CN115311229A

CN115311229A - Laser radar-based pavement disease detection and classification method and system and storage medium

Info

Publication number: CN115311229A
Application number: CN202210939656.4A
Authority: CN
Inventors: 卢金; 王晓南; 成剑华; 何源; 张奇源; 刘铭
Original assignee: Wuhan Zhongguan Automation Technology Co ltd
Current assignee: Wuhan Zhongguan Automation Technology Co ltd
Priority date: 2022-08-05
Filing date: 2022-08-05
Publication date: 2022-11-08

Abstract

The application provides a method, a system and a storage medium for detecting and classifying pavement diseases based on a laser radar, which comprises the following steps: acquiring 3D point cloud data of a road surface based on a laser radar technology; segmenting and clustering the 3D point cloud data, and extracting a damaged area of the pavement; collecting training samples in the damaged areas of the pavement, and training a pavement damage classification deep learning model by using the training samples; and carrying out pavement disease detection classification on the detection sample by using the trained pavement disease classification deep learning model. The method includes the steps of obtaining road surface 3D point cloud data based on a laser radar SLAM technology, automatically dividing the point cloud data, extracting disease areas, collecting training samples, training the models by using a deep learning technology, and performing prediction classification on detection samples by using the trained models, so that labor cost is saved, and efficiency is improved.

Description

Laser radar-based pavement disease detection and classification method and system and storage medium

Technical Field

The invention relates to the technical field of deep learning, in particular to a method, a system and a storage medium for detecting and classifying road surface diseases based on a laser radar.

Background

With the continuous development and progress of society, people's daily life is inseparable with the traffic density, and the road is the important component of traffic system, is the important infrastructure of the development that national economy relies on. In recent years, road construction in China has been developed with sudden and violent progress, the scale of the mileage of passing vehicles is continuously increased, the road mileage in China is continuously increased along with the increasing investment of capital construction in China, and the total road mileage in China is 519.81 kilometers and 16.1 kilometers in expressways by the end of 2020 years. And the running number of motor vehicles is rapidly increased, so that the damage speed of the road surface is increased, and the detection work of the road surface diseases is more and more heavy.

For a long time, the pavement disease detection adopts a field investigation method based on artificial vision, and the method has the defects of high cost, low accuracy, influence on traffic, insecurity, time waste and the like, and can not meet the requirements of the current pavement disease detection work. With the rapid development of computer technology, digital image processing technology has been supported by powerful techniques and is widely used in the fields of biomedicine, remote sensing aerospace, military police and the like, so that researchers at home and abroad begin to try to utilize image processing technology to investigate road surface disease data and develop a series of road surface damage detection systems successively.

However, the current system has the common defects that the applied detection algorithm needs to be optimized, the analysis of the road image still adopts a man-machine combination or even a complete manual mode, and the workload of data processing in the later period is very large and the time consumption is long. Only if an effective automatic pavement disease detection algorithm is designed to quickly and accurately acquire pavement disease data, the road can be ensured to be better met and meet the new automobile age.

At present, most of road surface disease detection algorithms are performed on highway road surface images, and researches on automatic disease detection on a large amount of complex urban road surface data are still few. Compared with an expressway, the service life of the urban road is influenced by more factors, such as the fact that urban road water supply and drainage pipelines and structures are staggered, well covers on the road surface are more, the conditions under the road are complex, vehicles run more frequently, and the like, therefore, the urban road has more road surface diseases, more complex road surface image noise and higher detection difficulty. In order to make cities more harmonious and beautiful and people more convenient to live, the urban pavement disease detection must move to the automatic and digital development direction, the damage condition of the urban pavement is timely grasped, roads are scientifically and reasonably maintained, the potential safety hazard of the roads is avoided, and the urban traffic condition is better improved. How to comprehensively utilize various image processing technologies to automatically detect urban road surface diseases with high precision and high efficiency is a core problem for determining the performance of a road surface damage monitoring system, and becomes a problem to be solved urgently for developing the road traffic industry in China.

Disclosure of Invention

The invention aims to overcome the technical defects, provides a method, a system and a storage medium for detecting and classifying pavement diseases based on a laser radar, and solves the technical problems that the traditional pavement disease detection mode in the prior art mainly adopts a manual or vision-based semi-automatic detection mode, the detection efficiency is low, and the detection cost is high.

In order to achieve the above technical objective, a first aspect of the present invention provides a method for detecting and classifying road surface diseases based on a laser radar, which includes the following steps:

acquiring 3D point cloud data of a road surface based on a laser radar technology;

segmenting and clustering the 3D point cloud data, and extracting a damaged area of the pavement;

collecting training samples in the damaged areas of the pavement, and using the training samples to train a pavement damage classification deep learning model;

and carrying out pavement disease detection classification on the detection sample by using the trained pavement disease classification deep learning model.

Compared with the prior art, the invention has the beneficial effects that:

the method includes the steps of obtaining road surface 3D point cloud data based on a laser radar SLAM technology, automatically dividing the point cloud data, extracting disease areas, collecting training samples, training the models by using a deep learning technology, and performing prediction classification on detection samples by using the trained models, so that labor cost is saved, and efficiency is improved.

According to some embodiments of the invention, segmenting and clustering the 3D point cloud data to extract the damaged area of the road surface comprises the following steps:

carrying out segmentation processing on the 3D point cloud data to obtain segmented data;

clustering the segmented data to obtain a plurality of category point clouds;

performing surface fitting processing on each category point cloud to obtain a fitted surface;

comparing the original point cloud of the 3D point cloud data with the fitting curved surface to generate a color difference map;

and connecting points of the color difference value exceeding a preset threshold value in the color difference image to obtain a damaged area of the pavement.

According to some embodiments of the invention, training samples are collected in the damaged area of the road surface, and the training samples are used for training a road surface damage classification deep learning model, comprising the following steps:

acquiring 3D point cloud data containing pavement diseases, converting the 3D point cloud data into a depth map, and using the depth map as the training sample;

and marking disease category labels on the training samples, and training the pavement disease classification deep learning model in a deep learning network by using the training samples.

According to some embodiments of the invention, segmenting the 3D point cloud data to obtain segmented data comprises:

generating a track map of the laser radar;

and segmenting the 3D point cloud data at preset intervals along the track direction of the track map to obtain segmented data.

According to some embodiments of the invention, the disease category label of the training sample comprises at least any one of:

transverse cracks, longitudinal cracks, block cracks, crazing, repairing, pits, ruts, and bulges.

According to some embodiments of the present invention, clustering the segmented data to obtain a plurality of category point clouds includes:

calculating a curvature value of 3D point cloud data of the segmented data;

sorting the 3D point cloud data from small to large according to the curvature values, finding out a minimum curvature value point and adding the minimum curvature value point to a seed point set;

searching adjacent points around each seed point, and calculating the normal angle difference between each adjacent point and the current seed point;

if the adjacent points pass the normal angle difference test and the curvature is smaller than a set threshold value, adding the adjacent points to a seed point set;

and setting the point number of the minimum point cluster and the point number of the maximum point cluster, and generating all the category point clouds of which the point numbers are between the point number of the minimum point cluster and the point number of the maximum point cluster.

According to some embodiments of the invention, the method for detecting and classifying the road surface diseases of the detection samples by using the trained deep learning model for classifying the road surface diseases comprises the following steps:

dividing 3D point cloud data to be detected into continuous point cloud segment sequences at intervals of a preset distance along a track direction;

and converting each point cloud segment into a depth map, and sending the depth map into the pavement disease classification deep learning model to detect and classify the pavement diseases of the detection sample.

According to some embodiments of the invention, the road surface disease classification deep learning model comprises:

a CNN model comprising: a convolutional layer for extracting features; a pooling layer for down-sampling; the full connecting layer is used for classifying pavement diseases;

an RNN model connected to an output of the CNN model, comprising: an input layer, a hidden layer, and an output layer.

In a second aspect, a technical solution of the present invention provides a laser radar-based pavement damage detection and classification system, including: the system comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the processor executes the computer program to realize the laser radar-based road surface damage detection and classification method according to any one of the first aspect.

In a third aspect, the present invention provides a computer-readable storage medium, where computer-executable instructions are stored, and the computer-executable instructions are configured to enable a computer to execute the method for detecting and classifying a road surface disease based on a lidar according to any of the first aspects.

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 above and/or additional aspects and advantages of the present invention will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings of which the abstract is to be fully consistent with one of the figures of the specification:

fig. 1 is a flowchart of a method for detecting and classifying road surface diseases based on a laser radar according to an embodiment of the present invention;

fig. 2 is a flowchart of a method for detecting and classifying road surface diseases based on a laser radar according to another embodiment of the present invention;

fig. 3 is a flowchart of a method for detecting and classifying road surface diseases based on a laser radar according to another embodiment of the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

It should be noted that although functional block divisions are provided in the system drawings and logical orders are shown in the flowcharts, in some cases, the steps shown and described may be performed in different orders than the block divisions in the systems or in the flowcharts. The terms first, second and the like in the description and in the claims, and the drawings described above, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order.

The invention provides a laser radar-based pavement disease detection and classification method, which is characterized in that 3D point cloud data of a pavement are obtained based on a laser radar SLAM technology, then the point cloud data are automatically segmented, a disease area is extracted, then training samples are collected, a model is trained by using a deep learning technology, and detection samples are predicted and classified by using the trained model, so that the labor cost is saved and the efficiency is improved.

The embodiments of the present invention will be further explained with reference to the drawings.

Referring to fig. 1, fig. 1 is a flowchart of a method for detecting and classifying road surface diseases based on a laser radar according to an embodiment of the present invention; the method for detecting and classifying the pavement diseases based on the laser radar comprises the following steps:

step S110, 3D point cloud data of a road surface are obtained based on a laser radar technology;

step S120, segmenting and clustering the 3D point cloud data, and extracting a damaged area of the road surface;

step S130, collecting training samples in the damaged areas of the road surface, and training a road surface damage classification deep learning model by using the training samples;

and step S140, carrying out pavement disease detection classification on the detection sample by using the trained pavement disease classification deep learning model.

In one embodiment, the method for detecting and classifying the pavement diseases based on the laser radar comprises the following steps: acquiring 3D point cloud data of a road surface based on a laser radar technology; segmenting and clustering the 3D point cloud data, and extracting a damaged area of the pavement; collecting training samples in the damaged areas of the road surface, and training a road surface damage classification deep learning model by using the training samples; and carrying out pavement disease detection classification on the detection sample by using the trained pavement disease classification deep learning model.

According to the method for detecting and classifying the pavement diseases based on the laser radar, provided by the invention, the 3D point cloud data of the pavement is obtained based on the SLAM technology, then the point cloud data is automatically segmented, the disease area is extracted, then the training sample is collected, the model is trained by using the deep learning technology, and the detection sample is subjected to prediction classification by using the trained model, so that the labor cost is saved and the efficiency is improved. The pavement disease classification deep learning model provided by the invention has stronger generalization capability and adaptability. Road conditions all over the country are different, samples of all road conditions are difficult to collect, the traditional machine learning algorithm is easy to generate the phenomenon of overfitting, and the classification effect of the samples which are greatly different from training samples is poor. A wider neural network model has good generalization capability. This is because wider networks have more subnetworks and have a greater potential for gradient coherence than smaller networks, resulting in better generalization.

1. The automatic segmentation method of the road surface point cloud data comprises the following steps:

in the point cloud data acquisition process, the laser radar SLAM system can generate a locus diagram of the laser radar according to IMU real-time integral operation. The raw point cloud data may be segmented every N meters along the track direction for subsequent processing.

The IMU measures acceleration and angular velocity through an accelerometer and a gyroscope, and obtains the transformation of rotation and displacement between two frames through integration;

during back-end nonlinear optimization, the pose needs to be optimized, IMU measurement values need to be retransmitted between the pose and the pose each time the pose is adjusted, reintegration is needed, time is consumed, and in order to avoid retransmitting the measurement values, a pre-integration strategy is adopted.

The IMU model is as follows:

wherein:

b _ωt bias n of gyroscope _ω Additive noise

b _at Offset n of accelerometer _a Additive noise

The continuous expression form of PVQ at the present time is as follows:

the median discrete expression of the PVQ at the current moment is as follows:

2. the method for clustering the point cloud of the road surface comprises the following steps:

region growing starts from the point where there is a minimum curvature value. Therefore, we have to calculate all curvature values and rank them. This is because the point of least curvature is located in the flat region, and growing from the flattest region may reduce the total number of regions. We now describe this process in detail:

1. the point cloud has unmarked points, the points are sorted according to the curvature values of the points, the point with the minimum curvature value is found, and the point with the minimum curvature value is added into the seed point set;

2. for each seed point, the algorithm finds all the neighbors of the perimeter. Calculating the normal angle difference SmoothnessThreshold between each neighboring point and the current seed point, and if the difference value is smaller than a set threshold value, the neighboring point is considered in a key manner, and the second step of test is carried out; the neighbor point passes the normal angle difference test and if its curvature is smaller than the threshold CurvatureThreshold set by us, this point is added to the set of seed points, i.e. belongs to the current plane.

3. The points that pass the two checks are removed from the original point cloud.

4. And setting the number min of the minimum point cluster, and setting the number of the maximum point cluster as max.

5. Repeating the steps 1-3, generating all planes with the points of min and max by the algorithm, and marking different colors on different planes for distinguishing.

6. And stopping the algorithm until the point clusters generated by the algorithm in the rest points cannot meet the min.

3. The method is used for establishing a deep learning model for training the classification of the pavement diseases:

at present, a Convolutional Neural Network (CNN) is a hot direction of deep learning, and is successfully applied to the field of vision-based target detection and classification. The CNN model is mainly generated by stacking three basic layers in different combinations. Convolutional Layer (Convolutional Layer) -the main role is to extract features; a Pooling Layer (Max Pooling Layer) -mainly plays a role of downsampling (down sampling), but does not damage the recognition result; fully Connected Layer (Fully Connected Layer) -the main role is classification.

RNN recurrent neural networks are applicable to temporal data and other types of sequence data. The data may be text, stock market data, or letters and words in speech recognition. The inputs and outputs of the RNN may be data of any length. LSTM is a variant of RNN that remembers a controlled amount of pre-training data.

The RNN model comprises an input x, a hidden layer s and an output layer o; u is a weight matrix from the input layer to the hidden layer, V is a weight matrix from the hidden layer to the output layer, and further includes a self-loop W which is a weight matrix using the last value of the hidden layer as the next input. The input samples of the CNN model are generally time-sequence independent, and the model itself has no time-sequence relation between the samples before and after mining. In order to utilize the time sequence relevance before the samples, the invention combines an RNN model and a CNN model, thereby improving the accuracy of classification.

Referring to fig. 2, fig. 2 is a flowchart of a method for detecting and classifying road surface diseases based on a laser radar according to another embodiment of the present invention; the method for detecting and classifying the pavement diseases based on the laser radar comprises the following steps:

step S210, performing segmentation processing on the 3D point cloud data to obtain segmented data;

step S220, clustering segmented data to obtain a plurality of category point clouds;

step S230, performing surface fitting processing on each category point cloud to obtain a fitting surface;

step S240, comparing the original point cloud of the 3D point cloud data with the fitting curved surface to generate a color difference map;

and S250, connecting points of which the color difference values exceed a preset threshold value in the color difference image to obtain a damaged area of the road surface.

The method comprises the following steps of segmenting and clustering the 3D point cloud data, and extracting the disease area of the pavement, wherein the method comprises the following steps: performing segmentation processing on the 3D point cloud data to obtain segmented data; clustering the segmented data to obtain a plurality of category point clouds; performing surface fitting processing on each category point cloud to obtain a fitted surface; comparing the original point cloud of the 3D point cloud data with the fitting curved surface to generate a color difference map; and connecting points of the color difference value exceeding a preset threshold value in the color difference image to obtain a damaged area of the pavement.

Referring to fig. 3, fig. 3 is a flowchart of a method for detecting and classifying road surface diseases based on a laser radar according to another embodiment of the present invention; the method for detecting and classifying the pavement diseases based on the laser radar comprises the following steps:

step S310, acquiring 3D point cloud data containing the road surface diseases, converting the data into a depth map and using the depth map as a training sample;

and step S320, marking disease category labels on the training samples, and training a pavement disease classification deep learning model in a deep learning network by using the training samples.

In one embodiment, the method for detecting and classifying the pavement diseases based on the laser radar comprises the following steps: acquiring 3D point cloud data of a road surface based on a laser radar technology; segmenting and clustering the 3D point cloud data, and extracting a damaged area of the pavement; collecting training samples in the damaged areas of the pavement, and training a pavement damage classification deep learning model by using the training samples; and carrying out pavement disease detection classification on the detection sample by using the trained pavement disease classification deep learning model.

Collecting training samples in the damaged areas of the road surface, and training a road surface damage classification deep learning model by using the training samples, wherein the method comprises the following steps: acquiring 3D point cloud data containing pavement diseases, converting the 3D point cloud data into a depth map, and using the depth map as a training sample; and marking disease category labels on the training samples, and training a pavement disease classification deep learning model in a deep learning network by using the training samples.

The method comprises the following steps of segmenting and clustering the 3D point cloud data, and extracting the disease area of the pavement, wherein the method comprises the following steps: carrying out segmentation processing on the 3D point cloud data to obtain segmented data; clustering the segmented data to obtain a plurality of category point clouds; performing surface fitting processing on each category point cloud to obtain a fitted surface; comparing the original point cloud of the 3D point cloud data with the fitting curved surface to generate a color difference map; and connecting points of the color difference value exceeding a preset threshold value in the color difference image to obtain a damaged area of the pavement. Collecting training samples in the damaged area of the pavement, and training a pavement damage classification deep learning model by using the training samples, wherein the method comprises the following steps: acquiring 3D point cloud data containing pavement diseases, converting the 3D point cloud data into a depth map, and using the depth map as a training sample; and marking disease category labels on the training samples, and training a pavement disease classification deep learning model in a deep learning network by using the training samples.

The method comprises the following steps of segmenting and clustering the 3D point cloud data, and extracting the disease area of the pavement, wherein the method comprises the following steps: performing segmentation processing on the 3D point cloud data to obtain segmented data; clustering the segmented data to obtain a plurality of category point clouds; performing surface fitting processing on each category point cloud to obtain a fitted surface; comparing the original point cloud of the 3D point cloud data with the fitting curved surface to generate a color difference graph; and connecting points of the color difference value exceeding a preset threshold value in the color difference image to obtain a damaged area of the pavement. The method for segmenting the 3D point cloud data to obtain segmented data comprises the following steps: generating a track map of the laser radar; and segmenting the 3D point cloud data at preset intervals along the track direction of the track map to obtain segmented data.

The method comprises the following steps of segmenting and clustering the 3D point cloud data, and extracting the disease area of the pavement, wherein the method comprises the following steps: performing segmentation processing on the 3D point cloud data to obtain segmented data; clustering the segmented data to obtain a plurality of category point clouds; performing surface fitting processing on each category point cloud to obtain a fitted surface; comparing the original point cloud of the 3D point cloud data with the fitting curved surface to generate a color difference map; and connecting points of the color difference value exceeding a preset threshold value in the color difference image to obtain a damaged area of the pavement. Collecting training samples in the damaged areas of the road surface, and training a road surface damage classification deep learning model by using the training samples, wherein the method comprises the following steps: acquiring 3D point cloud data containing pavement diseases, converting the 3D point cloud data into a depth map, and using the depth map as a training sample; and marking disease category labels on the training samples, and training a pavement disease classification deep learning model in a deep learning network by using the training samples. The disease category label of the training sample at least comprises any one of the following items: transverse cracks, longitudinal cracks, block cracks, crazing, repairing, pits, ruts, and bumps.

The method comprises the following steps of segmenting and clustering the 3D point cloud data, and extracting the disease area of the pavement, wherein the method comprises the following steps: carrying out segmentation processing on the 3D point cloud data to obtain segmented data; clustering the segmented data to obtain a plurality of category point clouds; performing surface fitting processing on each category point cloud to obtain a fitted surface; comparing the original point cloud of the 3D point cloud data with the fitting curved surface to generate a color difference graph; and connecting points of the color difference value exceeding a preset threshold value in the color difference image to obtain a damaged area of the pavement. Clustering segmented data to obtain a plurality of category point clouds, comprising the following steps: calculating curvature values of 3D point cloud data of the segmented data; sorting the 3D point cloud data from small to large according to the curvature values, finding out a minimum curvature value point, and adding the minimum curvature value point to the seed point set; searching adjacent points around each seed point, and calculating the normal angle difference between each adjacent point and the current seed point; if the neighboring points pass the normal angle difference test and the curvature is smaller than a set threshold value, adding the neighboring points to the seed point set; and setting the point number of the minimum point cluster and the point number of the maximum point cluster, and generating all category point clouds of which the point numbers are between the point number of the minimum point cluster and the point number of the maximum point cluster.

The method for clustering the point cloud of the road surface comprises the following steps: region growing starts from the point where there is a minimum curvature value. Therefore, we must compute all curvature values and rank them. This is because the point of least curvature is located in the flat region, and growing from the flattest region may reduce the total number of regions. We now describe this process in detail:

5. Repeating the steps 1-3, generating all planes with the points at min and max by the algorithm, and marking different colors on different planes for distinguishing.

In one embodiment, the method for detecting and classifying the pavement diseases based on the laser radar comprises the following steps: acquiring 3D point cloud data of a road surface based on a laser radar technology; carrying out segmentation and clustering processing on the 3D point cloud data, and extracting a damaged area of the pavement; collecting training samples in the damaged areas of the road surface, and training a road surface damage classification deep learning model by using the training samples; and carrying out pavement disease detection classification on the detection sample by using the trained pavement disease classification deep learning model. The method for detecting and classifying the pavement diseases of the detection samples by using the trained pavement disease classification deep learning model comprises the following steps: dividing 3D point cloud data to be detected into continuous point cloud segment sequences at intervals of a preset distance along a track direction; and converting each point cloud segment into a depth map, and sending the depth map into a pavement disease classification deep learning model to detect and classify the pavement diseases of the detection sample.

In one embodiment, the method for detecting and classifying the pavement diseases based on the laser radar comprises the following steps: acquiring 3D point cloud data of a road surface based on a laser radar technology; carrying out segmentation and clustering processing on the 3D point cloud data, and extracting a damaged area of the pavement; collecting training samples in the damaged areas of the road surface, and training a road surface damage classification deep learning model by using the training samples; and carrying out pavement disease detection classification on the detection sample by using the trained pavement disease classification deep learning model. The road surface disease classification deep learning model comprises the following steps: a CNN model comprising: a convolutional layer for extracting features; a pooling layer for down-sampling; the full-connection layer is used for classifying the pavement diseases; RNN model, with CNN model's output connection, include: an input layer, a hidden layer, and an output layer.

The invention also provides a laser radar-based pavement disease detection and classification system, which comprises: the system comprises a memory, a processor and a computer program which is stored in the memory and can run on the processor, wherein the processor executes the computer program to realize the laser radar-based pavement damage detection and classification method.

The processor and memory may be connected by a bus or other means.

The memory, which is a non-transitory computer readable storage medium, may be used to store non-transitory software programs as well as non-transitory computer executable programs. Further, the memory may include high speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid state storage device. In some embodiments, the memory optionally includes memory located remotely from the processor, and these remote memories may be connected to the processor through 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 above-described embodiments of the apparatus are merely illustrative, wherein the units illustrated as separate components may or may not be physically separate, i.e. may be located in one place, or may also be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment.

Furthermore, an embodiment of the present invention further provides a computer-readable storage medium, where the computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions, which are executed by a processor or a controller, for example, by a processor in the foregoing terminal embodiment, may cause the processor to execute the laser radar-based road surface damage detection and classification method in the foregoing embodiment.

One of ordinary skill in the art will appreciate that all or some of the steps, systems, and methods disclosed above may be implemented as software, firmware, hardware, and suitable combinations thereof. Some or all of the physical components may be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application specific integrated circuit. Such software may be distributed on computer readable media, which may include computer storage media (or non-transitory media) and communication media (or transitory media). The term computer storage media includes volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data, as is well known to those of ordinary skill in the art. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital Versatile Disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and which can accessed by a computer. In addition, communication media typically embodies computer readable instructions, data structures, program modules or other data in a modulated data signal such as a carrier wave or other transport mechanism and includes any information delivery media as known to those skilled in the art.

While the preferred embodiments of the present invention have been described, the present invention is not limited to the above embodiments, and those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present invention, and such equivalent modifications or substitutions are to be included within the scope of the present invention defined by the appended claims.

The above-described embodiments of the present invention should not be construed as limiting the scope of the present invention. Any other corresponding changes and modifications made according to the technical idea of the present invention should be included in the protection scope of the claims of the present invention.

Claims

1. A pavement disease detection and classification method based on a laser radar is characterized by comprising the following steps:

collecting training samples in the damaged areas of the road surface, and training a road surface damage classification deep learning model by using the training samples;

and carrying out pavement disease detection classification on a detection sample by using the trained pavement disease classification deep learning model.

2. The method for detecting and classifying the road surface diseases based on the laser radar as claimed in claim 1, wherein the 3D point cloud data is segmented and clustered to extract the disease areas of the road surface, and the method comprises the following steps:

clustering the segmented data to obtain a plurality of category point clouds;

comparing the original point cloud of the 3D point cloud data with the fitting curved surface to generate a color difference graph;

3. The laser radar-based road surface damage detection and classification method according to claim 1 or 2, characterized in that training samples are collected in the damaged area of the road surface, and a road surface damage classification deep learning model is trained by using the training samples, comprising the steps of:

4. The method for detecting and classifying the road surface diseases based on the laser radar as claimed in claim 2, wherein the step of performing segmentation processing on the 3D point cloud data to obtain segmented data comprises the following steps:

generating a track map of the laser radar;

5. The method for detecting and classifying the laser radar-based road surface diseases according to claim 3, wherein the disease category labels of the training samples at least comprise any one of the following:

6. The method for detecting and classifying the road surface diseases based on the laser radar as claimed in claim 2, wherein the step of clustering the segmented data to obtain a plurality of category point clouds comprises the steps of:

calculating a curvature value of 3D point cloud data of the segmented data;

7. The method for detecting and classifying the road surface diseases based on the laser radar as claimed in claim 1, wherein the trained deep learning model for classifying the road surface diseases is used for detecting and classifying the road surface diseases of the detection samples, and the method comprises the following steps:

and converting each point cloud segment into a depth map, and sending the depth map into the pavement disease classification deep learning model to detect and classify pavement diseases of the detection sample.

8. The method for detecting and classifying the road surface diseases based on the laser radar as claimed in claim 1, wherein the deep learning model for classifying the road surface diseases comprises:

9. The utility model provides a road surface disease detection classification system based on laser radar which characterized in that includes: a memory, a processor and a computer program stored on the memory and operable on the processor, the processor implementing the lidar-based road surface damage detection and classification method according to any one of claims 1 to 8 when executing the computer program.

10. A computer-readable storage medium storing computer-executable instructions for causing a computer to perform the method for lidar based pavement disease detection and classification according to any one of claims 1 to 8.