Construction road crack prediction method based on time sequence monitoring data

Abstract

The invention relates to the technical field of road monitoring, and discloses a construction road crack prediction method and system based on time sequence monitoring data, and the method comprises the steps: obtaining historical monitoring time based on a road monitoring device, obtaining a plurality of unit time periods based on the historical monitoring time, obtaining a plurality of historical images based on the unit time periods, and obtaining a noise reduction image based on the historical images, obtaining unit impact data based on the plurality of noise reduction images, summarizing the unit impact data to obtain a unit impact data set, obtaining a noise reduction image sequence, obtaining an initial crack area sequence based on the noise reduction image sequence, screening the initial crack area sequence to obtain a to-be-interpolated sequence, and performing interpolation calculation on the to-be-interpolated sequence to obtain a crack area evolution sequence; and constructing a crack evolution model according to the crack area evolution sequence and the unit impact data set, and obtaining a predicted crack value based on the crack evolution model. According to the invention, the road crack can be predicted by using historical data.

Description

Technical Field

[0001] The present invention relates to the technical field of road monitoring, and in particular to a construction road crack prediction method, system, electronic device and computer-readable storage medium based on time series monitoring data. Background Art

[0002] With the development of the times, all parties and regions have deeply implemented the policy of rural revitalization. Correspondingly, expanding the scope of urbanization and building roads to every home have become the people's day and night expectations. Therefore, road construction has become a key development area.

[0003] Before carrying out road construction, it is often necessary to evaluate and predict the roads to be constructed so that the constructed roads can meet national standards and usage standards, avoid cracks in the roads after construction in a short period of time, or even rapidly develop cracks after they appear, affecting traffic. In the prior art, construction is generally carried out based on the experience of the road construction team and existing construction standards, and for road cracks, measures are generally taken to control cracks after cracks appear.

[0004] Although the above method can complete road construction and predict and maintain cracks that occur, for areas where road construction has never been carried out, such as rural roads and urban fringe roads, it is relatively subjective based on experience alone. There is a lack of a method that can predict future road conditions and the evolution of possible cracks based on the actual situation of the road to be constructed, thereby realizing the prediction of road cracks using historical data. Summary of the invention

[0005] The present invention provides a construction road crack prediction method based on time series monitoring data and a computer-readable storage medium, the main purpose of which is to predict road cracks using historical data.

[0006] To achieve the above object, the present invention provides a construction road crack prediction method based on time series monitoring data, comprising:

[0007] receiving a road monitoring instruction, starting a road monitoring device according to the road monitoring instruction, obtaining a historical monitoring time based on the started road monitoring device, and dividing the historical monitoring time to obtain a plurality of unit time periods;

[0008] Extracting unit time periods from the plurality of unit time periods in sequence, and performing the following operations on the extracted unit time periods: acquiring a plurality of historical images based on the extracted unit time periods, extracting historical images from the plurality of historical images in sequence, and performing the following operations on the extracted historical images:

[0009] Performing a clearing operation on the extracted historical image to obtain a clear image, performing a denoising operation on the clear image to obtain a denoised image, extracting a set of vehicle images to be identified based on the denoised image, and calculating the extracted unit impact data of the unit time period based on the set of vehicle images to be identified;

[0010] Summarize the unit impact data to obtain the unit impact data set corresponding to the historical monitoring time, summarize the denoised images to obtain the denoised image sequence corresponding to the historical monitoring time, use the pre-built edge recognition algorithm to identify the denoised image sequence to obtain the initial crack area sequence, and filter the initial crack area sequence to obtain the sequence to be interpolated;

[0011] The interpolation method is used to interpolate the sequence to be interpolated, and the crack area evolution sequence is obtained;

[0012] A crack evolution model is constructed based on the crack area evolution sequence and the unit impact data set, and the time period to be predicted and the impact to be predicted are obtained. The time period to be predicted and the impact to be predicted are input into the crack evolution model to obtain the predicted crack value. The prediction of construction road cracks in the unit time period monitoring data is completed based on the predicted crack value.

[0013] Optionally, performing a clearing operation on the extracted historical image to obtain a clear image includes:

[0014] Multiple Gaussian difference curves are generated using the pre-built Gaussian difference operator and the extracted historical images, where the Gaussian difference operator is:

[0015]

[0016] Among them, z represents the horizontal coordinate of the historical image in the image coordinate system, and w represents the vertical coordinate of the historical image in the image coordinate system. and Both represent the variance of Gaussian filter, σ1 and σ2 both represent the standard deviation of Gaussian filter, and σ1≠σ2, DOG represents the Gaussian difference operator;

[0017] The zero-crossing point of each Gaussian difference curve in a plurality of Gaussian difference curves is obtained to obtain a plurality of zero-crossing points, a plurality of contour coordinates are generated according to the plurality of zero-crossing points, the plurality of contour coordinates are spliced ​​using a pre-built edge connection algorithm to obtain a historical image contour, an image transition area is obtained according to the historical image contour, and a clear image is obtained according to the image transition area, wherein the contour coordinates correspond one-to-one to the Gaussian difference curve.

[0018] Optionally, performing a noise reduction operation on the clear image to obtain a noise-reduced image includes:

[0019] Construct a mean filter unit and a median filter unit;

[0020] A standard image is obtained, and the spatial distribution characteristics of the pixel grayscale values ​​are extracted from the standard image. The spatial distribution characteristics of the pixel grayscale values ​​are introduced into a mean filter unit to obtain a fixed pattern noise, wherein the size of the filter kernel of the mean filter unit is 9×9;

[0021] A denoising operation is performed on the clear image using fixed pattern noise to obtain a denoised image, wherein the formula for a denoising operation is as follows:

[0022] k(x,y)=j(x,y)-δ×l(x,y)+β×h

[0023] Wherein, k(x, y) represents the pixel grayscale value corresponding to the pixel coordinate (x, y) in the primary denoised image, x represents the abscissa of the pixel coordinate (x, y), y represents the ordinate of the pixel coordinate (x, y), j(x, y) represents the pixel grayscale value corresponding to the pixel coordinate (x, y) in the clear image, l(x, y) represents the pixel grayscale value of the pixel coordinate (x, y) in the fixed pattern noise, δ and β represent the first correction coefficient and the second correction coefficient respectively, and δ=0.7, β=1.2, and h represents the average grayscale value of all pixels in the clear image;

[0024] The median filter unit is used to perform a secondary denoising operation on the primary denoised image to obtain a denoised image, wherein the size of the filter kernel of the median filter unit is 3×3.

[0025] Optionally, extracting a set of to-be-identified vehicle images based on the denoised image, and calculating the extracted unit impact data of a unit time period based on the set of to-be-identified vehicle images, comprises:

[0026] Using a target detection algorithm to identify the denoised image, a set of vehicle images to be identified is obtained, wherein the set of vehicle images to be identified includes a plurality of vehicle images to be identified;

[0027] The following operations are performed on the vehicle images to be identified in the vehicle image set:

[0028] Identify the vehicle model of the vehicle image to be identified, and determine whether the license plate number can be extracted from the vehicle image to be identified. If the license plate number can be extracted from the vehicle image to be identified, identify the extracted license plate number to obtain the target license plate number; otherwise, confirm the license plate number of the vehicle image to be identified as 1 to obtain the target license plate number;

[0029] Construct vehicle data according to vehicle model and target license plate number, summarize vehicle data, and obtain the extracted vehicle data set of unit time period;

[0030] Calculate unit impact data from the vehicle dataset.

[0031] Optionally, calculating unit impact data according to the vehicle data set includes:

[0032] Extracting vehicle data from the vehicle data set in sequence, and confirming the target license plate number of the extracted vehicle data;

[0033] If the target license plate number is 1, the extracted vehicle data is confirmed as identification data, and the vehicle data is removed from the vehicle data set;

[0034] If the target license plate number is not 1, the extracted vehicle data is confirmed as identification data, and the same license plate set is extracted from the vehicle data set, and the same license plate set is removed from the vehicle data set to obtain a retained identification data set, and the retained identification data set is confirmed to be a vehicle data set, and the step of sequentially extracting vehicle data from the vehicle data set is returned until the vehicle data set is an empty set;

[0035] The identification data are aggregated to obtain an identification data set of the extracted unit time period, and the identification data set is input into a pre-built prediction model to obtain unit impact data of the extracted unit time period.

[0036] Optionally, the using a pre-built edge recognition algorithm to identify the denoised image sequence to obtain an initial crack area sequence includes:

[0037] The denoised image sequence is screened to obtain an initial crack image sequence, and the following operations are performed on the initial crack images in the initial crack image sequence:

[0038] A pixel gradient sequence is obtained based on the initial crack image, a first pixel probability, a second pixel probability and a third pixel probability are calculated based on the pixel gradient sequence, a first-category pixel grayscale mean, a second-category pixel grayscale mean and a third-category pixel grayscale mean are calculated according to the first pixel probability, the second pixel probability and the third pixel probability respectively, and a maximum total inter-class variance is calculated using the first pixel probability, the second pixel probability, the third pixel probability, the first-category pixel grayscale mean, the second-category pixel grayscale mean and the third-category pixel grayscale mean;

[0039] A low threshold and a high threshold are calculated based on the maximum total inter-class variance, and the initial crack image is divided using the low threshold and the high threshold to obtain a strong edge point set and a non-edge point set;

[0040] Construct a crack image according to the strong edge point set and the non-edge point set, and calculate the initial crack area according to the crack image;

[0041] The initial crack areas are summarized to obtain the initial crack area sequence.

[0042] Optionally, the maximum total inter-class variance is calculated as follows:

[0043]

[0044] Where T represents the maximum total inter-class variance, ω j represents the j-th pixel probability, a j represents the mean grayscale value of the j-th class pixel corresponding to the j-th pixel probability, represents the mean gray value of all pixels in the initial crack image, and j represents the j-th type of pixel.

[0045] Optionally, the step of dividing the initial crack image by using a low threshold and a high threshold to obtain a strong edge point set and a non-edge point set includes:

[0046] Based on the low threshold and the high threshold, the non-edge pixel interval, the weak edge pixel interval and the strong edge pixel interval are constructed, the crack pixels are extracted from the initial crack image in sequence, and the following operations are performed on the extracted crack pixels:

[0047] Obtaining a crack grayscale value based on the extracted crack pixels, and comparing the crack grayscale value with the size of the non-edge pixel interval, the weak edge pixel interval, and the strong edge pixel interval;

[0048] If the crack grayscale value belongs to the non-edge pixel interval, the grayscale value of the extracted crack pixel is output as 0 to obtain a non-edge point;

[0049] If the crack gray value belongs to the weak edge pixel interval, a judgment neighborhood is obtained based on the extracted crack pixels, and the maximum neighborhood pixel gray value is extracted from the judgment neighborhood. If the maximum neighborhood pixel gray value is greater than the high threshold, the gray value of the extracted crack pixel is output as 255 to obtain a strong edge point. If the maximum neighborhood pixel gray value is not greater than the high threshold, the gray value of the extracted crack pixel is output as 0 to obtain a non-edge point.

[0050] If the crack gray value belongs to the strong edge pixel interval, the gray value of the extracted crack pixel is output as 255 to obtain a strong edge point;

[0051] Summarize the strong edge points to get the strong edge point set, and summarize the non-edge points to get the non-edge point set.

[0052] Optionally, the interpolation calculation of the sequence to be interpolated by using the interpolation method to obtain the fracture area evolution sequence includes:

[0053] Constructing an interpolation model, wherein the interpolation model includes: a generation unit and a discrimination unit, wherein the discrimination unit includes a main discrimination unit and an auxiliary discrimination unit;

[0054] Based on the interpolation model, the generation unit is started, and according to the initial fracture area sequence and the sequence to be interpolated and the missing data set, the missing data set, the initial fracture area sequence and the sequence to be interpolated are input into the started generation unit to obtain a plurality of initial interpolation areas, and the plurality of initial interpolation areas are inserted into the sequence to be interpolated to obtain a filled sequence, wherein the missing data set includes a plurality of missing data, and the initial interpolation areas correspond to the missing data one by one;

[0055] Detect the data distribution of the sequence to be interpolated to obtain the original distribution, detect the data distribution of the filled sequence to obtain the filled distribution, and import the original distribution and the filled distribution into the discriminant unit;

[0056] If the original distribution is consistent with the filled distribution, the filled sequence is confirmed as the fracture area evolution sequence based on the auxiliary discrimination unit;

[0057] If the original distribution is inconsistent with the filled distribution, a missing value set is identified from the multiple initial interpolation areas based on the main discrimination unit, the missing value set is confirmed as a missing data set, and the step of inputting the missing data set, the initial crack area sequence and the sequence to be interpolated into the generation unit is returned until the filled sequence is confirmed as a crack area evolution sequence based on the auxiliary discrimination unit.

[0058] To achieve the above object, the present invention also provides a construction road crack prediction system based on time series monitoring data, comprising:

[0059] The historical image acquisition module is used to receive a road monitoring instruction, start the road monitoring device according to the road monitoring instruction, obtain the historical monitoring time based on the started road monitoring device, divide the historical monitoring time to obtain a plurality of unit time periods, extract the unit time periods from the plurality of unit time periods in sequence, and perform the following operations on the extracted unit time periods: obtain a plurality of historical images based on the extracted unit time periods;

[0060] The impact data acquisition module is used to extract historical images from a plurality of historical images in sequence, and perform the following operations on the extracted historical images: perform a clearing operation on the extracted historical images to obtain a clear image, perform a noise reduction operation on the clear image to obtain a noise reduction image, extract a set of vehicle images to be identified based on the noise reduction image, and calculate the extracted unit impact data of the unit time period based on the set of vehicle images to be identified;

[0061] The crack area evolution sequence module is used to summarize the unit impact data to obtain the unit impact data set corresponding to the historical monitoring time, summarize the denoised images to obtain the denoised image sequence corresponding to the historical monitoring time, use the pre-built edge recognition algorithm to identify the denoised image sequence to obtain the initial crack area sequence, screen the initial crack area sequence to obtain the sequence to be interpolated, and use the interpolation method to perform interpolation calculation on the sequence to be interpolated to obtain the crack area evolution sequence;

[0062] The crack prediction module is used to build a crack evolution model based on the crack area evolution sequence and the unit impact data set, obtain the time period to be predicted and the impact to be predicted, input the time period to be predicted and the impact to be predicted into the crack evolution model, obtain the predicted crack value, and complete the prediction of construction road cracks in the unit time period monitoring data according to the predicted crack value.

[0063] In order to solve the above problem, the present invention further provides an electronic device, the electronic device comprising:

[0064] A memory storing at least one instruction;

[0065] The processor executes the instructions stored in the memory to implement the above-mentioned construction road crack prediction method based on time series monitoring data.

[0066] In order to solve the above problems, the present invention also provides a computer-readable storage medium, in which at least one instruction is stored. The at least one instruction is executed by a processor in an electronic device to implement the above-mentioned construction road crack prediction method based on time series monitoring data.

[0067] The present invention is to solve the problem described in the background technology. The present invention obtains historical monitoring time based on the road monitoring device after startup, divides the historical monitoring time, obtains multiple unit time periods, and makes full use of historical data to build a crack evolution model. The present invention performs a clearing operation on the extracted historical image to obtain a clear image, performs a noise reduction operation on the clear image to obtain a noise reduction image, extracts a vehicle image set to be identified based on the noise reduction image, and calculates the unit impact data of the extracted unit time period based on the vehicle image set to be identified. By reducing the noise and clarifying the historical image, a basis is provided for identifying the initial crack area. At the same time, the clear image corresponds to the vehicle image set to be identified, and multiple clear images correspond to multiple vehicle sets to be identified, so that it is possible to identify how many vehicles have passed in a unit time period, and it is also possible to identify the size of the vehicle, thereby calculating the unit impact data to quantify the impact on the road. The present invention uses a pre-built edge recognition algorithm to identify the denoised image sequence, obtains an initial crack area sequence, screens the initial crack area sequence, obtains a sequence to be interpolated, uses an interpolation method to interpolate the sequence to be interpolated, obtains a crack area evolution sequence, removes the abnormal initial crack area caused by weather, light, machine noise, etc. in the initial crack area sequence, and uses the interpolation method to improve the sequence to be interpolated, and constructs a crack area evolution sequence that can accurately represent the evolution of road cracks within the historical monitoring time. The present invention constructs a crack evolution model based on the crack area evolution sequence and the unit impact data set, obtains the time period to be predicted and the impact to be predicted, inputs the time period to be predicted and the impact to be predicted into the crack evolution model, obtains the predicted crack value, completes the prediction of the construction road cracks of the unit time period monitoring data according to the predicted crack value, and describes the evolution of the cracks within the historical monitoring time according to the time sequence, the development of the road crack area and the unit impact data set. Therefore, the present invention can realize the prediction of road cracks using historical data. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 A schematic diagram of a process flow of a construction road crack prediction and monitoring method based on time series monitoring data provided by an embodiment of the present invention;

[0069] Figure 2 A functional module diagram of a construction road crack prediction and monitoring device based on time series monitoring data provided by an embodiment of the present invention;

[0070] Figure 3 A schematic diagram of the structure of an electronic device for implementing the construction road crack prediction model method based on time series monitoring data provided by an embodiment of the present invention.

[0071] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0072] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0073] The embodiment of the present application provides a construction road crack prediction method based on time series monitoring data. The execution subject of the construction road crack prediction method based on time series monitoring data includes but is not limited to at least one of the electronic devices such as a server and a terminal that can be configured to execute the method provided by the embodiment of the present application. In other words, the construction road crack prediction method based on time series monitoring data can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc.

[0074] Reference Figure 1 FIG. 1 is a flow chart of a method for predicting cracks on a construction road based on time series monitoring data provided by an embodiment of the present invention. In this embodiment, the method for predicting cracks on a construction road based on time series monitoring data includes:

[0075] S1. Receive a road monitoring instruction, start a road monitoring device according to the road monitoring instruction, obtain historical monitoring time based on the started road monitoring device, divide the historical monitoring time into multiple unit time periods.

[0076] It is understandable that the road monitoring instruction is an instruction to start the road monitoring device, which is initiated by the monitoring personnel. The road monitoring device is a device for monitoring the road.

[0077] For example,

[0078] Construction needs to be carried out on a road that has been used by vehicles. Xiao Zhang, as a technician, issued a road monitoring instruction to extract the usage status of the road in the past period of time from the road monitoring device in order to understand the usage status of the road in the past period of time, so as to determine the possible cracks and crack evolution of the road after renovation.

[0079] It is understandable that the historical monitoring time is a period of time for monitoring a road in the past, and the unit time period is a time shorter than the historical monitoring time.

[0080] It should be noted that the road to be constructed in the embodiment of the present invention is a road that has a certain amount of traffic flow. Therefore, by obtaining historical images of the past period of time and analyzing the past historical images, it is possible to understand the usage of the road to be constructed in the past period of time, including the traffic flow of the original road and the cracks of the original road. It is also possible to similarly predict the time when cracks may appear after the road to be constructed is completed, as well as the evolution of cracks after they appear. The construction road crack evolution model constructed in the embodiment of the present invention occurs before direct construction of the current road.

[0081] S2. Extract unit time periods in sequence from the multiple unit time periods, and perform the following operations on the extracted unit time periods: obtain multiple historical images based on the extracted unit time periods, extract historical images in sequence from the multiple historical images, and perform the following operations on the extracted historical images: perform a sharpening operation on the extracted historical images to obtain a clear image, and perform a noise reduction operation on the clear image to obtain a noise reduced image.

[0082] It is understandable that the historical image is an image of the current road taken by the road monitoring device in a unit time period. In the embodiment of the present invention, the image acquisition frequency of the road monitoring device is preset, so in the extracted unit time period, the road monitoring device takes multiple images of the current road according to the image acquisition frequency, and the multiple images of the current road are multiple historical images. For example, the historical monitoring time is from the first month to the sixth month of a year, a total of six months, and the historical monitoring time is divided into days, then the unit time period is 1 day, and the image acquisition frequency is 10s, then 144 historical images will be collected in 1 day.

[0083] Furthermore, performing a clearing operation on the extracted historical image to obtain a clear image includes:

[0084] Multiple Gaussian difference curves are generated using the pre-built Gaussian difference operator and the extracted historical images, where the Gaussian difference operator is:

[0085]

[0086] Among them, z represents the horizontal coordinate of the historical image in the image coordinate system, and w represents the vertical coordinate of the historical image in the image coordinate system. and Both represent the variance of Gaussian filter, σ1 and σ2 both represent the standard deviation of Gaussian filter, and σ1≠σ2, DOG represents the Gaussian difference operator;

[0087] The zero-crossing point of each Gaussian difference curve in a plurality of Gaussian difference curves is obtained to obtain a plurality of zero-crossing points, a plurality of contour coordinates are generated according to the plurality of zero-crossing points, the plurality of contour coordinates are spliced ​​using a pre-built edge connection algorithm to obtain a historical image contour, an image transition area is obtained according to the historical image contour, and a clear image is obtained according to the image transition area, wherein the contour coordinates correspond one-to-one to the Gaussian difference curve.

[0088] It is understandable that the sharpening operation is an operation of obtaining a clear image using a Gaussian difference operator. The Gaussian difference curve is a curve obtained by convolving the Gaussian difference operator with the historical image. The Gaussian difference curve is a curve from the peak to the trough, and the zero-crossing point in the Gaussian difference curve corresponds to the contour coordinates in the generated historical image. This technology is a prior art and will not be described in detail here. The construction of the image coordinate system is also a prior art and will not be described in detail here.

[0089] It should be noted that the pre-built edge connection algorithm is used to splice the multiple contour coordinates: a smooth curve is used to splice the multiple contour coordinates in sequence according to a preset counterclockwise direction. The image formed by the spliced ​​closed curve is the historical image contour, and the image transition area is the area jointly formed by the inner and outer areas of the closed curve in the historical image contour.

[0090] For example, taking the rectangular coordinate system as an example, the multiple contour coordinates are: (0,0), (1,1) and 0,1). Taking (0,0) as the starting point, (0,0), (1,1), (0,1) and (0,0) are connected in sequence to form a closed curve. The image formed by the closed curve is the contour of the historical image.

[0091] Specifically, the reason why a clear image can be obtained in the image transition area is that in the historical image, the position of the image contour has not changed, but the spatial distribution of the pixel grayscale values ​​in the adjacent area of ​​the image contour has changed, so a clear image can be obtained based on the image transition area.

[0092] For example, the pixel coordinates of the pixel in the historical image contour are (0,0), and its grayscale value is 54. Assuming that in a clear image, the grayscale value of the pixel coordinate (0,1) should be 0. In the historical image, due to atmospheric disturbances, inaccurate focus, or vehicle shaking, the historical image is blurred, so the grayscale value of the pixel coordinate (0,1) is affected by the pixel coordinate (0,0), so that the grayscale value of the pixel coordinate (0,1) is also 54, and the pixel coordinate (0, 1) is classified as the image transition area.

[0093] Furthermore, performing a noise reduction operation on the clear image to obtain a noise-reduced image includes:

[0094] Construct a mean filter unit and a median filter unit;

[0095] A standard image is obtained, and the spatial distribution characteristics of the pixel grayscale values ​​are extracted from the standard image. The spatial distribution characteristics of the pixel grayscale values ​​are introduced into a mean filter unit to obtain a fixed pattern noise, wherein the size of the filter kernel of the mean filter unit is 9×9;

[0096] A denoising operation is performed on the clear image using fixed pattern noise to obtain a denoised image, wherein the formula for a denoising operation is as follows:

[0097] k(x,y)=j(x,y)-δ×l(x,y)+β×h

[0098] Wherein, k(x, y) represents the pixel grayscale value corresponding to the pixel coordinate (x, y) in the primary denoised image, x represents the abscissa of the pixel coordinate (x, y), y represents the ordinate of the pixel coordinate (x, y), j(x, y) represents the pixel grayscale value corresponding to the pixel coordinate (x, y) in the clear image, l(x, y) represents the pixel grayscale value of the pixel coordinate (x, y) in the fixed pattern noise, δ and β represent the first correction coefficient and the second correction coefficient respectively, and δ=0.7, β=1.2, and h represents the average grayscale value of all pixels in the clear image;

[0099] The median filter unit is used to perform a secondary denoising operation on the primary denoised image to obtain a denoised image, wherein the size of the filter kernel of the median filter unit is 3×3.

[0100] It should be noted that the mean filtering unit is a unit that performs a mean filtering operation using the average pixel value of the filter kernel, and the median filtering unit is a unit that performs a median filtering operation using the median in the filter kernel. Both mean filtering and median filtering are existing technologies and will not be described in detail here.

[0101] Furthermore, the standard image is an image of a complete road without foreign matter or cracks. The pixel grayscale value spatial distribution feature is a feature map used to describe the spatial distribution of grayscale values ​​in the standard image. The pixel grayscale value spatial distribution feature is constructed as follows: a three-dimensional spatial distribution constructed with the image column of the standard image as the Y axis, the image row of the standard image as the X axis, and the grayscale value of the pixel in the standard image as the Z axis, wherein the image column is a column composed of pixels in a column of the standard image, the image row is a row composed of pixels in a row of the standard image, and the correspondence between the Y axis and the X axis of the standard image in the three-dimensional spatial distribution is the same as that of the Y axis and the X axis of the standard image in the image coordinate system.

[0102] Optionally, in this embodiment, a camera device is used to capture an image without foreign objects or cracks on the road to be predicted, and a standard image can be obtained after mean filtering and median filtering are performed on the image.

[0103] Furthermore, fixed pattern noise is a common noise in image sensors, which is mainly manifested as deviations in the same pattern such as fixed position, fixed brightness, and fixed color in the image, which does not change with the change of the collected image. It is a special term in the professional field and will not be repeated here. After the spatial distribution characteristics of the pixel grayscale value are imported into the mean filter unit, the standard image is filtered by the mean filter unit, and the fixed pattern noise can be obtained through the filtered standard image. The primary denoising image is an image obtained by a clear image after a denoising operation. The primary denoising operation is a denoising operation to remove the fixed pattern noise in the clear image that is the same as the standard image. The pixel grayscale value is the grayscale value of the pixel, the pixel coordinates are the coordinates of the pixel in the image coordinate system, the first correction coefficient is a coefficient to ensure that the grayscale value after the primary denoising operation is within the range of [0, 255], and the second correction coefficient is a coefficient to adjust the overall grayscale value of the denoised image. The second correction coefficient solves the problem of insufficient overall brightness of the denoised image after denoising. The embodiment of the present invention defines the first correction coefficient and the second correction coefficient. Under different road materials, the first correction coefficient and the second correction coefficient can be adjusted according to the actual situation.

[0104] It should be explained that the average grayscale value is the average grayscale value of all pixels in the clear image. The secondary denoising operation is an operation of denoising the primary denoised image using a median filter unit.

[0105] S3. Extracting a set of to-be-identified vehicle images based on the denoised image, and calculating the extracted unit impact data of a unit time period based on the set of to-be-identified vehicle images.

[0106] Furthermore, extracting a set of vehicle images to be identified based on the denoised image, and calculating the extracted unit impact data of a unit time period based on the set of vehicle images to be identified, comprises:

[0107] Using a target detection algorithm to identify the denoised image, a set of vehicle images to be identified is obtained, wherein the set of vehicle images to be identified includes a plurality of vehicle images to be identified;

[0108] The following operations are performed on the vehicle images to be identified in the vehicle image set:

[0109] Identify the vehicle model of the vehicle image to be identified, and determine whether the license plate number can be extracted from the vehicle image to be identified. If the license plate number can be extracted from the vehicle image to be identified, identify the extracted license plate number to obtain the target license plate number; otherwise, confirm the license plate number of the vehicle image to be identified as 1 to obtain the target license plate number;

[0110] Construct vehicle data according to vehicle model and target license plate number, summarize vehicle data, and obtain the extracted vehicle data set of unit time period;

[0111] Calculate unit impact data from the vehicle dataset.

[0112] It is understandable that the target detection algorithm is a prior art, which can identify the vehicle in the image from the denoised image. Optionally, the target detection algorithm uses the YOLOX algorithm. The image of the vehicle to be identified is an image of a vehicle identified using the target detection algorithm. The target license plate number is the license plate number identified from the image of the vehicle to be identified. Vehicle models include: large buses, medium-sized buses, small buses, micro buses and non-motor vehicles. The classification standard of vehicle models is the Chinese automobile classification standard, which is a prior art and will not be repeated here.

[0113] It should be noted that the vehicle data is data used to describe the vehicle in the vehicle image to be identified. It should be noted that in the process of identifying vehicles using the target detection algorithm, multiple denoised images are obtained in a unit time period, and one denoised image corresponds to a set of vehicle images to be identified. The extracted unit time period includes multiple denoised images. Therefore, the vehicle data set corresponding to the unit time period includes all vehicle data of the multiple vehicle image sets to be identified in the unit time period. The unit impact data is the impact data of the vehicle on the road in the unit time period, which is calculated by the vehicle flow and vehicle model in the unit time period.

[0114] It is understandable that the current road defined in the embodiment of the present invention does not include traffic lights. Therefore, when the target detection algorithm is used to detect the set of vehicle images to be identified in the denoised image, the situation that the license plate number of the vehicle cannot be identified because it is blocked by other vehicles is relatively rare. In addition, there will be situations including but not limited to the following situations that lead to the inability to extract the license plate number: the vehicle is at the edge of the picture, resulting in the inability to identify the license plate number, the vehicle is not equipped with a license plate number, and the vehicle license plate number is blocked by stains, resulting in the inability to identify the license plate number, etc., which will not be given one by one here. If the vehicles whose license plates cannot be identified are ignored, the actual impact and actual traffic volume of the road will be much greater than the predicted unit impact data. If these vehicles whose license plates cannot be identified are included in the calculation range of the unit impact data, the actual impact and actual traffic volume of the road will be less than the predicted unit impact data. Therefore, according to the unit impact data at this time, the road is renovated and the road plan is formulated, so that the road can withstand greater future impacts and future traffic volume. When the actual impact and actual traffic volume are applied to the renovated road, the road cracks will appear later, thereby increasing the life of the road.

[0115] It should be noted that the to-be-identified vehicle image set is a collection of multiple to-be-identified vehicle images. When there is no vehicle traveling on the current road, the to-be-identified vehicle image set is an empty set.

[0116] Further, the calculating unit impact data according to the vehicle data set includes:

[0117] Extracting vehicle data from the vehicle data set in sequence, and confirming the target license plate number of the extracted vehicle data;

[0118] If the target license plate number is 1, the extracted vehicle data is confirmed as identification data, and the vehicle data is removed from the vehicle data set;

[0119] If the target license plate number is not 1, the extracted vehicle data is confirmed as identification data, and the same license plate set is extracted from the vehicle data set, and the same license plate set is removed from the vehicle data set to obtain a retained identification data set, and the retained identification data set is confirmed to be a vehicle data set, and the step of sequentially extracting vehicle data from the vehicle data set is returned until the vehicle data set is an empty set;

[0120] The identification data are aggregated to obtain an identification data set of the extracted unit time period, and the identification data set is input into a pre-built prediction model to obtain unit impact data of the extracted unit time period.

[0121] Furthermore, the identification data includes vehicle model and vehicle license plate number. In the identification data set, there are no two vehicles with the same target license plate number except for the vehicle with the target license plate number 1. The purpose is to use the identification data with the target license plate number 1 to improve the safety factor of the road's tolerable impact, and use the identification data with the target license plate number not 1 to improve the accuracy of the unit impact data.

[0122] For example, the vehicle data set is: large bus-1, medium bus-1, small bus-XXX01, small bus-XXX01, small bus-XXX01, small bus-XXX02, non-motor vehicle-1, non-motor vehicle-XXX09. The vehicle data is extracted from the vehicle data set in sequence. When the vehicle data is large bus-1, the target license plate number is confirmed to be 1, then large bus-1 is confirmed as identification data, and large bus-1 is removed from the vehicle data set. At this time, the vehicle data set is: medium bus-1, small bus-XXX01, small bus-XXX01, small bus-XXX01, small bus-XXX02, non-motor vehicle-1, non-motor vehicle-XXX09. When the vehicle data is medium-sized passenger car-1, it is confirmed that the target license plate number is 1, then medium-sized passenger car-1 is confirmed as identification data, and medium-sized passenger car-1 is removed from the vehicle data set. At this time, the vehicle data set is: small passenger car-XXX01, small passenger car-XXX01, small passenger car-XXX01, small passenger car-XXX02, non-motor vehicle-1, non-motor vehicle-XXX09. When the vehicle data is small passenger car-XXX01, the target license plate number is not 1, small passenger car-XXX01 is confirmed as identification data, and the same license plate set is extracted from the vehicle data set. The same license plate set is the set of vehicle data of all small passenger cars-XXX01 in the vehicle data set. In this example, the vehicle data includes 3 vehicle data of small passenger cars-XXX01, so the same license plate set is: small passenger car-XXX01, small passenger car-XXX01 and small passenger car-XXX01. Remove the same license plate set from the vehicle dataset to obtain the retained identification dataset, then the retained identification dataset is: small passenger car-XXX02, non-motor vehicle-1, non-motor vehicle-XXX09. And so on, until the vehicle dataset is an empty set. In this example, the final identification dataset is: large passenger car-1, medium passenger car-1, small passenger car-XXX01, small passenger car-XXX02, non-motor vehicle-1, non-motor vehicle-XXX09.

[0123] It can be understood that the prediction model is a model that predicts the impact based on the identification data set. For example, in the extracted unit time period, the identification data set is: large bus-1, medium bus-1, small bus-XXX01, small bus-XXX02, non-motor vehicle-1, non-motor vehicle-XXX09, then there is a large bus, a medium bus, two small buses, and two non-motor vehicles. Since vehicles of different masses have different impacts on the road, vehicles of different vehicle models need to be calculated separately. Based on one large bus, one medium bus, two small buses, and two non-motor vehicles, the impact on the road in the unit time period can be calculated.

[0124] S4. Summarize the unit impact data to obtain the unit impact data set corresponding to the historical monitoring time, summarize the denoised images to obtain the denoised image sequence corresponding to the historical monitoring time, use a pre-built edge recognition algorithm to identify the denoised image sequence to obtain an initial crack area sequence, filter the initial crack area sequence, and obtain a sequence to be interpolated.

[0125] It should be noted that the denoised image sequence is a sequence of multiple denoised images sorted in chronological order. The purpose of screening the initial crack area sequence is to eliminate the abnormal initial crack areas in the initial crack area sequence. The calculation of the crack area by image recognition method will be affected by weather, light, machine noise and other aspects. Therefore, not every initial crack image can accurately represent the actual area of ​​the road crack at that time, so it is necessary to screen the initial crack area sequence. After eliminating the outliers in the initial crack area sequence, the multiple initial crack areas retained are the sequence to be interpolated.

[0126] Furthermore, the method of using a pre-built edge recognition algorithm to identify the denoised image sequence to obtain an initial crack area sequence includes:

[0127] The denoised image sequence is screened to obtain an initial crack image sequence, and the following operations are performed on the initial crack images in the initial crack image sequence:

[0128] A pixel gradient sequence is obtained based on the initial crack image, a first pixel probability, a second pixel probability and a third pixel probability are calculated based on the pixel gradient sequence, a first-category pixel grayscale mean, a second-category pixel grayscale mean and a third-category pixel grayscale mean are calculated according to the first pixel probability, the second pixel probability and the third pixel probability respectively, and a maximum total inter-class variance is calculated using the first pixel probability, the second pixel probability, the third pixel probability, the first-category pixel grayscale mean, the second-category pixel grayscale mean and the third-category pixel grayscale mean;

[0129] A low threshold and a high threshold are calculated based on the maximum total inter-class variance, and the initial crack image is divided using the low threshold and the high threshold to obtain a strong edge point set and a non-edge point set;

[0130] Construct a crack image according to the strong edge point set and the non-edge point set, and calculate the initial crack area according to the crack image;

[0131] The initial crack areas are summarized to obtain the initial crack area sequence.

[0132] Furthermore, the calculation formula of the maximum total inter-class variance is:

[0133]

[0134] Where T represents the maximum total inter-class variance, ω jrepresents the j-th pixel probability, a j represents the mean grayscale value of the j-th class pixel corresponding to the j-th pixel probability, It represents the mean gray value of all pixels in the denoised image, and j represents the j-th type of pixel.

[0135] It should be noted that in the denoised image sequence, there must be images where vehicles block road cracks. Therefore, the denoised image sequence is screened to obtain an initial crack image sequence so that the initial crack images in the initial crack image sequence are all images where road cracks are exposed.

[0136] Furthermore, the pixel gradient sequence is a sequence of pixel grayscale values ​​in the initial crack image. This technology is prior art and will not be described in detail here. The first pixel probability, the second pixel probability and the third pixel probability are the probabilities of the pixels in the pixel gradient sequence being in the first class of pixels, the second class of pixels and the third class of pixels, respectively. In this embodiment, j represents the jth class of pixels. When j=1, it represents the first class of pixels, corresponding to the first pixel probability and the grayscale mean of the first class of pixels. When j=2, it represents the second class of pixels, corresponding to the second pixel probability and the grayscale mean of the second class of pixels. When j=3, it represents the third class of pixels, corresponding to the third pixel probability and the grayscale mean of the third class of pixels.

[0137] It should be noted that before calculating the first pixel probability, the second pixel probability and the third pixel probability, an initial high threshold and an initial low threshold are preset, wherein the initial high threshold is greater than the initial low threshold, and the initial high threshold and the initial low threshold are both grayscale values ​​in the pixel gradient sequence. The pixel gradient sequence is divided into three categories according to the initial high threshold and the initial low threshold, pixels corresponding to grayscale values ​​lower than the initial low threshold are first-category pixels, pixels corresponding to grayscale values ​​between the initial low threshold and the initial high threshold are second-category pixels, and pixels corresponding to grayscale values ​​higher than the initial high threshold are third-category pixels. At this time, the grayscale mean of the first-category pixels, the grayscale mean of the second-category pixels and the grayscale mean of the third-category pixels are the average grayscale value of all first-category pixels, the average grayscale value of all second-category pixels and the average grayscale value of all third-category pixels, respectively.

[0138] Furthermore, different initial high thresholds and initial low thresholds calculate different total inter-class variances. The total inter-class variance is a value used to determine the degree of dispersion among the first-class pixels, the second-class pixels, and the third-class pixels. When the maximum total inter-class variance is obtained, the difference in grayscale values ​​among the first-class pixels, the second-class pixels, and the third-class pixels is the largest. Therefore, the initial crack image is processed using the initial high threshold and the initial low threshold corresponding to the maximum total inter-class variance, so that the edge information of the crack can be extracted from the initial crack image. In an embodiment of the present invention, the initial high threshold and the initial low threshold corresponding to the maximum total inter-class variance are identified as the high threshold and the low threshold, and the initial crack image is divided using the high threshold and the low threshold.

[0139] Specifically, the crack image is a contour image of a road crack after edge detection, and the initial crack area is the actual crack area calculated based on the crack image. The technical means for achieving this is the prior art. For example, a smooth contour image of a road crack is drawn based on a strong edge set in the crack image, and the actual crack area of ​​the road crack is calculated based on the ratio of the crack image to the actual road crack. The initial crack area sequence is a sequence of initial crack areas sorted in chronological order.

[0140] Furthermore, the use of a low threshold and a high threshold to divide the initial crack image to obtain a strong edge point set and a non-edge point set includes:

[0141] Based on the low threshold and the high threshold, the non-edge pixel interval, the weak edge pixel interval and the strong edge pixel interval are constructed, the crack pixels are extracted from the initial crack image in sequence, and the following operations are performed on the extracted crack pixels:

[0142] Obtaining a crack grayscale value based on the extracted crack pixels, and comparing the crack grayscale value with the size of the non-edge pixel interval, the weak edge pixel interval, and the strong edge pixel interval;

[0143] If the crack grayscale value belongs to the non-edge pixel interval, the grayscale value of the extracted crack pixel is output as 0 to obtain a non-edge point;

[0144] If the crack gray value belongs to the weak edge pixel interval, a judgment neighborhood is obtained based on the extracted crack pixels, and the maximum neighborhood pixel gray value is extracted from the judgment neighborhood. If the maximum neighborhood pixel gray value is greater than the high threshold, the gray value of the extracted crack pixel is output as 255 to obtain a strong edge point. If the maximum neighborhood pixel gray value is not greater than the high threshold, the gray value of the extracted crack pixel is output as 0 to obtain a non-edge point.

[0145] If the crack gray value belongs to the strong edge pixel interval, the gray value of the extracted crack pixel is output as 255 to obtain a strong edge point;

[0146] Summarize the strong edge points to get the strong edge point set, and summarize the non-edge points to get the non-edge point set.

[0147] Furthermore, the non-edge pixel interval is an interval composed of grayscale values ​​greater than or equal to zero 0 and less than a low threshold, the weak edge pixel interval is an interval composed of grayscale values ​​greater than or equal to the low threshold and less than a high threshold, the strong edge pixel interval is an interval composed of grayscale values ​​greater than or equal to the high threshold and less than 255, the crack pixels are pixels in the initial crack image, and the crack grayscale value is the grayscale value of the crack pixels.

[0148] It can be understood that the judgment neighborhood is a set of pixels in the neighborhood around the extracted pixel, and the maximum neighborhood pixel grayscale value is the maximum grayscale value in the judgment neighborhood. For example, the judgment neighborhood takes a 3×3 window size. For ease of representation, the 3×3 window size is set to a 3×3 matrix, and the elements in the 3×3 matrix represent grayscale values. The grayscale value corresponding to the extracted pixel is located in the second row and second column. When the maximum grayscale value in the 3×3 matrix is ​​greater than the high threshold, the grayscale value of the extracted pixel is output as 255, otherwise, the grayscale value of the extracted pixel is output as 0.

[0149] Furthermore, the strong edge points are pixel points of the edge contour of the crack, and the non-edge points are pixel points of the edge contour of the non-crack.

[0150] S5. Use the interpolation method to perform interpolation calculation on the interpolation sequence to obtain the fracture area evolution sequence.

[0151] Furthermore, the interpolation method is used to perform interpolation calculation on the interpolation sequence to obtain the fracture area evolution sequence, including:

[0152] Constructing an interpolation model, wherein the interpolation model includes: a generation unit and a discrimination unit, wherein the discrimination unit includes a main discrimination unit and an auxiliary discrimination unit;

[0153] Based on the interpolation model, the generation unit is started, and according to the initial fracture area sequence and the sequence to be interpolated and the missing data set, the missing data set, the initial fracture area sequence and the sequence to be interpolated are input into the started generation unit to obtain a plurality of initial interpolation areas, and the plurality of initial interpolation areas are inserted into the sequence to be interpolated to obtain a filled sequence, wherein the missing data set includes a plurality of missing data, and the initial interpolation areas correspond to the missing data one by one;

[0154] Detect the data distribution of the sequence to be interpolated to obtain the original distribution, detect the data distribution of the filled sequence to obtain the filled distribution, and import the original distribution and the filled distribution into the discriminant unit;

[0155] If the original distribution is consistent with the filled distribution, the filled sequence is confirmed as the fracture area evolution sequence based on the auxiliary discrimination unit;

[0156] If the original distribution is inconsistent with the filled distribution, a missing value set is identified from the multiple initial interpolation areas based on the main discrimination unit, the missing value set is confirmed as a missing data set, and the step of inputting the missing data set, the initial crack area sequence and the sequence to be interpolated into the generation unit is returned until the filled sequence is confirmed as a crack area evolution sequence based on the auxiliary discrimination unit.

[0157] Furthermore, the interpolation model is a model for interpolating a sequence to be interpolated. The generation unit is a unit for generating interpolation values, the determination unit is a unit for determining whether the generated interpolation values ​​are valid interpolation values, the main determination unit is a unit for identifying an initial interpolation area that cannot be effectively used from a plurality of initial interpolation areas, and the auxiliary determination unit is a unit for determining data distribution.

[0158] It should be noted that the missing data set is the set of data removed from the initial crack area sequence, the initial interpolation area is the area value generated by the generation unit, and the filled sequence is the sequence after all the initial interpolation areas are inserted into the positions corresponding to the missing data set.

[0159] For example, the initial crack area sequence is: 10, 20, 16, 16, 16, 17, 28, 18; the initial crack area sequence is screened to obtain the sequence to be interpolated, and the sequence to be interpolated is: 10, [*], 16, 16, 16, 17, [*], 18, where [*] and [*] are two data to be interpolated. Since the sequence to be interpolated includes time information, [*] and [*] respectively constitute missing data sets with the corresponding time information. Assuming that the multiple initial interpolation areas generated by the generation unit are 12 and 17, the multiple initial interpolation areas are interpolated into the sequence to be interpolated, and the interpolated sequence is: 10, 12, 16, 16, 16, 17, 17, 18.

[0160] Furthermore, the original distribution is the data distribution of the sequence to be interpolated, and the filled distribution is the data distribution of the filled sequence.

[0161] Exemplarily, in the embodiment of the present invention, when the road cracks are not filled, the road cracks are getting bigger and bigger. Therefore, if the original distribution is a straight line obtained by fitting using the least squares method, when judging the filled distribution, the least squares method is also used to fit the data of the filled sequence into a straight line. When comparing the filled distribution and the original distribution, the slope and intercept corresponding to the original distribution are extracted, the slope corresponding to the original distribution is subtracted from the slope corresponding to the filled distribution and the absolute value is taken to obtain the slope absolute difference, the intercept corresponding to the original distribution is subtracted from the intercept corresponding to the filled distribution and the absolute value is taken to obtain the intercept absolute difference. If the slope absolute difference is less than a preset difference threshold and the intercept absolute difference is less than a preset intercept threshold, the original distribution is considered to be consistent with the filled distribution. Among them, the difference threshold is the maximum slope absolute difference for considering that the original distribution is consistent with the filled distribution, and the intercept threshold is the maximum intercept absolute difference for considering that the original distribution is consistent with the filled distribution.

[0162] It should be noted that the crack area evolution sequence is a sequence that can describe the evolution of road cracks after interpolation. The missing value set is a set of data whose distribution of the sequence to be interpolated is destroyed by multiple initial interpolation areas. For example, the sequence to be interpolated is: 10, [*], 16, 16, 16, 17, [*], 18, where [*] and [*] are two data that need to be interpolated. Assuming that the multiple initial interpolation areas generated by the generation unit are 22 and 17, and after judgment, 22 has destroyed the data distribution of the sequence to be interpolated, the main discrimination unit can be used to confirm the data corresponding to 22 as an element in the missing value set, and the missing value set can be confirmed as a missing data set.

[0163] S6. Construct a crack evolution model based on the crack area evolution sequence and the unit impact data set, obtain the time period to be predicted and the impact to be predicted, input the time period to be predicted and the impact to be predicted into the crack evolution model, obtain the predicted crack value, and complete the prediction of the construction road cracks of the unit time monitoring data according to the predicted crack value.

[0164] It should be noted that the crack area evolution sequence is a sequence of the areas of road cracks sorted in chronological order and is constructed by multiple crack area evolution sequences. For example, at time a, the corresponding initial crack image sequence is: 0, 0, 1, 2; at time (a+1), the corresponding initial crack image sequence is: 2, 2, 2.2, 3; there is no missing data at time a and time (a+1), then the crack area evolution sequence is: 0, 0, 1, 2, 2, 2, 2.2, 3.

[0165] Furthermore, the unit impact data set is a collection of multiple unit impact data. In this embodiment, a unit impact data corresponds to a unit time period, and the unit time period corresponds to an initial crack image sequence. The initial crack area corresponds to the initial crack image one by one. For example, the unit impact data corresponding to the unit time period is at, t is tons, and the initial crack image sequence is: 0, 0, 1, 2. At this time, the initial crack area sequence corresponds to the unit impact data at. Therefore, the evolution of cracks during the historical monitoring time can be described based on the time sequence, the development of the road crack area and the unit impact data set. Therefore, the crack evolution model is a model used to describe the evolution of cracks during the historical monitoring time.

[0166] Furthermore, the time period to be predicted is the time period set by a person for predicting the expected road crack situation, the impact to be predicted is the data on the impact on the road set by a person, and the predicted crack value is the area of ​​the road crack predicted after inputting the time period to be predicted and the impact to be predicted into the crack evolution model.

[0167] The present invention is to solve the problem described in the background technology. The present invention obtains historical monitoring time based on the road monitoring device after startup, divides the historical monitoring time, obtains multiple unit time periods, and makes full use of historical data to build a crack evolution model. The present invention performs a clearing operation on the extracted historical image to obtain a clear image, performs a noise reduction operation on the clear image to obtain a noise reduction image, extracts a vehicle image set to be identified based on the noise reduction image, and calculates the unit impact data of the extracted unit time period based on the vehicle image set to be identified. By reducing the noise and clarifying the historical image, a basis is provided for identifying the initial crack area. At the same time, the clear image corresponds to the vehicle image set to be identified, and multiple clear images correspond to multiple vehicle sets to be identified, so that it is possible to identify how many vehicles have passed in a unit time period, and it is also possible to identify the size of the vehicle, thereby calculating the unit impact data to quantify the impact on the road. The present invention uses a pre-built edge recognition algorithm to identify the denoised image sequence, obtains an initial crack area sequence, screens the initial crack area sequence, obtains a sequence to be interpolated, uses an interpolation method to interpolate the sequence to be interpolated, obtains a crack area evolution sequence, removes the abnormal initial crack area caused by weather, light, machine noise, etc. in the initial crack area sequence, and uses the interpolation method to improve the sequence to be interpolated, and constructs a crack area evolution sequence that can accurately represent the evolution of road cracks within the historical monitoring time. The present invention constructs a crack evolution model based on the crack area evolution sequence and the unit impact data set, obtains the time period to be predicted and the impact to be predicted, inputs the time period to be predicted and the impact to be predicted into the crack evolution model, obtains the predicted crack value, completes the prediction of the construction road cracks of the unit time period monitoring data according to the predicted crack value, and describes the evolution of the cracks within the historical monitoring time according to the time sequence, the development of the road crack area and the unit impact data set. Therefore, the present invention can realize the prediction of road cracks using historical data.

[0168] like Figure 2 , which is a functional module diagram of a construction road crack prediction system based on time series monitoring data provided by an embodiment of the present invention.

[0169] The construction road crack prediction system 100 based on time series monitoring data of the present invention can be installed in an electronic device. According to the functions implemented, the construction road crack prediction system 100 based on time series monitoring data can include a historical image acquisition module 101, an impact data acquisition module 102, a crack area evolution sequence module 103 and a crack prediction module 104. The module of the present invention can also be called a unit, which refers to a series of computer program segments that can be executed by an electronic device processor and can complete fixed functions, which are stored in the memory of the electronic device.

[0170] The historical image acquisition module 101 is used to receive a road monitoring instruction, start a road monitoring device according to the road monitoring instruction, obtain a historical monitoring time based on the started road monitoring device, divide the historical monitoring time into multiple unit time periods, extract unit time periods from the multiple unit time periods in sequence, and perform the following operations on the extracted unit time periods: obtain multiple historical images based on the extracted unit time periods;

[0171] The impact data acquisition module 102 is used to extract historical images from a plurality of historical images in sequence, and perform the following operations on the extracted historical images: perform a clearing operation on the extracted historical images to obtain a clear image, perform a noise reduction operation on the clear image to obtain a noise reduction image, extract a set of vehicle images to be identified based on the noise reduction image, and calculate the extracted unit impact data of the unit time period based on the set of vehicle images to be identified;

[0172] The crack area evolution sequence module 103 is used to summarize the unit impact data to obtain the unit impact data set corresponding to the historical monitoring time, summarize the denoised images to obtain the denoised image sequence corresponding to the historical monitoring time, identify the denoised image sequence using a pre-built edge recognition algorithm to obtain an initial crack area sequence, screen the initial crack area sequence to obtain a sequence to be interpolated, and perform interpolation calculation on the sequence to be interpolated using an interpolation method to obtain a crack area evolution sequence;

[0173] The crack prediction module 104 is used to build a crack evolution model based on the crack area evolution sequence and the unit impact data set, obtain the time period to be predicted and the impact to be predicted, input the time period to be predicted and the impact to be predicted into the crack evolution model, obtain the predicted crack value, and complete the prediction of the construction road cracks in the unit time period monitoring data according to the predicted crack value.

[0174] In detail, each module in the construction road crack prediction system 100 based on time series monitoring data in the embodiment of the present invention is used in the same manner as described above. Figure 1 The same technical means are used as the construction road crack prediction method based on time series monitoring data described in, and can produce the same technical effects, so they will not be repeated here.

[0175] like Figure 3 , is a schematic diagram of the structure of an electronic device for implementing a construction road crack prediction method based on time series monitoring data provided by an embodiment of the present invention.

[0176] The electronic device 1 may include a processor 10, a memory 11 and a bus 12, and may also include a computer program stored in the memory 11 and executable on the processor 10, such as a construction road crack prediction method program based on time series monitoring data.

[0177] Among them, the memory 11 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, mobile hard disk, multimedia card, card-type memory (for example: SD or DX memory, etc.), magnetic memory, disk, optical disk, etc. The memory 11 may be an internal storage unit of the electronic device 1 in some embodiments, such as a mobile hard disk of the electronic device 1. The memory 11 may also be an external storage device of the electronic device 1 in other embodiments, such as a plug-in mobile hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), etc. equipped on the electronic device 1. Further, the memory 11 also includes an internal storage unit of the electronic device 1 and an external storage device. The memory 11 can not only be used to store application software and various types of data installed in the electronic device 1, such as the code of the construction road crack prediction method program based on time series monitoring data, but also can be used to temporarily store data that has been output or is to be output.

[0178] The processor 10 may be composed of an integrated circuit in some embodiments, for example, a single packaged integrated circuit, or a plurality of packaged integrated circuits with the same or different functions, including one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and combinations of various control chips, etc. The processor 10 is the control core (Control Unit) of the electronic device, and uses various interfaces and lines to connect various components of the entire electronic device, and executes or executes programs or modules stored in the memory 11 (such as a construction road crack prediction method program based on time series monitoring data, etc.), and calls data stored in the memory 11 to execute various functions of the electronic device 1 and process data.

[0179] The bus 12 may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The bus 12 may be divided into an address bus, a data bus, a control bus, etc. The bus 12 is configured to realize connection and communication between the memory 11 and at least one processor 10, etc.

[0180] Figure 3 Only an electronic device with components is shown, and those skilled in the art will understand that Figure 3The structure shown does not constitute a limitation on the electronic device 1, and may include fewer or more components than shown in the figure, or combine certain components, or arrange the components differently.

[0181] For example, although not shown, the electronic device 1 may also include a power source (such as a battery) for supplying power to each component. Preferably, the power source may be logically connected to the at least one processor 10 through a power management device, so that the power management device can realize functions such as charging management, discharging management, and power consumption management. The power source may also include any components such as one or more DC or AC power sources, recharging devices, power failure detection circuits, power converters or inverters, power status indicators, etc. The electronic device 1 may also include a variety of sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be repeated here.

[0182] Furthermore, the electronic device 1 may also include a network interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), which is generally used to establish a communication connection between the electronic device 1 and other electronic devices.

[0183] Optionally, the electronic device 1 may further include a user interface, which may be a display, an input unit (such as a keyboard), or a standard wired interface or a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, and an OLED (Organic Light-Emitting Diode) touch device. The display may also be appropriately referred to as a display screen or a display unit, which is used to display information processed in the electronic device 1 and to display a visual user interface.

[0184] It should be understood that the embodiment is for illustration only and the scope of the patent application is not limited to this structure.

[0185] The construction road crack prediction method program based on time series monitoring data stored in the memory 11 of the electronic device 1 is a combination of multiple instructions. When running in the processor 10, it can achieve:

[0186] receiving a road monitoring instruction, starting a road monitoring device according to the road monitoring instruction, obtaining a historical monitoring time based on the started road monitoring device, and dividing the historical monitoring time to obtain a plurality of unit time periods;

[0187] Extracting unit time periods from the plurality of unit time periods in sequence, and performing the following operations on the extracted unit time periods: acquiring a plurality of historical images based on the extracted unit time periods, extracting historical images from the plurality of historical images in sequence, and performing the following operations on the extracted historical images:

[0188] Performing a clearing operation on the extracted historical image to obtain a clear image, performing a denoising operation on the clear image to obtain a denoised image, extracting a set of vehicle images to be identified based on the denoised image, and calculating the extracted unit impact data of the unit time period based on the set of vehicle images to be identified;

[0189] Summarize the unit impact data to obtain the unit impact data set corresponding to the historical monitoring time, summarize the denoised images to obtain the denoised image sequence corresponding to the historical monitoring time, use the pre-built edge recognition algorithm to identify the denoised image sequence to obtain the initial crack area sequence, and filter the initial crack area sequence to obtain the sequence to be interpolated;

[0190] The interpolation method is used to interpolate the sequence to be interpolated, and the crack area evolution sequence is obtained;

[0191] A crack evolution model is constructed based on the crack area evolution sequence and the unit impact data set, and the time period to be predicted and the impact to be predicted are obtained. The time period to be predicted and the impact to be predicted are input into the crack evolution model to obtain the predicted crack value. The prediction of construction road cracks in the unit time period monitoring data is completed based on the predicted crack value.

[0192] Specifically, the specific implementation method of the processor 10 for the above instructions can refer to Figures 1 to 3 The description of the relevant steps in the corresponding embodiments will not be repeated here.

[0193] Furthermore, if the module / unit integrated in the electronic device 1 is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. The computer-readable storage medium can be volatile or non-volatile. For example, the computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a mobile hard disk, a magnetic disk, an optical disk, a computer memory, and a read-only memory (ROM).

[0194] The present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor of an electronic device, the computer program can implement:

[0195] receiving a road monitoring instruction, starting a road monitoring device according to the road monitoring instruction, obtaining a historical monitoring time based on the started road monitoring device, and dividing the historical monitoring time to obtain a plurality of unit time periods;

[0196] Extracting unit time periods from the plurality of unit time periods in sequence, and performing the following operations on the extracted unit time periods: acquiring a plurality of historical images based on the extracted unit time periods, extracting historical images from the plurality of historical images in sequence, and performing the following operations on the extracted historical images:

[0197] Performing a clearing operation on the extracted historical image to obtain a clear image, performing a denoising operation on the clear image to obtain a denoised image, extracting a set of vehicle images to be identified based on the denoised image, and calculating the extracted unit impact data of the unit time period based on the set of vehicle images to be identified;

[0198] Summarize the unit impact data to obtain the unit impact data set corresponding to the historical monitoring time, summarize the denoised images to obtain the denoised image sequence corresponding to the historical monitoring time, use the pre-built edge recognition algorithm to identify the denoised image sequence to obtain the initial crack area sequence, and filter the initial crack area sequence to obtain the sequence to be interpolated;

[0199] The interpolation method is used to interpolate the sequence to be interpolated, and the crack area evolution sequence is obtained;

[0200] A crack evolution model is constructed based on the crack area evolution sequence and the unit impact data set, and the time period to be predicted and the impact to be predicted are obtained. The time period to be predicted and the impact to be predicted are input into the crack evolution model to obtain the predicted crack value. The prediction of construction road cracks in the unit time period monitoring data is completed based on the predicted crack value.

[0201] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, systems and methods can be implemented in other ways. For example, the system embodiments described above are only illustrative, and actual implementation may have other division methods.

[0202] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0203] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0204] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0205] In addition, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in the system claim can also be implemented by one unit or device through software or hardware. The second and other words are used to indicate names, but not to indicate any particular order.

[0206] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A construction road crack prediction method based on time series monitoring data, characterized in that: The method comprises: receiving a road monitoring instruction, starting a road monitoring device according to the road monitoring instruction, obtaining a historical monitoring time based on the started road monitoring device, and dividing the historical monitoring time to obtain a plurality of unit time periods; Extracting unit time periods from the plurality of unit time periods in sequence, and performing the following operations on the extracted unit time periods: acquiring a plurality of historical images based on the extracted unit time periods, extracting historical images from the plurality of historical images in sequence, and performing the following operations on the extracted historical images: Performing a clearing operation on the extracted historical image to obtain a clear image, performing a denoising operation on the clear image to obtain a denoised image, extracting a set of vehicle images to be identified based on the denoised image, and calculating the extracted unit impact data of the unit time period based on the set of vehicle images to be identified; Summarize the unit impact data to obtain the unit impact data set corresponding to the historical monitoring time, summarize the denoised images to obtain the denoised image sequence corresponding to the historical monitoring time, use the pre-built edge recognition algorithm to identify the denoised image sequence to obtain the initial crack area sequence, and filter the initial crack area sequence to obtain the sequence to be interpolated; The interpolation method is used to interpolate the sequence to be interpolated, and the crack area evolution sequence is obtained; A crack evolution model is constructed based on the crack area evolution sequence and the unit impact data set, and the time period to be predicted and the impact to be predicted are obtained. The time period to be predicted and the impact to be predicted are input into the crack evolution model to obtain the predicted crack value. The prediction of construction road cracks in the unit time period monitoring data is completed based on the predicted crack value.

2. The construction road crack prediction method based on time series monitoring data as claimed in claim 1, characterized in that: The step of performing a clearing operation on the extracted historical image to obtain a clear image includes: Multiple Gaussian difference curves are generated using the pre-built Gaussian difference operator and the extracted historical images, where the Gaussian difference operator is: Among them, z represents the horizontal coordinate of the historical image in the image coordinate system, and w represents the vertical coordinate of the historical image in the image coordinate system. and Both represent the variance of Gaussian filter, σ1 and σ2 both represent the standard deviation of Gaussian filter, and σ1≠σ2, DOG represents the Gaussian difference operator; The zero-crossing point of each Gaussian difference curve in a plurality of Gaussian difference curves is obtained to obtain a plurality of zero-crossing points, a plurality of contour coordinates are generated according to the plurality of zero-crossing points, the plurality of contour coordinates are spliced ​​using a pre-built edge connection algorithm to obtain a historical image contour, an image transition area is obtained according to the historical image contour, and a clear image is obtained according to the image transition area, wherein the contour coordinates correspond one-to-one to the Gaussian difference curve.

3. The construction road crack prediction method based on time series monitoring data as claimed in claim 2, characterized in that: The performing of a noise reduction operation on the clear image to obtain a noise-reduced image includes: Construct a mean filter unit and a median filter unit; A standard image is obtained, and the spatial distribution characteristics of the pixel grayscale values ​​are extracted from the standard image. The spatial distribution characteristics of the pixel grayscale values ​​are introduced into a mean filter unit to obtain a fixed pattern noise, wherein the size of the filter kernel of the mean filter unit is 9×9; A denoising operation is performed on the clear image using fixed pattern noise to obtain a denoised image, wherein the formula for a denoising operation is as follows: k(x,y)=j(x,y)-δ×l(x,y)+β×h Wherein, k(x, y) represents the pixel grayscale value corresponding to the pixel coordinate (x, y) in the primary denoised image, x represents the abscissa of the pixel coordinate (x, y), y represents the ordinate of the pixel coordinate (x, y), j(x, y) represents the pixel grayscale value corresponding to the pixel coordinate (x, y) in the clear image, l(x, y) represents the pixel grayscale value of the pixel coordinate (x, y) in the fixed pattern noise, δ and β represent the first correction coefficient and the second correction coefficient respectively, and δ=0.7, β=1.2, and h represents the average grayscale value of all pixels in the clear image; The median filter unit is used to perform a secondary denoising operation on the primary denoised image to obtain a denoised image, wherein the size of the filter kernel of the median filter unit is 3×3.

4. The construction road crack prediction method based on time series monitoring data as claimed in claim 3, characterized in that: The extracting of a set of vehicle images to be identified based on the denoised image, and calculating the extracted unit impact data of a unit time period based on the set of vehicle images to be identified, comprises: Using a target detection algorithm to identify the denoised image, a set of vehicle images to be identified is obtained, wherein the set of vehicle images to be identified includes a plurality of vehicle images to be identified; The following operations are performed on the vehicle images to be identified in the vehicle image set: Identify the vehicle model of the vehicle image to be identified, and determine whether the license plate number can be extracted from the vehicle image to be identified. If the license plate number can be extracted from the vehicle image to be identified, identify the extracted license plate number to obtain the target license plate number; otherwise, confirm the license plate number of the vehicle image to be identified as 1 to obtain the target license plate number; Construct vehicle data according to vehicle model and target license plate number, summarize vehicle data, and obtain the extracted vehicle data set of unit time period; Calculate unit impact data from the vehicle dataset.

5. The construction road crack prediction method based on time series monitoring data as claimed in claim 4, characterized in that: The calculating unit impact data according to the vehicle data set includes: Extracting vehicle data from the vehicle data set in sequence, and confirming the target license plate number of the extracted vehicle data; If the target license plate number is 1, the extracted vehicle data is confirmed as identification data, and the vehicle data is removed from the vehicle data set; If the target license plate number is not 1, the extracted vehicle data is confirmed as identification data, and the same license plate set is extracted from the vehicle data set, and the same license plate set is removed from the vehicle data set to obtain a retained identification data set, and the retained identification data set is confirmed to be a vehicle data set, and the step of sequentially extracting vehicle data from the vehicle data set is returned until the vehicle data set is an empty set; The identification data are aggregated to obtain an identification data set of the extracted unit time period, and the identification data set is input into a pre-built prediction model to obtain unit impact data of the extracted unit time period.

6. The construction road crack prediction method based on time series monitoring data according to claim 1, characterized in that: The method of using a pre-built edge recognition algorithm to identify the denoised image sequence to obtain an initial crack area sequence includes: The denoised image sequence is screened to obtain an initial crack image sequence, and the following operations are performed on the initial crack images in the initial crack image sequence: A pixel gradient sequence is obtained based on the initial crack image, a first pixel probability, a second pixel probability and a third pixel probability are calculated based on the pixel gradient sequence, a first-category pixel grayscale mean, a second-category pixel grayscale mean and a third-category pixel grayscale mean are calculated according to the first pixel probability, the second pixel probability and the third pixel probability respectively, and a maximum total inter-class variance is calculated using the first pixel probability, the second pixel probability, the third pixel probability, the first-category pixel grayscale mean, the second-category pixel grayscale mean and the third-category pixel grayscale mean; A low threshold and a high threshold are calculated based on the maximum total inter-class variance, and the initial crack image is divided using the low threshold and the high threshold to obtain a strong edge point set and a non-edge point set; Construct a crack image according to the strong edge point set and the non-edge point set, and calculate the initial crack area according to the crack image; The initial crack areas are summarized to obtain the initial crack area sequence.

7. The construction road crack prediction method based on time series monitoring data according to claim 6, characterized in that: The calculation formula of the maximum total inter-class variance is: Where T represents the maximum total inter-class variance, ω j represents the j-th pixel probability, a j represents the mean grayscale value of the j-th class pixel corresponding to the j-th pixel probability, represents the mean gray value of all pixels in the initial crack image, and j represents the j-th type of pixel.

8. The construction road crack prediction method based on time series monitoring data according to claim 6, characterized in that: The method of dividing the initial crack image by using a low threshold and a high threshold to obtain a strong edge point set and a non-edge point set includes: Based on the low threshold and the high threshold, the non-edge pixel interval, the weak edge pixel interval and the strong edge pixel interval are constructed, the crack pixels are extracted from the initial crack image in sequence, and the following operations are performed on the extracted crack pixels: Obtaining a crack grayscale value based on the extracted crack pixels, and comparing the crack grayscale value with the size of the non-edge pixel interval, the weak edge pixel interval, and the strong edge pixel interval; If the crack grayscale value belongs to the non-edge pixel interval, the grayscale value of the extracted crack pixel is output as 0 to obtain a non-edge point; If the crack gray value belongs to the weak edge pixel interval, a judgment neighborhood is obtained based on the extracted crack pixels, and the maximum neighborhood pixel gray value is extracted from the judgment neighborhood. If the maximum neighborhood pixel gray value is greater than the high threshold, the gray value of the extracted crack pixel is output as 255 to obtain a strong edge point. If the maximum neighborhood pixel gray value is not greater than the high threshold, the gray value of the extracted crack pixel is output as 0 to obtain a non-edge point. If the crack gray value belongs to the strong edge pixel interval, the gray value of the extracted crack pixel is output as 255 to obtain a strong edge point; Summarize the strong edge points to get the strong edge point set, and summarize the non-edge points to get the non-edge point set.

9. The construction road crack prediction method based on time series monitoring data according to claim 8, characterized in that: The method of using the interpolation method to perform interpolation calculation on the interpolation sequence to obtain the fracture area evolution sequence includes: Constructing an interpolation model, wherein the interpolation model includes: a generation unit and a discrimination unit, wherein the discrimination unit includes a main discrimination unit and an auxiliary discrimination unit; Based on the interpolation model, the generation unit is started, and according to the initial fracture area sequence and the sequence to be interpolated and the missing data set, the missing data set, the initial fracture area sequence and the sequence to be interpolated are input into the started generation unit to obtain a plurality of initial interpolation areas, and the plurality of initial interpolation areas are inserted into the sequence to be interpolated to obtain a filled sequence, wherein the missing data set includes a plurality of missing data, and the initial interpolation areas correspond to the missing data one by one; Detect the data distribution of the sequence to be interpolated to obtain the original distribution, detect the data distribution of the filled sequence to obtain the filled distribution, and import the original distribution and the filled distribution into the discriminant unit; If the original distribution is consistent with the filled distribution, the filled sequence is confirmed as the fracture area evolution sequence based on the auxiliary discrimination unit; If the original distribution is inconsistent with the filled distribution, a missing value set is identified from the multiple initial interpolation areas based on the main discrimination unit, the missing value set is confirmed as a missing data set, and the step of inputting the missing data set, the initial crack area sequence and the sequence to be interpolated into the generation unit is returned until the filled sequence is confirmed as a crack area evolution sequence based on the auxiliary discrimination unit.

10. A construction road crack prediction system based on time series monitoring data, characterized in that: The system comprises: The historical image acquisition module is used to receive a road monitoring instruction, start the road monitoring device according to the road monitoring instruction, obtain the historical monitoring time based on the started road monitoring device, divide the historical monitoring time to obtain a plurality of unit time periods, extract the unit time periods from the plurality of unit time periods in sequence, and perform the following operations on the extracted unit time periods: obtain a plurality of historical images based on the extracted unit time periods; The impact data acquisition module is used to extract historical images from a plurality of historical images in sequence, and perform the following operations on the extracted historical images: perform a clearing operation on the extracted historical images to obtain a clear image, perform a noise reduction operation on the clear image to obtain a noise reduction image, extract a set of vehicle images to be identified based on the noise reduction image, and calculate the extracted unit impact data of the unit time period based on the set of vehicle images to be identified; The crack area evolution sequence module is used to summarize the unit impact data to obtain the unit impact data set corresponding to the historical monitoring time, summarize the denoised images to obtain the denoised image sequence corresponding to the historical monitoring time, use the pre-built edge recognition algorithm to identify the denoised image sequence to obtain the initial crack area sequence, screen the initial crack area sequence to obtain the sequence to be interpolated, and use the interpolation method to perform interpolation calculation on the sequence to be interpolated to obtain the crack area evolution sequence; The crack prediction module is used to build a crack evolution model based on the crack area evolution sequence and the unit impact data set, obtain the time period to be predicted and the impact to be predicted, input the time period to be predicted and the impact to be predicted into the crack evolution model, obtain the predicted crack value, and complete the prediction of construction road cracks in the unit time period monitoring data according to the predicted crack value.

Images