A computer-readable storage medium storing computer-executable instructions for causing a computer to perform the road traffic crack monitoring method according to any one of claims 1 to 3.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210192638.4A CN114565793B (en) | 2022-02-28 | 2022-02-28 | Road traffic crack monitoring method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210192638.4A CN114565793B (en) | 2022-02-28 | 2022-02-28 | Road traffic crack monitoring method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114565793A CN114565793A (en) | 2022-05-31 |
| CN114565793B true CN114565793B (en) | 2023-05-23 |
Family
ID=81716330
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210192638.4A Active CN114565793B (en) | 2022-02-28 | 2022-02-28 | Road traffic crack monitoring method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114565793B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110276756A (en) * | 2019-06-25 | 2019-09-24 | 百度在线网络技术（北京）有限公司 | Road surface crack detection method, device and equipment |
| CN112233105A (en) * | 2020-10-27 | 2021-01-15 | 江苏科博空间信息科技有限公司 | Road crack detection method based on improved FCN |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB2570049A (en) * | 2016-10-20 | 2019-07-10 | Landmark Graphics Corp | Classifying well data using a support vector machine |
| WO2018122589A1 (en) * | 2016-12-30 | 2018-07-05 | 同济大学 | Method for detecting degree of development of asphalt pavement fracture based on infrared thermal image analysis |
| CN112785578A (en) * | 2021-01-26 | 2021-05-11 | 汕头大学 | Road crack detection method and system based on U-shaped codec neural network |
| CN113177611B (en) * | 2021-05-24 | 2022-11-01 | 河北工业大学 | Pavement disease rapid inspection method based on mechanical index and artificial neural network |
| CN113850228A (en) * | 2021-10-18 | 2021-12-28 | 北方民族大学 | Road crack detection method and system based on multi-mode fusion |
-
2022
- 2022-02-28 CN CN202210192638.4A patent/CN114565793B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110276756A (en) * | 2019-06-25 | 2019-09-24 | 百度在线网络技术（北京）有限公司 | Road surface crack detection method, device and equipment |
| CN112233105A (en) * | 2020-10-27 | 2021-01-15 | 江苏科博空间信息科技有限公司 | Road crack detection method based on improved FCN |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114565793A (en) | 2022-05-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Tian et al. | An automatic car accident detection method based on cooperative vehicle infrastructure systems | |
| Chen et al. | Analysis of factors affecting the severity of automated vehicle crashes using XGBoost model combining POI data | |
| Cano-Ortiz et al. | Machine learning algorithms for monitoring pavement performance | |
| CN108001456A (en) | Object Classification Adjustment Based On Vehicle Communication | |
| CN110866427A (en) | Vehicle behavior detection method and device | |
| US12014423B1 (en) | Using vehicle data, geographic area type data, and vehicle collision data in determining an indication of whether a vehicle in a vehicle collision is a total loss | |
| CN115294767B (en) | Real-time detection and traffic safety early warning method and device for expressway lane line | |
| CN108460685A (en) | Method and apparatus for excavating the correlation rule between vehicle insurance risks and assumptions | |
| CN115909281A (en) | Matching fusion obstacle detection method and system, electronic device and storage medium | |
| CN110838230A (en) | Mobile video monitoring method, monitoring center and system | |
| CN117649632B (en) | Expressway event identification method and device based on multi-source traffic data | |
| CN116631186A (en) | Expressway traffic accident risk assessment method and system based on dangerous driving event data | |
| CN116289444A (en) | Intelligent pavement disease real-time monitoring and predicting device and method | |
| CN117372979A (en) | Road inspection method, device, electronic equipment and storage medium | |
| Kumar et al. | Moving Vehicles Detection and Tracking on Highways and Transportation System for Smart Cities | |
| Lincy et al. | Road Pothole Detection System | |
| CN114565793B (en) | Road traffic crack monitoring method and system | |
| CN113701642A (en) | Method and system for calculating appearance size of vehicle body | |
| CN115147618A (en) | Method for generating saliency map, method and device for detecting abnormal object | |
| US20230314169A1 (en) | Method and apparatus for generating map data, and non-transitory computer-readable storage medium | |
| CN116946183A (en) | Commercial vehicle driving behavior prediction method considering driving capability and vehicle equipment | |
| Peng et al. | A Method for Vehicle Collision Risk Assessment through Inferring Driver's Braking Actions in Near-Crash Situations | |
| Ma et al. | High-resolution traffic sensing with autonomous vehicles | |
| CN111354191B (en) | Lane driving condition determining method, device and equipment and storage medium | |
| CN114968189A (en) | Platform for perception system development of an autopilot system |
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 |