CN115482474A

CN115482474A - Bridge deck vehicle load identification method and system based on high-altitude aerial image

Info

Publication number: CN115482474A
Application number: CN202211021016.1A
Authority: CN
Inventors: 沈明燕; 党浩鹏; 舒小娟; 李贞贤; 孙华
Original assignee: Hunan University of Science and Technology
Current assignee: Hunan University of Science and Technology
Priority date: 2022-08-24
Filing date: 2022-08-24
Publication date: 2022-12-16
Anticipated expiration: 2042-08-24
Also published as: CN115482474B

Abstract

The invention provides a bridge deck vehicle load identification method and system based on high-altitude aerial images, and belongs to the field of bridge deck operation safety. The method comprises the steps of collecting a bridge deck vehicle image shot at high altitude, and constructing and calibrating a vehicle type and vehicle weight comparison database; after the bridge deck vehicle image is preprocessed, vehicle shadow filtering is carried out, and the filtered image is input into a vehicle identification model; identifying the vehicle through the vehicle identification model, and outputting the identified vehicle parameters; the vehicle parameters include: vehicle type, vehicle direction, vehicle number and vehicle distance; and obtaining the load distribution of the bridge deck vehicles by a statistical method according to the constructed vehicle comparison database and the identified vehicle parameters. The invention not only reduces the calculation amount of bridge deck vehicle load identification, but also improves the identification efficiency and the identification accuracy, and is suitable for instantaneous/long-term vehicle load monitoring of various bridges.

Description

Bridge deck vehicle load identification method and system based on high-altitude aerial image

Technical Field

The invention belongs to the field of bridge operation safety, and particularly relates to a bridge deck vehicle load identification method and system based on high-altitude aerial images.

Background

The vehicle load distribution of the bridge deck is an important influence factor for evaluating the safety level of the bridge during operation, and is also an important basis for bridge design, state evaluation, maintenance and reinforcement.

In the prior art, the bridge deck vehicle load statistical analysis is mainly performed in a manual statistical mode and a video acquisition mode in which a camera is installed on a monitoring rod. The method based on manual statistics has the defects of long time consumption, large workload, low working efficiency and the like; the traditional video acquisition mode is that three-dimensional data of vehicles are acquired by installing a plurality of groups of cameras on a monitoring rod or a bridge upright rod, due to the limitation of the height of a traffic auxiliary facility, the observation visual field of the cameras in the bridge length direction is limited, the overall distribution condition of bridge deck vehicles at a certain moment is difficult to acquire, the vehicle weight information of the vehicles cannot be acquired, and the distribution condition of the bridge deck vehicles and the parameters such as the speed of the vehicles cannot be accurately measured. In addition, the traffic monitoring task is usually completed by multiple groups of cameras in a coordinated manner, and once one of the cameras fails, vehicle information may be lost, and the statistical result of traffic volume may be affected.

Disclosure of Invention

In view of this, embodiments of the present invention provide a bridge deck vehicle load identification method and system based on a high-altitude aerial image, which reduce the amount of computation and improve the identification efficiency and the identification accuracy.

In order to achieve the above purpose, the embodiment of the invention adopts the following technical scheme:

in a first aspect, an embodiment of the present invention provides a bridge deck vehicle load identification method based on a high-altitude aerial image, including the following steps:

a bridge deck vehicle load identification method based on high-altitude aerial images is characterized by comprising the following steps:

constructing a vehicle type and vehicle weight comparison database;

collecting a bridge deck vehicle image shot at high altitude;

preprocessing the received bridge deck vehicle image, and then filtering vehicle shadows to obtain a clean bridge deck vehicle image;

constructing a vehicle identification model, inputting the clean bridge deck vehicle image into the vehicle identification model to identify bridge deck vehicles, and outputting the identified vehicle parameters; the vehicle parameters include: vehicle type, vehicle direction, vehicle number and vehicle distance;

and obtaining the load distribution of the bridge deck vehicles by a statistical method according to the constructed vehicle comparison database and the identified vehicle parameters.

As a preferred embodiment of the invention, the acquisition of the aerial image of the bridge deck vehicle comprises

S21, arranging an unmanned aerial vehicle above the bridge floor, wherein the unmanned aerial vehicle carries a camera device through a holder, and the visual field range of the camera device is the whole bridge floor;

and S22, adjusting flight parameters to enable the camera to shoot the bridge floor, collecting each vehicle image of the whole bridge floor, and transmitting the bridge floor vehicle image to an upper computer.

As a preferred embodiment of the invention, the height of the overhead photography is 50-200 m.

As a preferred embodiment of the present invention, the vehicle shadow filtering includes:

step S31, obtaining a three-color channel red R, green G and blue B brightness value matrix of the image according to the bridge deck vehicle image, wherein the m-th line number and the three-color channel brightness value of the n-th row of pixel points are represented as R _(m,n) ,G _(m,n) ,B _(m,n) ；

Step S32, counting the brightness values of three-color channels of the bridge floor, the vehicles with various colors and the shadow parts of the vehicles in the aerial photography picture to obtain the maximum value and the minimum value as representative values, wherein the shadow parts are expressed as

Step S33, judging whether each pixel point in the image simultaneously satisfies

If the two conditions are met, judging the shadow;

step S34, the brightness value (R) of the three-color channel of the pixel point of the shadow part _shadow ,G _shadow ,B _shadow ) Replacing the bridge deck three-color channel brightness value or brightness mean value

The shadow filtering is done.

As a preferred embodiment of the invention, the vehicle identification model is constructed by adopting a YOLO-V3 network structure.

As a preferred embodiment of the present invention, the YOLO-V3 network structure has no pooling layer, and outputs 3 different scales of feature maps in terms of output tensor.

As a preferred embodiment of the present invention, the 3 feature maps with different scales are implemented by dividing the original image with 3 different grids, including 16 × 16 grids for large objects, 26 × 26 grids for medium objects, and 52 × 52 grids for fine objects.

As a preferred embodiment of the present invention, the identifying the bridge deck vehicles comprises the following steps:

step S411, extracting first features based on the appearance and the outline of the vehicle, primarily classifying the vehicle in the image based on the first features, adopting the vehicle type with obvious features and small quantity, identifying and reading the position of the vehicle in the short time difference, and judging the advancing direction of the traffic flow according to the position of the vehicle in the picture;

step S412, determining the width of lanes by two vehicles with the farthest distance according to the advancing direction of each traffic flow, correspondingly dividing different lanes according to the width of the lanes, dividing each vehicle into respective lanes, and identifying the inter-vehicle distance in each lane;

step S413, extracting a second feature based on the aspect ratio of the vehicle, identifying the vehicle type of each vehicle in the image, and calculating the number of vehicles of each vehicle type.

As a preferred embodiment of the present invention, the inter-vehicle distance parameter is based on the principle of pinhole imaging, and the size V of the image of the camera is related to the size of the object U, the size Φ of the pinhole, the distance W from the object to the camera, and the distance X from the image to the camera, and is related as follows:

and the vehicle identification model obtains the inter-vehicle distance in the images according to the principle and the received bridge deck vehicle images.

In a second aspect, an embodiment of the present invention further provides a bridge deck vehicle load identification system based on a high-altitude aerial image, where the system includes: the vehicle comparison database module, the image acquisition module, the image processing module, the vehicle identification module and the vehicle data analysis module; wherein the content of the first and second substances,

the image acquisition module is used for acquiring an image of the bridge deck vehicle shot by plane at high altitude;

the vehicle comparison database module is used for constructing a vehicle type and vehicle weight comparison database;

the image processing module is used for preprocessing the received bridge deck vehicle image and then filtering vehicle shadows to obtain a clean bridge deck vehicle image;

the vehicle identification module is used for constructing a vehicle identification model, inputting the clean bridge deck vehicle image into the vehicle identification model to identify the bridge deck vehicle, and outputting the identified vehicle parameters; the vehicle parameters include: vehicle type, vehicle direction, vehicle number and vehicle distance;

and the vehicle data analysis module is used for obtaining the bridge deck vehicle load distribution through a statistical method according to the constructed vehicle comparison database and the identified vehicle parameters.

The technical scheme provided by the embodiment of the invention has the following beneficial effects:

according to the bridge deck vehicle load identification method and system based on the high-altitude aerial photography image, shadow filtering is firstly carried out on the acquired high-altitude aerial photography bridge deck vehicle image, vehicle identification errors caused by vehicle shadows are eliminated on the premise that the height of the vehicle is ignored, calculated amount is reduced, identification efficiency and identification accuracy are improved, and the method and system are suitable for instantaneous/long-term vehicle load monitoring of various bridges. Compared with a fixed camera, the vehicle load identification method based on the high-altitude aerial images has higher flexibility, and the distribution condition of all vehicles on the whole bridge floor at a certain moment can be observed.

Of course, it is not necessary for any product or method to achieve all of the above-described advantages at the same time for practicing the invention.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.

FIG. 1 is a flow chart of a bridge deck vehicle load identification method based on high-altitude aerial images according to an embodiment of the invention;

FIG. 2 is a schematic structural diagram of a bridge deck vehicle load identification system based on high-altitude aerial images according to an embodiment of the invention;

FIG. 3 is a RGB value statistical chart of the shadow area in the embodiment of the present invention;

FIG. 4 is a statistical chart of RGB values of the bridge floor background area in the embodiment of the present invention;

FIG. 5 is a result of vehicle type recognition training in an embodiment of the present invention;

FIG. 6 is a chart illustrating statistics of the number of cars in the first period of time according to the embodiment of the present invention;

FIG. 7 is a chart illustrating the statistics of the number of vehicles in the second time period according to the embodiment of the present invention;

FIG. 8 is a chart showing statistics of the number of cars in the third time period according to the embodiment of the present invention;

FIG. 9 is a comparison of vehicle separation for different vehicle types in an embodiment of the present invention;

Detailed Description

After finding the above problems, the inventors of the present application have conducted intensive studies on a statistical method for vehicle load distribution on a bridge deck in the prior art. Researches show that the machine vision statistical method has many advantages far exceeding the manual statistical method, and with the rapid development and application of machine vision technology, many traffic identification achievements based on image processing technology exist, but bridge deck camera shooting-based images cannot meet the requirements of bridge deck vehicle load analysis.

Because the unmanned aerial vehicle has advantages such as small and mobility is strong, can overcome the defect of traditional image acquisition system, more and more are applied to intelligent transportation system. The main task of traffic monitoring is the identification of the number of vehicles, the height information of the vehicles can be ignored, the two-dimensional characteristic information of the vehicles obtained through aerial images can meet the task requirement of traffic monitoring, the calculation amount is greatly reduced, and the working efficiency of a vehicle identification model is improved. However, in the aerial image collected at high altitude under natural lighting conditions, shadow images of a part of vehicles are stuck with the vehicle images, and vehicle shadows may be recognized as vehicles, so that the accuracy of model recognition is reduced, and the error of traffic statistics becomes large.

It should be noted that the above prior art solutions have defects which are the results of practical and careful study by the inventors, and therefore, the discovery process of the above problems and the solutions proposed by the following embodiments of the present invention to the above problems should be the contribution of the inventors to the present invention in the course of the present invention.

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. It is to be understood that the described embodiments are merely a few embodiments of the invention, and not all embodiments. The components of embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations. It should be noted that the embodiments and features of the embodiments of the present invention may be combined with each other without conflict.

It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, it need not be further defined or explained in subsequent figures. In the description of the present invention, the terms "first," "second," "third," "fourth," etc. are used merely to distinguish one description from another, and are not intended to indicate or imply relative importance.

After the deep analysis, the embodiment of the invention provides a bridge deck vehicle load identification method and a bridge deck vehicle load identification system based on high-altitude aerial images, the height of a vehicle is ignored based on the collected aerial top view images, the calculated amount is reduced on the premise of meeting the identification requirement, the problem of limited visual field height in the traditional identification means is solved, and the identification efficiency is improved; in the identification process, the difference between the vehicle shadow and the vehicle and bridge floor background pigment values is utilized to eliminate the error of vehicle shadow identification, realize the rapid identification of bridge floor vehicle load, effectively improve the detection and identification accuracy, and be suitable for the instantaneous/long-term vehicle load monitoring of various bridges.

Referring to fig. 1, the bridge deck vehicle load identification method based on the high-altitude aerial image provided by the embodiment of the invention comprises the following steps:

s1, constructing a vehicle type and vehicle weight comparison database.

In this step, a vehicle type and vehicle weight comparison database is constructed, and vehicle types and vehicle weights are paired one by one. The vehicle types are classified according to the length-width ratio of the vehicles. Preferably, the vehicle model and vehicle weight comparison database contains seven types of vehicle models, the vehicle models are shown in table 1, and the vehicle axle weight distribution is shown in table 2.

TABLE 1

TABLE 2

Vehicle model code	Number of axes	Gross weight/KN	Axle weight/kN	Wheelbase/mm
					V1	2	19	9+10	2749
V2	2	60	21+39	4320
					V3	2	130	60+70	4000
V4	2	140	65+75	4300
					V5	2	160	55+105	4000
V6	3	250	50+100+100	4100+1360
					V7	4	310	45+95+85+85	1860+3560+1350
V8	5	420	40+125+85+85+85	3600+5010+1310+1310
					V9	6	490	30+95+95+90+90+90	3270+1350+6230+1310+1310

And S2, acquiring a bridge deck vehicle image shot at high altitude.

In this step, it specifically includes:

and S21, arranging an unmanned aerial vehicle above the bridge floor, wherein the unmanned aerial vehicle carries a camera device through a holder, and the visual field range of the camera device is the whole bridge floor.

In this step, cloud platform and the camera device that unmanned aerial vehicle carried can bow to whole bridge floor and take a photograph. Through the position and the camera device's of adjusting unmanned aerial vehicle angle simultaneously, no matter which position that unmanned aerial vehicle is in the overhead bridge floor, all can realize the shooting to whole bridge floor.

In the step, the camera device can ignore the details of the vehicle, but needs to completely shoot the contour line of the vehicle; meanwhile, real-time vehicle shadows need to be acquired simultaneously. The adjustment flight parameters are expressed by the ratio of the length of the vehicle to the flight height, and clear images containing full-bridge vehicles are obtained by adjusting the flight parameters. The height of the overhead camera is 50-200 m. The high-altitude shooting through the unmanned aerial vehicle ignores the influence of vehicle height, only contains the length and the width information of vehicle, can reduce the calculated amount under the prerequisite that satisfies the detection requirement, improves detection efficiency.

And S3, preprocessing the received bridge deck vehicle image, and then carrying out vehicle shadow filtering to obtain a clean bridge deck vehicle image.

In this step, the vehicle shadow filtering includes two steps of shadow detection and elimination, and the vehicle shadow detection is realized by the difference between the RGB values of the bridge floor, the vehicle and the vehicle shadow. Through vehicle shadow filtering, the influence of the vehicle shadow on the identification result is reduced, and the vehicle identification accuracy is improved. The method specifically comprises the following steps:

step S31, obtaining a three-color channel red (R), green (G) and blue (B) brightness value matrix of the image according to the bridge deck vehicle image, wherein the mth line number and the nth row of pixel pointsIs expressed as R _(m,n) ,G _(m,n) ,B _(m,n) ；

Step S32, counting the brightness values of three-color channels of the bridge floor, the vehicles with various colors and the shadow parts of the vehicles in the aerial photography picture to obtain the maximum value and the minimum value as representative values, wherein the shadow parts are represented as

In this step, the brightness values of the three-color channels of the vehicles of each color are set with thresholds, and the thresholds are listed in table 3.

TABLE 3

If the two are satisfied, the shadow is determined.

The shadow filtering is done.

S4, constructing a vehicle identification model, inputting the clean bridge deck vehicle image into the vehicle identification model to identify bridge deck vehicles, and outputting the identified vehicle parameters; the vehicle parameters include: the vehicle type, the vehicle direction, the vehicle number and the vehicle distance.

In the step, the vehicle identification model is constructed by adopting a YOLO-V3 network structure. Preferably, the YOLO-V3 network structure has no pooling layer, and the YOLO-V3 network structure outputs 3 feature maps of different scales, y respectively, in terms of output tensor ₁ 、y ₂ And y ₃ . In the embodiment, multiple scales are adopted to detect different targets, and the finer the grid unit is, the finer the object can be detected. Preferably, in the present embodiment, 3 different grids are used to implement the division of the original image, 16 × 16 is for large objects, 26 × 26 is for medium objects, 52 × 52 is the finest of the three grids, and is for small objects.

The YOLO model loss function is determined by respective characteristics, and end-to-end training is achieved through one loss function.

The vehicle type and vehicle direction parameters are used for firstly identifying the traffic flow advancing direction of the bridge deck vehicle image, the vehicle identification model analyzes the length-width ratio of the vehicle, and therefore the purpose of identifying the traffic flow advancing direction is achieved, and the problem that the identification effect cannot meet the requirement due to the fact that the images are different in direction is solved. The method comprises the steps of determining the different lanes and the single lane by vehicles with different distances, determining the width of the lanes by two vehicles with the farthest distances, and correspondingly dividing the different lanes according to the lane widths, so that the purpose of dividing each vehicle into respective lanes is achieved. The vehicle type parameter includes an aspect ratio (L/B) of the vehicle. And distinguishing different vehicle types after training according to the length-width ratio of each vehicle in the picture for the types of the vehicles in the image.

Specifically, the identification of the bridge deck vehicles includes the following steps:

step S411, extracting first features based on the appearance and the outline of the vehicle, primarily classifying the vehicle in the image based on the first features, recognizing and reading the position of the vehicle in the short time difference by adopting the vehicle types with obvious features and small quantity, and judging the traffic flow advancing direction according to the position of the vehicle in the picture;

step S412, determining the width of a lane by two vehicles with the farthest distance according to the advancing direction of each traffic flow, correspondingly dividing different lanes according to the width of the lane, dividing each vehicle into respective lanes, and identifying the inter-vehicle distance in each lane, wherein the identified parameters can also include the vehicle speed;

The vehicle quantity parameter, YOLO, adopts a deep learning detection algorithm based on a regression method to identify the number of vehicles in the image. The vehicle number identification is to call a test-detector function in a detector, modify batch processing, count the number of detection targets returned to each picture, increase a return value, and finally recompile the darknet, so that the tested image can display the classification and number of the detected objects. The YOLO algorithm theme idea is that an excellent classifier in a network structure is utilized to obtain a target object range in a picture, and then an upsampling and loss function is continuously iterated to enable a target to be accurate to a certain position. The vehicle number detection is simple, and counting is completed according to the position relation between each judgment target and the detection line.

The inter-vehicle distance parameter is based on the principle of pinhole imaging, the size of the image (V) of the camera is related to the size of the object (U), the size (phi) of the pinhole, the distance (W) between the object and the camera and the distance (X) between the image and the camera, and the relationship is as follows:

In the embodiment, an image set with fixed interval time is adopted, a certain vehicle is fixed in the image set, the image set is shot with the same time interval and the same shooting height, the shot images are identified, the vehicle center is calibrated to be a characteristic point through the positions of the two images, and the vehicle speed of the vehicle in driving is obtained through the position difference value of the characteristic point and the time interval.

And S5, obtaining the load distribution of the bridge deck vehicles by a statistical method according to the constructed vehicle comparison database and the identified vehicle parameters.

In the step, the identified vehicle parameters are matched with a calibrated vehicle type and vehicle weight comparison database, the identified vehicles are converted into corresponding vehicle loads, and the distribution condition of the bridge deck vehicle loads is obtained by combining vehicle characteristic information such as vehicle number, vehicle distance and the like through a statistical method.

Based on the same idea, an embodiment of the present invention further provides a bridge deck vehicle load identification system based on a high-altitude aerial image, as shown in fig. 2, the system includes: the vehicle comparison system comprises a vehicle comparison database module, an image acquisition module, an image processing module, a vehicle identification module and a vehicle data analysis module.

The image acquisition module is used for acquiring a bridge deck vehicle image shot at high altitude. Preferably, the image acquisition module comprises: the system comprises an unmanned aerial vehicle, a cloud deck and a camera device;

the image processing module is used for preprocessing the received bridge deck vehicle image, then carrying out vehicle shadow filtering to obtain a clean bridge deck vehicle image and sending the clean bridge deck vehicle image to the vehicle identification module;

In the embodiment, each module is realized by a processor, and when the storage is needed, the storage is added appropriately. The Processor may be, but is not limited to, a microprocessor MPU, a Central Processing Unit (CPU), a Network Processor (NP), a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA), other programmable logic devices, discrete gates, transistor logic devices, discrete hardware components, and the like. The Memory may include a Random Access Memory (RAM) or a Non-Volatile Memory (NVM), such as at least one disk Memory. Alternatively, the memory may be at least one memory device located remotely from the processor.

It should be noted that, the bridge deck vehicle load identification system and method based on the high-altitude aerial photography image described in this embodiment correspond to each other, and the description and limitation of the system are also applicable to the method, and are not described herein again.

In order to facilitate understanding of the embodiments of the present invention, the following description will be further explained by taking specific examples as examples with reference to the drawings, and the examples do not limit the technical solutions of the present invention.

And taking a Hunan Tan four-bridge as an engineering background, acquiring related image information by using an unmanned aerial vehicle for the Hunan Tan four-bridge and analyzing the load distribution of the bridge deck vehicles. The method comprises the following steps:

step S1, a vehicle type and vehicle weight comparison database is constructed, and the data in table 1 and table 2 are adopted in this embodiment.

And S2, arranging an unmanned aerial vehicle over the Hunan Tan four-bridge, wherein the unmanned aerial vehicle carries a camera through a cradle head. The camera and pan/tilt parameters are shown in tables 4 and 5, respectively.

TABLE 4

TABLE 5

Adjusting flight parameters to lift the unmanned aircraft to a flight height of 200m, and taking aerial images of the bridge surface of.

In this embodiment, the images of the flying heights of 50m, 100m, 150m, 200m and 300m were respectively tested, and the recognition rate at each flying height is shown in table 6.

TABLE 6

Flying height/m	Recognition rate
			50	99％
100	97％
		150	90％
200	89％
		300	5％

Analysis of the recognition results of the images taken at different flying heights of 50m, 100m, 150m, 200m and 300m revealed that the taken image recognition was substantially ineffective when the flying height reached 300 m. The recognition effect of the images shot below 300m can basically meet the research requirement, the lowest accuracy rate also reaches 89%, and the model also has good robustness. In consideration of the overall research of the bridge and the reasons of the image width and the model identification accuracy, the images are acquired at the height of 200m through comparison of the images shot at different flight heights.

And S3, preprocessing the received bridge deck vehicle image, performing vehicle shadow filtering, selecting 35 shadow areas in the shadow image as shadow RGB value samples, randomly selecting 3 points in each area as samples, taking the average value of the 3 points in each area as the representative value of the area, and performing statistics, wherein the statistical result is shown in FIG 3. Similarly, statistical analysis of the bridge deck RGB values was performed, the results of which are shown in FIG. 4.

As can be seen from FIG. 3, the R value of the shadow area of the vehicle on the bridge deck ranges from 80 to 100, the G value ranges from 105 to 125, and the B value ranges from 125 to 145. The results of FIG. 4 show that the R value, G value and B value of the bridge floor background area fluctuate in the range of 160-185 under different brightness conditions, the R value, G value and B value of the bridge floor background area are respectively averaged to obtain the average value of the R value is 169, the average value of the G value is 172 and the average value of the B value is 168.

And (3) carrying out point taking identification on common vehicles with different colors, wherein the value intervals of the R value, the G value and the B value of the vehicles with various colors are shown in a table 7.

As can be seen from the statistical results in fig. 3-4 and table 7, the chromaticity value of the vehicle shadow at the position close to the vehicle is smaller compared to the position of the vehicle, and the image is displayed in a darker color, whereas the lighter color is, i.e., the gradient of the shadow changes. According to the shadow gradient principle, the shadow detection is to detect the R value, the G value and the B value in a region, and the R value, the G value and the B value in a certain region in an image can only be in accordance with the value range of the R value, the G value and the B value of a shadow part, so that the shadow can be identified. A corresponding MATLAB program is programmed to realize the process of identifying and replacing the shadow, and the pseudo code for identifying and replacing the shadow is shown in a table 7.

TABLE 7

S4, constructing a vehicle identification model, inputting the clean bridge deck vehicle image into the vehicle identification model to identify bridge deck vehicles, and outputting the identified vehicle parameters; the vehicle parameters include: vehicle type, vehicle direction, vehicle number and vehicle distance.

Wherein, the vehicle type discernment is as follows:

the acquired aerial images are input as the input end of the recognition model, the bridge deck vehicle model recognition is carried out by utilizing a YOLO V3 network, and the vehicle model recognition training effect is shown in the attached figure 5.

The vehicle number is identified as follows:

the single cruising time of the unmanned aerial vehicle is 20 minutes, the flight time is limited, the following images collected in several time periods of 9.

As can be seen from fig. 6-8, for each time interval vehicle number statistics, compared with the manual statistics-based method, the vehicle number statistics based on the unmanned aerial vehicle are all smaller than the value based on the manual statistics, the minimum error is 2.96%, the maximum error is 5.11%, but the total vehicle number statistics error is 3.86%, which is within the error tolerance range.

The inter-vehicle distance recognition is as follows:

the analysis is carried out on the vehicle data within 200m of the bridge length, the acquired inter-vehicle distance information is identified by using a YOLO model as a research data set, the acquired data is data of 5-day acquisition period, 3000 images are taken as research samples for the acquired data, and 600 images are respectively taken for each time period for detection, wherein the time periods are. The averaging process for the inter-vehicle distance in the image is performed according to equation (1).

Average pitch:

in the formula, L is a collection bridge length area; Σ n is the total number of vehicles;

is the average pitch.

For Hunan Tan four-bridge, the method is used for obtaining the number of vehicles, vehicle types and vehicle distance data among different lanes in each time period, and counting the total data. The statistical analysis of the inter-vehicle distance information in 3000 acquired images, the average inter-vehicle distance in a single image is obtained through the inter-vehicle distance of a single image, the average inter-vehicle distance in a single image is obtained, 3000 images in a sample are subjected to data processing by adopting the same method, 3000 groups of data in each lane are subjected to mean processing, the 95% confidence coefficient is ensured to obtain the mean value of the inter-vehicle distance, and the result is shown in figure 9.

Vehicle speed identification is as follows:

getting 7 in the morning in the time period of vehicle speed acquisition at the peak time of Hunan Tan four-axle: 30 to 9:30 hours, 5 nights: 30 to 7: and in the period of 30, processing the acquired images to establish a speed identification data set, listing the speed identification data set in a table 8, and identifying the 200m flying height and the images in a fixed time period through an image identification model to further obtain the speed per hour of the same vehicle.

TABLE 8

Vehicle model	Vehicle with wheels	Two-type vehicle	Three-type vehicle	Four-wheel vehicle	Five-type vehicle	Six-wheeled vehicle	Seven-type vehicle
								Vehicle number/vehicle	2368	97	261	71	24	4	94
Average vehicle speed km/h	79.5	66.5	70.2	63.4	70.6	68.3	69.4

Step S5, vehicle load distribution statistics: since the traffic jam operating state has the least adverse effect on the bridge structure, the subsequent analysis is the traffic flow in the dense operating state. The method comprises the following steps of (1) adopting a regression analysis method for automobile load research of a bridge, identifying the vehicle types, the vehicle numbers and the vehicle distances of various vehicles in an image at the moment through a model, comparing the vehicle types with vehicle weights, analyzing and researching the vehicle weight information of various vehicles in a database, establishing a sample with the capacity of 1500 images by collecting images at different time periods, and researching load calculation by adopting a method for calculating load concentration, wherein the main formula is as follows:

the proportion of each vehicle type can be calculated according to the following formula:

the load concentration calculation formula is as follows:

in formulae (2) and (3): t is t _i The vehicle weight average value of each vehicle type is obtained; l is the image acquisition length; z is the number of samples; and m is the total number of vehicles.

The vehicle weight and the distribution of each vehicle type are fitted, and the fitting is in a normal distribution form, so that the distribution function of the vehicle weight and the distribution of each vehicle type is as follows:

and (4) carrying out lane-dividing statistics on 1500 samples to obtain vehicle type and vehicle number data of each lane, and calculating the vehicle type ratio of each bridge by taking the vehicle type and vehicle number data as the basis. The method is characterized in that a sample with relatively small sample capacity is established by calculating the load concentration of each bridge, the purpose is mainly to select the dense state of the bridge during the operation period for analysis, the selection of small samples needs to follow the principle of more vehicles, and the working conditions with different acquisition lengths are analyzed, so that the load concentration of the bridge during the operation period is obtained.

When the four-axle concentration is calculated, the four-axle traffic flow forms a plurality of vehicle types including one-type vehicles to seven-type vehicles, so after the vehicle weight data are referred, the concentration of the one-type vehicles to the four-type vehicles is calculated by adopting uniformly distributed loads, the five-type vehicles to the seven-type vehicles with equal weight are calculated as concentrated loads, the conditions of small acquisition length and dense traffic flow are adopted, and the dense traffic flow under the fixed acquisition length is selected for the concentration calculation. The calculated load density for each lane of the four-axle is shown in table 9.

TABLE 9

The heavy vehicle load research is carried out by taking a four-axle heavy vehicle as a basis, taking 1500 samples established for the four axles in the previous section as research objects, carrying out load calculation on the heavy vehicles under different acquisition lengths, carrying out statistics on the intensive traffic flows under different acquisition lengths and carrying out concentrated load conversion. The number of heavy vehicles and the vehicle types of each lane in 1500 samples were counted, as shown in table 10.

TABLE 10

The study on the heavy vehicle load is the traffic flow load effect in the intensive operation state, and the four-bridge heavy vehicle load is used as the concentrated force to carry out loading study in view of the fact that the four-bridge belongs to a large-span bridge. According to the method, the load of the heavy vehicle needs to be researched, the uniform load fitting of a single vehicle is carried out according to the wheel base and the wheel weight distribution of the heavy vehicle, so that the wheel base, the wheel weight and other data of each vehicle type need to be counted, and the wheel base and the wheel weight distribution of the heavy vehicle type are shown in a table 11.

TABLE 11

Vehicle model	Gross weight/KN	Wheelbase/mm	Axle weight/KN
				Five-type vehicle	321	1860+3560+1350	49+100+86+86
Six-wheeled vehicle	439	3600+5010+1310+1310	43+132+88+88+88
				Seven-type vehicle	495	3270+1350+6230+1310+1310	30+99+99+89+89+89

And (3) counting the heavy vehicle lanes in the sample obtained by counting, analyzing the heavy vehicle working condition in the dense state, and after the data are sorted, obtaining the most dense heavy vehicle flow image in the single lane, wherein the number of the heavy vehicles is shown in a table 12.

TABLE 12

The load values uniformly distributed on the single vehicle of each vehicle type are obtained by combining the wheel base and the wheel load data of each vehicle type, as shown in a table 13:

watch 13

According to the technical scheme, the bridge deck vehicle load identification method and the bridge deck vehicle load identification system based on the high-altitude aerial images, provided by the embodiment of the invention, firstly carry out shadow filtering on the acquired high-altitude aerial bridge deck vehicle images, eliminate errors of vehicle shadow identification on the premise of neglecting the height of the vehicle, reduce the calculated amount, improve the identification efficiency and the identification accuracy, and are suitable for instantaneous/long-term vehicle load monitoring of various bridges.

The above description is only for the specific embodiment of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention are included in the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A bridge deck vehicle load identification method based on high-altitude aerial images is characterized by comprising the following steps:

constructing a vehicle type and vehicle weight comparison database;

collecting a bridge deck vehicle image shot at high altitude;

2. The bridge deck vehicle load identification method based on high-altitude aerial images of claim 1, wherein the collecting of the high-altitude aerial bridge deck vehicle images comprises

3. The bridge deck vehicle load identification method based on the high-altitude aerial images as claimed in claim 2, wherein the height of the overhead shot is 50-200 m.

4. The bridge deck vehicle load identification method based on the high-altitude aerial image according to claim 1, wherein the vehicle shadow filtering comprises the following steps:

If the two shadow masks are satisfied at the same time, judging the shadow;

The shadow filtering is done.

5. The bridge deck vehicle load identification method based on the high-altitude aerial image as claimed in claim 1, wherein the vehicle identification model is constructed by adopting a YOLO-V3 network structure.

6. The bridge deck vehicle load identification method based on the high-altitude aerial image is characterized in that the YOLO-V3 network structure is not provided with a pooling layer, and 3 feature maps with different scales are output in terms of output tensors.

7. The bridge deck vehicle load identification method based on the high-altitude aerial images as claimed in claim 5, wherein the 3 feature maps with different scales are implemented by dividing the original image by using 3 different grids, including 16 × 16 grids for large objects, 26 × 26 grids for medium objects, and 52 × 52 grids for small objects.

8. The bridge deck vehicle load identification method based on the high-altitude aerial image as claimed in claim 5, wherein the step of identifying the bridge deck vehicle comprises the following steps:

9. The bridge deck vehicle load identification method based on the high-altitude aerial images as claimed in claim 8, wherein the inter-vehicle distance parameter is based on the principle of small hole imaging, and the size V of the image of the camera is related to the size of the object U, the size phi of the small hole, the distance W between the object and the camera, and the distance X between the image and the camera, and the relationship is as follows:

10. A bridge deck vehicle load identification system based on high altitude aerial images, the system comprising: the vehicle comparison database module, the image acquisition module, the image processing module, the vehicle identification module and the vehicle data analysis module; wherein the content of the first and second substances,

the image acquisition module is used for acquiring a bridge deck vehicle image shot at high altitude;

the vehicle identification module is used for constructing a vehicle identification model, inputting the clean bridge deck vehicle image into the vehicle identification model to identify the bridge deck vehicle and outputting the identified vehicle parameters; the vehicle parameters include: vehicle type, vehicle direction, vehicle number and vehicle distance;