CN108760740B - Quick detection method for road surface skid resistance based on machine vision - Google Patents
Quick detection method for road surface skid resistance based on machine vision Download PDFInfo
- Publication number
- CN108760740B CN108760740B CN201810548514.9A CN201810548514A CN108760740B CN 108760740 B CN108760740 B CN 108760740B CN 201810548514 A CN201810548514 A CN 201810548514A CN 108760740 B CN108760740 B CN 108760740B
- Authority
- CN
- China
- Prior art keywords
- road surface
- skid
- model
- image
- picture
- 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
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
Landscapes
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Traffic Control Systems (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a method for rapidly detecting the anti-skid performance of a road surface based on machine vision, which is used for rapidly detecting the anti-skid performance of the road surface based on the method of machine vision, namely extracting texture information from a road surface picture, screening out components with obvious anti-skid effect on the road surface, eliminating the influence of noise textures and other apparent characteristics, establishing a relevant model of the noise textures and the anti-skid indexes of the road surface, and realizing the pre-estimation of the anti-skid performance based on image recognition. Compared with the prior art, the method can effectively judge whether the anti-skid performance of the pavement meets the safety standard, has high measurement speed and low cost, is suitable for large-scale road disease investigation and detection, and can effectively improve traffic safety.
Description
Technical Field
The invention relates to the technical field of pavement quality detection and automatic information acquisition, in particular to a method for quickly detecting the skid resistance of a pavement based on machine vision.
Background
The anti-skid performance of the road surface is one of the most important factors influencing the driving safety, and the influence is more obvious particularly in the rain and snow environment. The road surface skid resistance is generally considered to be the sliding resistance of a tire when the tire is locked, or the resistance of a vehicle on a road surface when the tire does not roll. According to the accident data of 1973 in the UK, the friction coefficient of the expressway is improved by 0.15, and the traffic accident rate in rainy days is reduced from the original 56 times/kilometer to 29 times/kilometer. The TRRL RR76 report shows: for every 10% increase in the skid resistance of the pavement, the accident rate will be reduced by 13%. The anti-skid property not only affects the traffic safety by itself, but also the decay condition and the volatility under different environments are important parameters affecting the traffic safety.
Currently, in the research of the road surface anti-skid performance detection method, the characterization parameters of the anti-skid performance are mainly divided into two types: firstly, the road surface index: the road surface structure depth, the response index: coefficient of road surface friction. The former detection is mainly divided into three categories: cross section type, volume method. The profiler mainly directly measures the texture depth of a road and calculates the construction depth of the road by detecting the vertical section of a rough structure. The devices commonly seen on the market are laser profilers, light scanners and the like. The sand spreading method spreads standard sand with known volume on the road surface, pushes the sand into a circle as much as possible by a flat plate, and the ratio of the volume of the sand to the average area of the spread circle is the structural depth of the road surface. Through experimental comparison, the correlation coefficient of the depth of the pavement structure obtained by the laser profiler and the sand laying method is only about 0.45, namely the correlation coefficient and the depth of the pavement structure are not obviously related. Therefore, although the sand-laying method is simple to operate, the metering parameters under different environments are difficult to compare due to the influence of individual differences in the operation process. In recent years, some countries adopt glass balls as a substitute for sand, and the reproducibility of measured values can be improved to a certain extent. Although the structural depth is the characteristic of the road surface, the action mechanism is complex, and the relation with the actual driving quality is difficult to establish after the parameters are obtained. Therefore, the friction coefficient test is recommended in more occasions.
The measurement parameters of the friction coefficient also vary depending on the detection method, and generally mainly include: the lock wheel trailer method, the yaw wheel method, the braking distance method, and the pendulum method. The wheel locking trailer method is the most intuitive detection method of the friction coefficient, its test procedure is to drag the single-wheeled or double-wheeled trailer equipped with standard test tire on the wet road surface of sprinkling water with the required measuring speed by the towing vehicle; the locking test wheel can obtain a slip coefficient SN to represent the steady-state anti-slip capability of the road surface by measuring the traction force and dividing the traction force by the vertical force acting on the tire. The method is a discrete measuring method, has certain requirements on the quality of a tested road section, and is slow.
The deflection pulley method is a continuous anti-skid detection method, and effectively improves the detection efficiency of the anti-skid of the road surface. The device is mainly characterized in that two standard test tires which rotate freely are arranged on a trailer and deflect a certain angle to the driving direction of a vehicle. When the automobile is driven on a wet road surface at a certain speed under the traction of the automobile, the lateral friction force between the test tire and the road surface is acted by the lateral friction force, and the lateral anti-skid coefficient SFC is obtained by dividing the recorded lateral friction force by the load acted on the test wheel. The method can improve the detection efficiency in the use process, but the measurement speed has standard requirements and needs a special and non-vehicle test environment.
The braking distance method is to measure the anti-slip coefficient directly through the brake of the vehicle, mainly utilizes the test vehicle to run on the wet road surface at a certain speed, when four wheels are braked, the distance from the deceleration sliding of the vehicle to the stop can be used for expressing the constant unsteady anti-slip capability, and is expressed by an automatic distance number SDN. The method can obtain the unsteady anti-slip coefficient, but because each parameter of the tested vehicle has uniqueness, the repeatability of the detection site is low, the contrast among different vehicles is poor, and uniform deployment and uniformity of a large number of vehicle parameters are lacked.
The pendulum instrument method is a pavement anti-skid detection method invented by TRRL in the UK and gradually popularized in the world, and is one of the most widely applied means of the engineering at present. The skid resistance value of the road surface is measured according to the principle that when the pendulum falls from a certain height, potential energy is converted into mechanical energy and overcomes the friction force to do work. The rubber block is arranged on the bottom surface of the hammer, and when the pendulum bob freely swings down from a certain height, the surface of the sliding block is in contact with the test pavement. Part of energy is consumed due to friction between the two, so that the pendulum bob can only swing back to a certain height and is read out from the instrument and represented by a sliding resistance value BPN. However, in the using process, the placing height is difficult to accurately judge, and the same friction area is kept, so that a certain error is caused, and the measuring area of the method is small and is static discrete measurement. Therefore, it is mostly used in engineering acceptance.
The traditional detection method is time-consuming, labor-consuming and expensive, so that the anti-skid performance of the road pavement cannot be detected in time, and potential safety hazards are caused.
Disclosure of Invention
The invention aims to overcome the defects in the prior art, and provides a method for rapidly detecting the skid resistance of a pavement based on machine vision, which can effectively judge whether the skid resistance of the pavement meets the safety standard, has the advantages of high measurement speed, low cost, suitability for large-scale road disease investigation and detection and can effectively improve traffic safety.
The purpose of the invention can be realized by the following technical scheme:
a method for quickly detecting the skid resistance of a road surface based on machine vision comprises the steps of extracting texture information from a road surface picture, screening out components with obvious skid resistance on the road surface, eliminating the influence of noise textures and other apparent characteristics, establishing a correlation model of picture texture parameters and road surface skid resistance indexes, and realizing the prediction of the skid resistance based on image recognition.
Preferably, the method comprises: the method comprises the steps of carrying out quantitative characterization on local differences of each pixel of an image by a local binary feature extraction method, counting binary feature distribution of all pixels, extracting main fluctuation features by utilizing mixed Gaussian distribution, realizing noise filtering, carrying out model training by utilizing three feature vectors of extracted normal distribution weight, average value and variance, and establishing a correlation between normal distribution parameters and pavement skid resistance by adopting a random forest;
meanwhile, the data and the anti-skid index of actual measurement are added on the basis of the existing image decomposition depth network by adopting a transfer learning method.
Preferably, in order to avoid the influence of the sample size on the detection result, the two methods are respectively adopted to compare the measurement errors in the actual operation, and a model with a smaller error is selected as the estimation model.
Preferably, the method specifically comprises the following steps:
(1) selecting high-definition video acquisition equipment, GPS equipment, a test vehicle and a shooting frame number;
(2) arranging the high-definition video acquisition equipment in front of a test vehicle, wherein a lens of the video equipment faces the road surface;
(3) the method comprises the steps that uniform-speed driving is conducted on a road section to be measured, high-definition video pictures of a road surface are collected in real time through high-definition video collection equipment, position information is collected through GPS equipment, data matching of an image and a GPS is conducted through a timestamp, and measured data are stored in a memory card of the high-definition video collection equipment;
(4) extracting local binary characteristics (LBP) of the collected picture, converting the original picture into a gray mode, comparing the gray difference between any pixel point and 8 surrounding pixel points, and marking as 1 if the surrounding gray is greater than or equal to the gray of a central pixel, otherwise, marking as 0;
(5) forming a binary sequence by starting from a 12 o' clock position in a clockwise direction according to the obtained binary value, and converting the binary sequence into a decimal number;
(6) carrying out the LBP extraction method on all pixel points in the picture, counting the number of decimal digits, and drawing a histogram;
(7) performing mixed Gaussian distribution GMM conversion on the histogram, converting the original graphic representation into two normal distribution superposition results, and counting the weight, the mean value and the variance of the two normal distributions;
(8) establishing a road surface skid resistance and normal distribution parameter model by adopting a random forest algorithm, and calculating an average prediction error;
(9) cutting an original image, converting the original image into a plurality of images, carrying out sample training by adopting a transfer learning model to obtain a correlation model of the road surface anti-skid performance and the image fragments, and calculating an average prediction error;
(10) and under the condition of different sample sizes, respectively carrying out prediction error calculation by the algorithms in 9 and 10, selecting a model with a smaller prediction error as a detection model, and carrying out pavement skid resistance detection.
Preferably, the high-definition video acquisition equipment is connected with a test vehicle through bolts.
Preferably, the converted image in step (9) is an image of 150 × 3 pixels.
Compared with the prior art, the method for rapidly detecting the anti-skid performance of the pavement through the vehicle-mounted high-definition video acquisition equipment, the GPS and the like based on the machine vision can effectively solve the problems of time and labor consumption, high price and the like of the traditional detection method, finds the decay of the anti-skid performance of the pavement in time and avoids accidents. The method specifically solves the problems that texture information is extracted from a road surface picture, components with obvious anti-skid effect on the road surface are screened out, the influence of noise textures and other apparent characteristics is brought forward, a correlation model of the noise textures and the anti-skid indexes of the road surface is established, and the estimation of the anti-skid performance based on other images is realized.
Drawings
FIG. 1 is a flow chart of the algorithm of the present invention;
FIG. 2 is a flow chart of local binary feature extraction;
FIG. 3 is an illustration of an LBP feature statistical histogram;
FIGS. 4(a) and 4(b) are graphs showing the results of Gaussian mixture distribution calculations;
FIG. 5 is a schematic diagram of the framework structure of the VGGNet model;
FIG. 6 is a schematic diagram of cross validation error results.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, not all, embodiments of the present invention. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, shall fall within the scope of protection of the present invention.
The invention relates to a method for rapidly detecting the skid resistance of a road surface based on machine vision, and the specific flow is shown in figure 1. The detection method mainly utilizes the methods of local binary characteristics, mixed Gaussian distribution, random forests, transfer learning and the like to realize the function of estimating the anti-skid performance of the road surface through the road surface picture.
Roughly divided into three main steps:
firstly, extracting texture features based on image recognition and associating the texture features with anti-skid performance
Skid resistance is a property of pavement that is primarily caused by the texture of the pavement. However, various other factors, such as tire pressure, accumulated water, temperature, driving load, vehicle speed, pavement foundation allocation, construction conditions and the like, have great influence on the resistance, so that the real-time detection of the skid resistance of the pavement is particularly important. The road surface has different functional problems when the wavelength is less than 0.5mm, and the road surface is called as a fine structure to provide adhesive force for the tire tread so as to ensure the anti-skid capability of the road surface. The road surface with the wavelength of 0.5-50mm is of a coarse structure, and mainly ensures the safety of high-speed running in rainy days. It is generally believed that the build wavelength of less than 50mm is called the micro-texture, which provides substantial skid resistance to the pavement. The effect of macro-texture on slip resistance is complex and often appears as a combined effect with micro-texture.
Since the anti-skid performance is mainly determined by the apparent property, the invention considers that the apparent feature of the road surface is extracted by an image recognition method. In order to ensure accurate extraction of macro-micro features, high-definition video acquisition equipment is proposed, the resolution is not less than 720P, and the frame number is not less than 10 frames per second, so that accurate coverage of each position and reliable identification of textures can be ensured.
Firstly, local binary feature extraction is performed on the acquired picture, as shown in fig. 2. The method mainly comprises the following steps: firstly, converting an RPG image into a gray level image; secondly, taking any central pixel and 8 surrounding pixels in the surrounding rectangular range; comparing the gray difference between the peripheral pixel and the central pixel, if the gray of the peripheral pixel is greater than or equal to the central pixel, marking as 1, otherwise marking as 0; starting from the 12 o' clock position in the clockwise direction, forming values of each element into a binary sequence (such as 11001000), and converting the binary sequence into decimal (between 0 and 256); fifthly, all elements in the image are subjected to the operation, the appearance frequency of each decimal number is counted, and a histogram is drawn, as shown in figure 3.
Because the LBP characteristics of the pictures are microscopic characteristics, wherein part of the characteristics are strongly related to the skid resistance of the road surface, and the other characteristics are noise, the histogram is converted into the superposition of two Gaussian distributions by adopting mixed Gaussian distribution, and the weight, the mean value and the variance are extracted to be used as training parameters. In the calculation process, the parameters are calculated by using an EM heuristic algorithm until convergence, and the calculation result of the mixed Gaussian distribution is shown in fig. 4(a) and fig. 4 (b). And finally, carrying out timestamp label matching on the obtained calculation parameters and the GPS position, and determining objective anti-skid performance indexes of the GPS position to carry out model regression.
And (3) taking 6 obtained attributes including the weight, the mean value and the variance as a training set, taking the actual pavement skid resistance value BPN as a label, and performing model training by adopting a random forest model. In the maintenance and management process, maintenance personnel pay more attention to whether the road surface anti-skid performance meets the standard, so that the label is subjected to binary conversion, the road sections meeting the anti-skid requirements are marked as 1, the road sections not meeting the anti-skid requirements are marked as 0, and the error of random forest prediction is calculated.
Secondly, establishing a transfer learning model based on image characteristics
Deep learning forms a more abstract class or feature of high-level representation properties by combining low-level features to discover a distributed feature representation of the data. Deep learning can realize direct processing of image information, relevant features are automatically extracted through methods such as cluster analysis, and the like, but if the sample size is not enough, problems such as over-fitting or under-fitting may exist, so that the obtained attribute class features are not accurate. Therefore, the invention adopts a transfer learning method on the basis of deep learning. The transfer learning is to transfer the model parameters which are well trained to a new model to help the new model training. Considering that most image texture features are relevant, the learned model parameters (such as VGGNet) can be shared with the new model through transfer learning so as to accelerate and optimize the learning efficiency of the model. The VGGNet network is adopted in the technology, and the technology is mainly characterized in that a 3x3 small convolution kernel and a 2x2 maximum pooling layer are repeatedly stacked, so that a 16-19-layer deep convolution neural network is successfully constructed. The expansibility is strong, and the generalization of migrating other picture data is good. The structure is simple, the whole network uses convolution kernel size and maximum pooling size with the same size, and the model structure is shown in FIG. 5.
Firstly, cutting the picture into small blocks of 150 × 3, labeling according to an actual anti-slip value, inputting the picture into VGGNet for training, directly obtaining a deep learning model, and recording model errors.
Thirdly, comparing errors of LBP and VGGNet under different sample sizes
For the learning process, the size of the sample size directly determines the reliability of the result. However, in the actual calibration process, it is difficult to realize the collection of a large number of samples, and the sample size is different in sensitivity to different models. Through tests, the two models can effectively reflect the anti-skid performance of the road surface, in actual operation, the calibration sample pictures are respectively processed, the estimated errors of the two models are compared, and the model with better error is selected as the actual measurement model. In actual test, the video acquisition device is directly used for acquiring road surface photo information, and the photos are led into the model system for platform calculation to generate a road surface anti-skid performance estimated value.
Example of the implementation
(1) And arranging a high-definition camera on a test vehicle, driving at a recommended test speed of 50km/h, and acquiring video and GPS data.
Selecting a typical asphalt road in a Shanghai city region, carrying out video acquisition by using a detection vehicle, wherein the test speed is 50m/h, the picture resolution is 6000 x 4000 x3, the test frame number is 20 frames per second, and fixing video acquisition equipment on a front bar of the vehicle, wherein the direction of the video acquisition equipment faces downwards to the road surface. The skid resistance of the calibration road surface is represented by the BPN measured by the pendulum instrument.
(2) Texture feature extraction based on local binary feature and mixed Gaussian distribution
And importing the measurement picture by using an LBP calculation model through Matlab software, converting the measurement picture into a gray image, performing LBP calculation, and drawing a histogram. Storing the histogram data into SQL data, and extracting characteristic parameters by adopting a Gaussian mixture distribution model, wherein the result is shown in table 1. And inputting the standard BPN and the calculated characteristic parameters into a random forest for calculation, and estimating a prediction error by using 10-fold cross validation, wherein the error result is shown in FIG. 6.
TABLE 1
(3) Antiskid performance estimation based on VGGNet migration learning model
The video capture pictures were cut to 150 × 3 size and put into the model for sample training to obtain prediction errors, as shown in table 2.
TABLE 2
| Method of producing a composite material | Recall ratio of | Predicting positive rate | Accuracy of |
| Convolutional neural network | 0.6485 | 0.7616 | 0.6514 |
| VGGNet network | 0.8030 | 0.8863 | 0.8114 |
(4) Model comparison and actual testing
According to the model algorithm, the texture recognition accuracy rate is 86% and the migration learning accuracy rate is 81% under the existing sample condition, therefore, the LBP texture recognition model is adopted to carry out actual test, the prediction results in 5 road sections of the actual test are consistent with the actual anti-skid performance, and the method can be used for rapid anti-skid performance detection.
While the invention has been described with reference to specific embodiments, the invention is not limited thereto, and various equivalent modifications and substitutions can be easily made by those skilled in the art within the technical scope of the invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (5)
1. A road surface anti-skid performance quick detection method based on machine vision is characterized in that the method is used for quickly detecting the road surface anti-skid performance based on the machine vision, namely texture information is extracted from a road surface picture, components with obvious anti-skid effect on the road surface are screened out, the influence of noise textures and other apparent characteristics is eliminated, a correlation model of picture texture parameters and road surface anti-skid indexes is established, and the anti-skid performance estimation based on image recognition is realized;
the method comprises the following steps: the method comprises the steps of carrying out quantitative characterization on local differences of each pixel of an image by a local binary feature extraction method, counting binary feature distribution of all pixels, extracting main fluctuation features by utilizing mixed Gaussian distribution, realizing noise filtering, carrying out model training by utilizing three feature vectors of extracted normal distribution weight, average value and variance, and establishing a correlation between normal distribution parameters and pavement skid resistance by adopting a random forest;
meanwhile, the data and the anti-skid index of actual measurement are added on the basis of the existing image decomposition depth network by adopting a transfer learning method.
2. The method of claim 1, wherein in order to avoid the influence of the sample size on the detection result, two methods are respectively adopted to compare the measurement errors in actual operation, and a model with smaller error is selected as the prediction model.
3. The method according to claim 1, characterized in that it comprises in particular the steps of:
(1) selecting high-definition video acquisition equipment, GPS equipment, a test vehicle and a shooting frame number;
(2) arranging the high-definition video acquisition equipment in front of a test vehicle, wherein a lens of the video equipment faces the road surface;
(3) the method comprises the steps that uniform-speed driving is conducted on a road section to be measured, high-definition video pictures of a road surface are collected in real time through high-definition video collection equipment, position information is collected through GPS equipment, data matching of an image and a GPS is conducted through a timestamp, and measured data are stored in a memory card of the high-definition video collection equipment;
(4) extracting local binary characteristics (LBP) of the collected picture, converting the original picture into a gray mode, comparing the gray difference between any pixel point and 8 surrounding pixel points, and marking as 1 if the surrounding gray is greater than or equal to the gray of a central pixel, otherwise, marking as 0;
(5) forming a binary sequence by starting from a 12 o' clock position in a clockwise direction according to the obtained binary value, and converting the binary sequence into a decimal number;
(6) carrying out the LBP extraction method on all pixel points in the picture, counting the number of decimal digits, and drawing a histogram;
(7) performing mixed Gaussian distribution GMM conversion on the histogram, converting the original graphic representation into two normal distribution superposition results, and counting the weight, the mean value and the variance of the two normal distributions;
(8) establishing a road surface skid resistance and normal distribution parameter model by adopting a random forest algorithm, and calculating an average prediction error;
(9) cutting an original image, converting the original image into a plurality of images, carrying out sample training by adopting a transfer learning model to obtain a correlation model of the road surface anti-skid performance and the image fragments, and calculating an average prediction error;
(10) and (3) under the condition of different sample sizes, respectively carrying out the algorithm and the model in the steps 8 and 9 to carry out prediction error calculation, selecting the model with smaller prediction error as a detection model, and carrying out pavement skid resistance detection.
4. The method of claim 3, wherein the high definition video capture device is bolted to the test vehicle.
5. The method according to claim 3, wherein the converted image of step (9) is an image of 150 x3 pixels.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810548514.9A CN108760740B (en) | 2018-05-31 | 2018-05-31 | Quick detection method for road surface skid resistance based on machine vision |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810548514.9A CN108760740B (en) | 2018-05-31 | 2018-05-31 | Quick detection method for road surface skid resistance based on machine vision |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108760740A CN108760740A (en) | 2018-11-06 |
| CN108760740B true CN108760740B (en) | 2020-10-02 |
Family
ID=64001048
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810548514.9A Active CN108760740B (en) | 2018-05-31 | 2018-05-31 | Quick detection method for road surface skid resistance based on machine vision |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108760740B (en) |
Families Citing this family (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20190362510A1 (en) * | 2018-05-24 | 2019-11-28 | Lu Sun | Method and system for evaluating friction coefficient and skid resistence of a surface |
| CN109671077B (en) * | 2018-12-24 | 2020-12-08 | 上海智能交通有限公司 | Method and system for detecting anti-skid performance of asphalt pavement |
| CN110765909B (en) * | 2019-10-14 | 2023-06-02 | 同济大学 | Road surface estimation method based on vehicle-mounted camera auxiliary distributed driving electric vehicle |
| CN112098409B (en) * | 2020-09-17 | 2023-04-07 | 国网河南省电力公司濮阳供电公司 | Hydrophobicity live-line testing method for composite insulator of power transmission line |
| CN112818563B (en) * | 2021-02-25 | 2022-08-05 | 同济大学 | Pavement skid resistance evaluation method based on friction contact surface estimation |
| CN112927204B (en) * | 2021-02-25 | 2022-09-20 | 同济大学 | Pavement water seepage performance evaluation method based on key water seepage point identification |
| CN113434954B (en) * | 2021-06-15 | 2022-09-20 | 同济大学 | Calibration method of vibrating type pavement flatness test vehicle |
| CN115128074B (en) * | 2022-07-05 | 2024-02-20 | 山东高速集团有限公司创新研究院 | Construction method of asphalt pavement anti-skid performance prediction system and constructed system thereof |
| CN115825410B (en) * | 2022-11-22 | 2023-12-22 | 重庆大学 | Method for estimating action noise level of tire road surface based on road surface structure |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE102007017293A1 (en) * | 2007-04-12 | 2008-10-16 | Continental Automotive Gmbh | Evaluation device for use in motor vehicle, has determination unit determining roadway limits using both image information captured by black-and-white camera and color camera which is aligned in rear side of vehicle |
| CN103114514A (en) * | 2013-01-31 | 2013-05-22 | 长安大学 | Grooved texture depth detection algorithm for cement concrete pavement |
-
2018
- 2018-05-31 CN CN201810548514.9A patent/CN108760740B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE102007017293A1 (en) * | 2007-04-12 | 2008-10-16 | Continental Automotive Gmbh | Evaluation device for use in motor vehicle, has determination unit determining roadway limits using both image information captured by black-and-white camera and color camera which is aligned in rear side of vehicle |
| CN103114514A (en) * | 2013-01-31 | 2013-05-22 | 长安大学 | Grooved texture depth detection algorithm for cement concrete pavement |
Non-Patent Citations (2)
| Title |
|---|
| Measuring skid resistance of hot mix asphalt using the aggregate image measurement system (AIMS);Victor M.C. Araujo等;《Construction and Building Materials》;20150828;全文 * |
| 基于数字图像技术的露石混凝土路面纹理构造抗滑性能;宋永朝等;《哈尔滨工业大学学报》;20150228;第47卷(第2期);第127页第2.2、4节 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108760740A (en) | 2018-11-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108760740B (en) | Quick detection method for road surface skid resistance based on machine vision | |
| CN111985494B (en) | Road disease detection and classification method based on lightweight vehicle-mounted terminal | |
| Lekshmipathy et al. | Vibration vs. vision: Best approach for automated pavement distress detection | |
| US10987977B2 (en) | Systems and methods for measuring tire wear using embedded image sensors | |
| US11208101B2 (en) | Device and method for measuring transverse distribution of wheel path | |
| CN101807345A (en) | Traffic jam judging method based on video detection technology | |
| CN112435463B (en) | Vehicle type and vehicle weight classification method based on road internet of things monitoring | |
| She et al. | Feasibility study of asphalt pavement pothole properties measurement using 3D line laser technology | |
| CN111783638A (en) | System and method for detecting number of vehicle axles and identifying vehicle type | |
| Abou Chacra et al. | Fully automated road defect detection using street view images | |
| CN111735524A (en) | Tire load obtaining method based on image recognition, vehicle weighing method and system | |
| CN112070756B (en) | Three-dimensional road surface disease measuring method based on unmanned aerial vehicle oblique photography | |
| Yun et al. | Crack recognition and segmentation using morphological image-processing techniques for flexible pavements | |
| CN110738636A (en) | Method and system for testing anti-sliding performance of asphalt mixture based on digital image technology | |
| CN115482474A (en) | Bridge deck vehicle load identification method and system based on high-altitude aerial image | |
| Shtayat et al. | Using supervised machine learning algorithms in pavement degradation monitoring | |
| Zhang et al. | Winter road surface condition classification using convolutional neural network (CNN): Visible light and thermal image fusion | |
| CN114971166A (en) | Lightweight road pavement service quality inspection system | |
| Elunai et al. | Asphalt concrete surfaces macrotexture determination from still images | |
| CN112085725A (en) | Residual film residual quantity detection method and early warning system based on heuristic iterative algorithm | |
| CN117011756A (en) | Video rainfall inversion method based on migration learning method | |
| Li et al. | Road Damage Evaluation via Stereo Camera and Deep Learning Neural Network | |
| CN116805409A (en) | Method for identifying road surface state and evaluating flatness by using driving video | |
| KY et al. | Road pothole detection using yolov2 model | |
| CN115272930A (en) | Ground penetrating radar-based road surface state evaluation method |
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 |