Method and system for detecting and estimating lane load of target bridge based on remote sensing image

Abstract

The invention relates to a method and a system for estimating a target bridge lane load based on remote sensing image detection. The method comprises the following steps: collecting a remote sensing image set of a to-be-researched area to extract a bridge and standard load automobile characteristic information thereof; based on the obtained bridge and automobile feature information, obtaining a bridge training sample set and a standard lorry image training sample set of the to-be-researched area according to artificial visual interpretation; training the deep learning model based on the training sample set, and constructing a recognition model; preliminarily identifying and matching corresponding grades of the beam bridge in the to-be-studied area and the standard lorry in the to-be-studied area based on the identification model; obtaining a remote sensing image of the bridge according to the recorded spatial position information based on the target bridge obtained through preliminary matching so as to carry out span extraction and calculation, and obtaining basic information of the target bridge; and estimating a target bridge lane load based on the basic information of the target bridge. The problem of data-lacking bridge vehicle load estimation based on remote sensing images is solved.

Description

Technical Field

[0001] The present application relates to the field of remote sensing detection technology, and in particular to a method and system for estimating target bridge lane load based on remote sensing image detection. Background Art

[0002] As a comprehensive earth observation technology, remote sensing has advantages that other technologies cannot match, such as wide coverage, large amount of information, and strong timeliness. At present, remote sensing technology has been widely used in many fields such as terrain monitoring, water and soil resource surveys, and road monitoring in uninhabited areas or areas with insufficient data. However, in the relevant research on obtaining information in areas without data, various scholars often rely on geographic spatial characteristics and spectral characteristics to obtain parameters and apply them to model or mechanism research. This research technology route is complex and lacks timeliness. Few studies apply remote sensing to the estimation of attribute information such as urban road infrastructure. In the context of the era of remote sensing big data, deep learning methods have strong learning ability, can obtain the characteristic information of ground objects for object discrimination, and have a wide coverage. They can perform a large amount of data processing and calculation through computers. Therefore, they are widely used in many fields such as ground object recognition, scene recognition, and face recognition.

[0003] Although deep learning can obtain information in areas without data based on big data, there are still some problems in its application in urban bridge load estimation. On the one hand, deep learning requires a large number of samples for model training to extract target features, and current technology cannot support the acquisition of attribute information of target objects, such as lane loads of urban roads and bridges. On the other hand, deep learning is easily limited by the quality of remote sensing images when combined with remote sensing images, and often misses the effective features of objects, resulting in inaccurate information acquisition, inability to accurately estimate urban bridge lane loads, and reduced work efficiency. Summary of the invention

[0004] The present application provides a method and system for estimating the lane load of a target bridge based on remote sensing image detection. Through the existing remote sensing image data and actual information, based on the deep learning model, a deep learning model with urban bridges as the main identification object is constructed, so as to realize the grade classification and monitoring of urban bridges, realize the identification and extraction of urban bridges in specific areas, and solve the problem of the lack of database of urban bridges and their standard load-carrying vehicles. Finally, combined with the information such as the clear span of urban bridges and the number of lanes of urban bridges extracted from drone remote sensing images, the lane load capacity of urban bridges is effectively estimated, and the problem of vehicle load estimation of bridges with missing data based on remote sensing images is solved, which provides technical support for obtaining road traffic information, path planning, etc. in special and complex geographical environments, and improves work efficiency.

[0005] In the first aspect, the present application provides a method for estimating the lane load of a target bridge based on remote sensing image detection, comprising the following steps: collecting a remote sensing image set of the area to be studied to extract characteristic information of the bridge and its standard load vehicle; based on the obtained characteristic information of the bridge and the standard load vehicle, obtaining a bridge training sample set and a standard load vehicle image training sample set in the area to be studied according to manual visual interpretation; training a deep learning model based on the training sample set to construct a recognition model; based on the recognition model, preliminary identification and matching of the corresponding grades of the beam bridges and standard load vehicles in the area to be studied are performed to determine the specific grade information of the beam bridges in the area to be studied and record their image spatial position information; based on the target bridge obtained by the preliminary matching, obtaining a remote sensing image of the target bridge by using a drone according to the recorded image spatial position information to obtain basic information of the target bridge, wherein the basic information is used to display basic construction data of the target bridge; estimating the lane load of the target bridge according to the bridge design standard based on the basic information of the target bridge.

[0006] Optionally, a remote sensing image set of the area to be studied is collected to extract characteristic information of bridges and their standard load vehicles, including: collecting the actual bridge construction environment in the area to be studied to obtain corresponding remote sensing images; performing data preprocessing on the remote sensing images, and cropping them according to visual interpretation to obtain a remote sensing image set of the area to be studied to extract characteristic information of bridges and their standard load vehicles in the area to be studied.

[0007] Optionally, based on the obtained bridge and standard load vehicle characteristic information, a bridge training sample set and a standard load vehicle image training sample set of the area to be studied are obtained according to manual visual interpretation, including: obtaining typical images of bridges containing lane number information in the area to be studied as a bridge training sample set based on the bridge characteristic information and manual visual interpretation; obtaining images of standard load vehicles of different grades in the area to be studied as standard load vehicle image training sample sets based on the standard load vehicle characteristic information and combining prior knowledge and vehicle management data, wherein the standard load vehicles include Class A standard load vehicles and Class B standard load vehicles.

[0008] Optionally, the deep learning model is trained based on the training sample set to construct a recognition model, including: taking the training sample set as input data based on a U-Net model, training and optimizing the model to construct a recognition model.

[0009] Optionally, based on the recognition model, preliminary identification and matching are performed on the corresponding grades of the beam bridges and standard trucks in the study area to determine the specific grade information of the beam bridges in the study area and record the image spatial position information thereof, including: determining whether there is a beam bridge in the image to be studied, and if so, verifying the time attribute of the identified beam bridge, determining the time range for bridge construction to match the calculation formula in the bridge design standard determined within the corresponding time, otherwise, discarding the image; based on the determination of the presence of a beam bridge, continuing to determine whether there is a Class A standard truck and / or a Class B standard truck in the image to be studied, and if not, discarding the image; if so, if a beam bridge and a Class A standard truck appear in the image at the same time, the bridge is determined to be a Class A bridge; if a beam bridge, a Class A standard truck and a Class B standard truck appear in the image at the same time, the bridge is determined to be a Class A bridge; if a beam bridge, a Class A standard truck and a Class B standard truck appear in the image at the same time, the bridge is determined to be a Class A bridge; if a beam bridge and a Class B standard truck appear in the image at the same time, the bridge is determined to be a Class B bridge; recording the spatial position information of the beam bridge that has passed the determination and verification of the time attribute.

[0010] Optionally, it is determined whether there is a beam bridge in the image to be studied. If so, the time attribute of the identified beam bridge is verified, and the time range of the bridge construction is determined to match the calculation formula in the bridge design standard determined within the corresponding time, including: based on the image to be studied, first confirm whether there is a beam bridge. If not, continue to select the image to be studied at a specific time interval according to demand to confirm whether there is a beam bridge; if so, obtain the image time of the image to be studied; based on the obtained image time, determine the time range of the beam bridge when it is designed or constructed to match the calculation formula in the bridge design standard determined within the corresponding time, wherein the bridge design standard is continuously updated and optimized over time, and each set of bridge design standards has different regulations on standard trucks.

[0011] Optionally, based on the target bridge obtained through preliminary matching, a remote sensing image of the target bridge is obtained by using a drone according to the recorded spatial position information of the image to obtain basic information of the target bridge, including: obtaining a two-dimensional side view image of the target bridge in the area to be studied, and performing image processing; obtaining the clear span of the target bridge according to the distance calculation in the image based on the scaling ratio of the drone remote sensing image, and obtaining the number of lanes and piers through manual visual interpretation to obtain basic information of the target bridge, wherein the basic information includes the classification of bridges in the area to be studied, the clear span of bridges, the number of piers, and the number of lanes.

[0012] Optionally, the lane load of the target bridge is estimated according to the bridge design standard based on the basic information of the target bridge, including: substituting the obtained basic information of the target bridge into the bridge lane load calculation formula in the corresponding bridge design standard to estimate the lane load of the target bridge, and the calculation formula is expressed as: total vehicle load of target bridge = number of lanes × single lane load × load coefficient × number of spans.

[0013] In the second aspect, the present application provides a system for estimating the lane load of a target bridge based on remote sensing image detection, the system comprising: an acquisition module, used to collect a remote sensing image set of the area to be studied to extract characteristic information of the bridge and its standard load vehicle; a training sample module, used to obtain a bridge training sample set and a standard load vehicle image training sample set in the area to be studied based on the obtained characteristic information of the bridge and the standard load vehicle according to manual visual interpretation; a model construction module, used to train the deep learning model based on the training sample set to construct a recognition model; a matching module, used to perform preliminary identification and matching of the corresponding grades of the beam bridge in the area to be studied and the standard load vehicle in the area to be studied based on the recognition model, so as to determine the specific grade information of the beam bridge in the area to be studied and record its image spatial position information; a span extraction module, used to obtain the remote sensing image of the target bridge through a drone based on the recorded image spatial position information based on the target bridge obtained by preliminary matching, so as to obtain the basic information of the target bridge, wherein the basic information is used to display the basic construction data of the target bridge; a load estimation module, used to estimate the lane load of the target bridge according to the bridge design standard based on the basic information of the target bridge.

[0014] In a third aspect, the present application also provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-described method when executing the computer program.

[0015] In a fourth aspect, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above-mentioned method are implemented.

[0016] This application has at least the following advantages:

[0017] By collecting remote sensing image sets and actual data of the area to be studied, the characteristic information of bridges and standard load vehicles in the area to be studied is obtained, and a training sample set is obtained. The recognition model is constructed in combination with the deep learning algorithm to identify and match beam bridges and vehicles, and to judge the grade of beam bridges. The recognition and extraction of bridges in specific areas is realized, and the problem of the lack of a database of urban bridges and their standard load-carrying vehicles is solved. Then, UAV remote sensing images are used to supplement and verify satellite remote sensing images to obtain basic information such as the span, length, and width of the target bridge. According to the bridge design standards corresponding to the target bridge, the lane load capacity of the bridge in the area to be studied is effectively estimated, providing technical support for obtaining road traffic information and path planning in special and complex geographical environments, thereby improving work efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A diagram showing an application environment of a method for estimating a target bridge lane load based on remote sensing image detection in one embodiment;

[0019] Figure 2 A schematic diagram of a flow chart showing steps of a method for estimating a target bridge lane load based on remote sensing image detection in one embodiment;

[0020] Figure 3 A flowchart showing a method for estimating a target bridge lane load based on remote sensing image detection in one embodiment;

[0021] Figure 4 A flowchart showing the process of extracting characteristic information of a bridge and its standard load vehicle in one embodiment;

[0022] Figure 5 A flowchart showing the construction of a training sample set in one embodiment;

[0023] Figure 6 A schematic diagram showing a multi-span beam bridge in one embodiment;

[0024] Figure 7 A sample diagram showing a standard truck for bridge design in one embodiment;

[0025] Figure 8 A schematic diagram of identification and matching of bridges and standard trucks in a research area in one embodiment;

[0026] Fig. 9 A schematic diagram of lane load in one embodiment;

[0027] Fig.10 A structural block diagram showing a system for estimating a target bridge lane load based on remote sensing image detection in one embodiment;

[0028] Fig.11 FIG. 4 is a schematic structural diagram of a computer device in one embodiment. DETAILED DESCRIPTION

[0029] The present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0030] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when "including" and / or "comprising" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0031] To make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the embodiments of the present application will be described in detail below in conjunction with the accompanying drawings. However, it will be appreciated by those skilled in the art that in the embodiments of the present application, many technical details are proposed in order to enable the reader to better understand the present application. However, even without these technical details and various changes and modifications based on the following embodiments, the technical scheme claimed in the present application can also be implemented. The markings of the following embodiments are for convenience of description and should not constitute any limitation to the specific implementation of the present application. The various embodiments can be combined with each other and quoted from each other without contradiction.

[0032] Figure 2 A flow chart of a method for estimating target bridge lane load based on remote sensing image detection provided in an embodiment of the present application. The method can be performed as follows: Figure 1 The user space file server is executed in the system shown. Figure 2 , Figure 3 As shown, the method may include the following steps:

[0033] S201, collecting a remote sensing image set of the area to be studied to extract characteristic information of the bridge and its standard load vehicle;

[0034] S202, based on the obtained bridge and standard load vehicle characteristic information, obtain a bridge training sample set and a standard load vehicle image training sample set of the area to be studied according to manual visual interpretation;

[0035] S203, training the deep learning model based on the training sample set to build a recognition model;

[0036] S204, preliminarily identifying and matching the corresponding grades of the beam bridges and standard trucks in the study area based on the recognition model, so as to determine the specific grade information of the beam bridges in the study area and record the spatial position information of the images;

[0037] S205, based on the target bridge obtained by preliminary matching, a remote sensing image of the target bridge is obtained by using a drone according to the recorded image spatial position information to obtain basic information of the target bridge, wherein the basic information is used to display basic construction data of the target bridge;

[0038] S206. Estimate the lane load of the target bridge based on the basic information of the target bridge and the bridge design standard.

[0039] This application mainly provides an urban bridge load estimation method based on deep learning and target detection of geographic thinking for existing remote sensing image data and actual data, realizes the grade classification and monitoring of urban bridges, and finally, combines the urban bridge clear span, urban bridge lane number and other information extracted from UAV remote sensing images to effectively estimate the lane load capacity of urban bridges. On the basis of simulation and verification in the study area with the city as the scale, it is applied to the identification of bridges in special scenarios, providing technical support for collecting information on environmental elements in special scenarios.

[0040] The following is a detailed description of each step:

[0041] Please refer to Figure 2 , Figure 3 As shown, step S201, collecting a remote sensing image set of the area to be studied to extract characteristic information of the bridge and its standard load vehicle;

[0042] In this embodiment, it should be noted that the actual bridge construction environment in the area to be studied is collected to obtain the corresponding remote sensing images; the remote sensing images are preprocessed and cropped according to visual interpretation to obtain a sample study area remote sensing image set, including a large parking lot and target objects of bridges in the area to be studied, so as to extract the characteristic information of the bridges and their standard load vehicles in the area to be studied.

[0043] Specifically, refer to Figure 4 As shown in the figure, the satellite remote sensing data collected is the Sentinel-1 series data. After image preprocessing such as atmospheric correction, noise removal, impact enhancement and fusion, the remote sensing image set in the area to be studied is obtained by cropping based on visual interpretation, which serves as the basic data for the subsequent construction of various sample data sets. Manual interpretation is currently the most commonly used image extraction method in China, such as land survey, geological survey, water conservancy survey, etc. Specifically, remote sensing images represent the differences between different objects through the high and low differences and spatial changes of brightness values ​​or pixel values, which is the physical basis for distinguishing different image objects. The remote sensing image set here should contain target objects such as large parking lots and urban bridges in space. In order to ensure the accuracy of target recognition results, the remote sensing images to be tested should have high temporal resolution (coverage period <7 days), and the satellite monitoring time range should be no less than 30 days.

[0044] Please continue to refer to Figure 2-Figure 5 As shown, step S202, based on the obtained bridge and standard load vehicle feature information, obtain a bridge training sample set and a standard load vehicle image training sample set of the area to be studied according to manual visual interpretation, including:

[0045] Step S2021: based on the bridge feature information and according to manual visual interpretation, typical images of bridges in the area to be studied containing information on the number of lanes are obtained as a bridge training sample set;

[0046] In this embodiment, it should be noted that typical images of urban bridges containing lane number information are obtained from the remote sensing image set of the sample study area based on manual visual interpretation as the bridge training sample set of the area to be studied, such as Figure 6 As shown, the bridges in the area to be studied should be beam bridges, that is, single-span or multi-span bridges with consistent span standards and using beams or truss beams as the main load-bearing structure.

[0047] Step S2022: Perform manual visual interpretation based on the characteristic information of standard load vehicles and combine prior knowledge and vehicle management data to obtain images of standard load vehicles of different grades in the study area as standard load vehicle image training sample sets, wherein standard load vehicles include Class A standard load vehicles and Class B standard load vehicles.

[0048] In this embodiment, it should be noted that, in addition, in combination with the remote sensing image set of the sample study area, remote sensing images of actual large parking lots, such as vehicle parking areas of car 4S stores, are extracted, and standard load vehicle images are obtained as training sample sets based on manual visual interpretation.

[0049] Among them, according to industry standards, urban bridges are divided into two categories according to traffic capacity and design standards: "Urban-A Class" and "Urban-B Class". Urban-A Class bridges usually have a large traffic capacity and can accommodate a large number of vehicles or pedestrians. They often involve important nodes in the urban transportation network. Whether they are smooth or not directly affects the smoothness and efficiency of urban transportation. Urban-B Class bridges are usually used to connect different areas, streets or important facilities in the city. They are major secondary traffic arteries or channels connecting important areas. For bridges of different levels, corresponding levels of load vehicles can be allowed to pass. Therefore, by obtaining images of urban-A Class standard load vehicles and urban-B Class standard load vehicles and combining prior knowledge and vehicle management data as training sample sets, the training sample sets include A Class standard load vehicle training sample sets and B Class standard load vehicle training sample sets.

[0050] Specifically, Figure 7As shown in the figure, the design standard data of trucks corresponding to the A-class and B-class bridge design standards are obtained by combining prior knowledge and vehicle management data, including the length, width, height and actual load of standard trucks and other vehicle characteristic information standards. Based on the obtained vehicle information standards, the standard load vehicle images obtained from the remote sensing vehicle images are divided into the A-class standard truck training sample set and the B-class standard truck training sample set according to their length, width, height and other vehicle characteristic information standards.

[0051] Reference Figure 2 , Figure 3 As shown, step S203, training the deep learning model based on the training sample set to construct a recognition model, includes: taking the training sample set as input data based on the U-Net model, training the optimization model to construct the recognition model.

[0052] In this embodiment, it should be noted that the deep learning model adopts the U-Net model, which is a convolutional neural network model, mainly used for processing tasks such as image recognition and segmentation, and can extract local features and spatial relationships of images and videos. The model data set contains only positive samples and performs data enhancement training through steps such as random cropping, random inversion, random scaling and random color transformation. It is necessary to set the sliding window and hyperparameters according to the remote sensing image to be studied, and select an effective algorithm for target recognition. It consists of two parts, an encoder Q and a decoder Q, which are connected by a jump connection in the middle. The encoder (Encoder) is responsible for downsampling and feature extraction. It gradually reduces the feature map size of the input image through a series of convolution and maximum pooling operations while increasing the abstraction of the features. After each downsampling, the size of the feature map is halved and the number of channels is doubled. Decoder (Decoder): used for image size restoration and segmentation. It restores the size of the feature map to the original input size through transposed convolution and convolution layers, and uses convolution layers to extract high-level features of the input image. The structure of the decoder is symmetrical with the encoder, and the size of the feature map is gradually increased through upsampling operations while reducing the number of channels. The Skip Connection crops the feature map of the corresponding layer in the encoder and concatenates it with the feature map of the corresponding layer in the decoder. This connection helps to fuse image features of the same layer and size, helping the decoder to better restore the detailed information of the image.

[0053] The three types of training sample data sets constructed are input into the U-Net model respectively. During the training process, the model preprocesses the input image, i.e., the training sample set, such as normalization and resizing. In order to ensure the accuracy of model training, the binary cross-entropy loss function (Binary Cross-Entropy Loss) is combined with the Dice loss function, which is not only more sensitive to boundary clarity and the model's prediction accuracy for small areas, but also relatively easy to calculate the gradient, which is conducive to the model training and optimization process. In addition, the training process combines multiple optimizers and learning strategies, continuously adjusts the optimization parameters, and selects the optimal solution to determine the final recognition model.

[0054] Reference Figure 2 , Figure 3 As shown, step S204 is to preliminarily identify and match the corresponding grades of the beam bridges and standard trucks in the study area based on the recognition model to determine the specific grade information of the beam bridges in the study area and record their image space position information.

[0055] In this embodiment, it should be noted that the trained deep learning model is used to perform preliminary identification of bridges and standard trucks in the area to be studied within a specific time period to determine the specific grade information of the beam bridges in the area to be studied and record their image spatial position information. To ensure the accuracy of the target recognition results, the remote sensing image to be tested should have a high temporal resolution (coverage period <7 days), and the satellite monitoring time range should be no less than 30 days. The identified beam bridge is matched with the standard truck to obtain the spatial feature information of the target object.

[0056] Preliminary identification and matching of beam bridges and standard truck grades in the study area, such as Figure 8 As shown, first determine whether there is a beam bridge in the image to be studied. If so, verify and identify the time attribute of the beam bridge, determine the time range of bridge construction, and match the calculation formula in the bridge design standard determined within the corresponding time. Otherwise, discard the image, or reselect a standard truck for ground feature information extraction, and retrain the deep learning model.

[0057] Among them, the time attribute of the beam bridge obtained by verification and identification is firstly to confirm whether there is a beam bridge based on the image to be studied. If not, the image to be studied is continuously selected at a specific time interval according to the needs until it is determined that there is a beam bridge, and then the time of the image to be studied is obtained; based on the obtained time of the image to be studied, the time range of the beam bridge during design or construction is determined to match the calculation formula in the bridge design standard determined in the corresponding time. Among them, the bridge design standard is continuously updated and optimized over time, and each set of bridge design standards has different provisions for standard trucks. Therefore, the urban bridge design standard should be selected according to the time scale of the image to match the bridge design standard determined in the corresponding time.

[0058] Specifically, there are standards such as the 1993 Urban Bridge Design Criteria (CJJ 11-1993), the 1998 Urban Bridge Design Load Standard (CJJ 77-1998), the 2011 Urban Bridge Design Code (CJJ 11-2011), and the current Highway Bridge Design Code (GB 50010) and Urban Road Bridge Design Code (JTGD J11-2005 / 2014). Each set of standards has slightly different provisions for standard trucks.

[0059] In one example, image A is selected from the remote sensing image library. The time of image A is June 2001. The target bridge is not found in the recognition of image A. Then image B is selected from the remote sensing image library for target recognition. The time of image B is 2002. And so on, image C, image D, etc. are recognized. The image selection time interval can be 1 year or 2 years, which can be selected according to the needs until the target bridge is recognized in image N. Assuming that the acquisition time of image N is March 2010, if the target bridge is recognized, it is determined that the construction time of this bridge is up to 2010. At this point, the "Bridge Construction Specifications" used when the bridge was built can be matched, and then subsequent calculations can be performed.

[0060] Secondly, on the basis of judging whether there is a beam bridge, it is further judged whether there is a Class A standard truck and / or a Class B standard truck in the image to be studied. If not, the image is discarded; if so, if a beam bridge and a Class A standard truck appear in the image at the same time, the bridge is judged as a Class A bridge; if a beam bridge, a Class A standard truck and a Class B standard truck appear in the image at the same time, the bridge is judged as a Class A bridge; if a beam bridge and a Class B standard truck appear in the image at the same time, the bridge is judged as a Class B bridge;

[0061] In this embodiment, it should be noted that Class A bridges are identified first. If they are not Class A bridges, Class B bridges are identified because the load capacity of Class A bridges is better than that of Class B bridges. If Class A and Class B standard trucks appear at the same time, it means that the bridge is at least a Class A bridge.

[0062] Finally, the spatial position information of the beam bridge that has been identified and verified by time attributes, that is, the longitude and latitude position information, is recorded to locate the target bridge so that the bridge span can be obtained using a drone in the next step.

[0063] like Figure 2 , Figure 3 As shown, step S205, based on the target bridge obtained by preliminary matching, a remote sensing image of the target bridge is obtained by using a drone according to the recorded image spatial position information to obtain basic information of the target bridge;

[0064] In this embodiment, it should be noted that a multi-rotor drone is selected to photograph the beam bridge in the historical time verification identification result, obtain a two-dimensional side view image of the target bridge in the study area, and perform image processing, including image correction, image enhancement, etc.; based on the scaling ratio of the drone remote sensing image, the clear distance between two adjacent piers (or abutments) at the design flood level of the target bridge, that is, the clear span of the target bridge, is obtained according to the distance calculation in the image, and at the same time, the number of lanes and piers is obtained through manual visual interpretation to obtain basic information of the target bridge, such as Figure 6 As shown, the basic information includes the classification of bridges in the area to be studied, the clear span of bridges, the number of piers, and the number of lanes.

[0065] like Figure 2 , Figure 3 As shown, step S206, based on the basic information of the target bridge and according to the bridge design standard, the lane load of the target bridge is estimated.

[0066] In this embodiment, it should be noted that the basic information of the target bridge is brought into the bridge lane load calculation formula in the corresponding bridge design standard to estimate the target bridge lane load. The calculation formula is expressed as: total vehicle load of the target bridge = number of lanes × single lane load × load coefficient × number of spans.

[0067] Among them, Fig. 9 As shown in the figure, the vehicle load on the bridge is divided into “concentrated load (P k )” and “Uniform load (q k )". Concentrated load refers to a local concentrated load acting on the bridge, usually represented by a point load or a load concentrated in a smaller area, such as a moving vehicle, pedestrian, equipment, etc. Uniformly distributed load refers to a uniformly distributed load acting on the bridge structure. Unlike concentrated load, uniformly distributed load is a load uniformly distributed along a certain length or area. Bending moment refers to the ability or result of a material or structure to bend when subjected to external force. When an object is subjected to lateral force or bending moment, it bends, and the moment generated at this time is called bending moment. Bending moment is used to describe the bending state and stress conditions of various parts of the structure, usually a moment along the axis of the cross section. Shear force refers to a force acting on a structure or material, which produces a shear effect along the parallel direction of the cross section of the structure. In structures such as beams or bridges, shear force is a force parallel to the force plane along the cross section, which can be used to characterize the shear force in the stress state of the structure.

[0068] The reference examples of urban bridge single lane load standards and estimation formulas are as follows:

[0069] Table 1 Urban bridge single lane load standards and estimation formulas in "Urban Bridge Design Load Standards" (CJJ 77-1998):

[0070] Table 1

[0071]

[0072] Table 2 Urban bridge single lane load standards and estimation formulas in the "Urban Bridge Design Code" (CJJ 11-2011):

[0073] Table 2

[0074]

[0075] Implementation principle of this embodiment: The above steps are mainly based on obtaining the bridge and car feature information in the area to be studied by collecting remote sensing image sets in the area to be studied and combining prior knowledge and vehicle management data to obtain a training sample set, in which the car load is divided into Class A standard truck and Class B standard truck to match different grades of bridges, so the training sample set is divided into bridge training sample set, Class A standard truck training sample set, and Class B standard truck training sample set. Combined with the deep learning algorithm, a recognition model is constructed based on the classified training sample set to identify and match the beam bridge and the car grade, and determine whether the grade of the beam bridge is a Class A bridge or a Class B bridge, thereby realizing the recognition and extraction of urban bridges in a specific area and solving the problem of the lack of a database of urban bridges and their standard trucks. Then, UAV remote sensing images are used to supplement and verify satellite remote sensing images to obtain basic information such as the span, length, and width of the target bridge. According to the bridge design standards corresponding to the target bridge, the lane load capacity of the urban bridge is effectively estimated, which solves the problem of vehicle load estimation on bridges with insufficient data based on remote sensing images. It provides technical support for obtaining road traffic information and route planning in special and complex geographical environments, thereby improving work efficiency.

[0076] Reference Fig.10 As shown, the present application also provides a system for estimating the target bridge lane load based on remote sensing image detection, which may include: an acquisition module 301, a training sample module 302, a model building module 303, a matching module 304, a span extraction module 305, and a load estimation module 306. The main functions of each component module are as follows:

[0077] An acquisition module 301 is used to collect a remote sensing image set of the area to be studied to extract characteristic information of the bridge and its standard load vehicle;

[0078] A training sample module 302 is used to obtain a bridge training sample set and a standard load vehicle image training sample set of the area to be studied based on the obtained bridge and standard load vehicle feature information and according to manual visual interpretation;

[0079] A model building module 303 is used to train the deep learning model based on the training sample set to build a recognition model;

[0080] Matching module 304, used for preliminarily identifying and matching the beam bridges in the study area and the corresponding grades of the standard trucks in the study area based on the recognition model, so as to determine the specific grade information of the beam bridges in the study area and record the image space position information thereof;

[0081] The span extraction module 305 obtains the remote sensing image of the target bridge through the drone according to the recorded image spatial position information based on the target bridge obtained by preliminary matching to obtain basic information of the target bridge, wherein the basic information is used to display the basic construction data of the bridge;

[0082] The load estimation module 306 is used to estimate the lane load of the target bridge based on the basic information of the target bridge according to the bridge design standard.

[0083] like Fig.11 , is a block diagram of a computer device according to an embodiment of the present application. The computer device is intended to represent various forms of digital computers or mobile devices. The digital computer may include a desktop computer, a portable computer, a workbench, a personal digital assistant, a server, a mainframe computer, and other suitable computers. The mobile device may include a tablet computer, a smart phone, a wearable device, etc.

[0084] like Fig.11 As shown, the device 600 includes a computing unit 601, a ROM 602, a RAM 603, a bus 604, and an input / output (I / O) interface 605, and the computing unit 601, ROM 602 and RAM 603 are connected to each other through the bus 604. The input / output (I / O) interface 605 is also connected to the bus 604.

[0085] The computing unit 601 can perform various processes in the method embodiment of the present application according to the computer instructions stored in the read-only memory (ROM) 602 or the computer instructions loaded from the storage unit 608 to the random access memory (RAM) 603. The computing unit 601 can be various general and / or special processing components with processing and computing capabilities. The computing unit 601 may include, but is not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. In some embodiments, the method provided in the embodiment of the present application can be implemented as a computer software program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 608.

[0086] RAM 603 may also store various programs and data required for the operation of device 600. Part or all of a computer program may be loaded and / or installed on device 600 via ROM 602 and / or communication unit 609.

[0087] The input unit 606, output unit 607, storage unit 608 and communication unit 609 in the device 600 can be connected to the I / O interface 605. The input unit 606 can be, for example, a keyboard, a mouse, a touch screen, a microphone, etc. The output unit 607 can be, for example, a display, a speaker, an indicator light, etc. The device 600 can exchange information, data, etc. with other devices through the communication unit 609.

[0088] It should be noted that the device may also include other components necessary for normal operation, or may only include components necessary for implementing the solution of the present application, rather than all the components shown in the figure.

[0089] Various implementations of the systems and techniques described herein can be realized in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof.

[0090] The computer instructions for implementing the method of the present application can be written in any combination of one or more programming languages. These computer instructions can be provided to the computing unit 601, so that when the computer instructions are executed by the computing unit 601 such as a processor, the steps involved in the method embodiment of the present application are executed.

[0091] The computer-readable storage medium provided in the present application may be a tangible medium that may contain or store computer instructions for executing the steps involved in the method embodiments of the present application. The computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, and other forms of storage media.

[0092] The above specific implementations do not constitute a limitation on the protection scope of this application. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of this application should be included in the protection scope of this application.

Claims

1. A method for estimating target bridge lane load based on remote sensing image detection, characterized in that: The following steps are involved: A remote sensing image set of the area to be studied is collected to extract characteristic information of bridges and standard load vehicles; based on the obtained characteristic information of bridges and standard load vehicles, a bridge training sample set and a standard load vehicle image training sample set of the area to be studied are obtained according to manual visual interpretation; Training the deep learning model based on the training sample set to build a recognition model; Based on the recognition model, preliminary recognition and matching of the corresponding grades of the beam bridges and standard trucks in the study area are performed to determine the specific grade information of the beam bridges in the study area and record the image spatial position information thereof; Based on the target bridge obtained by preliminary matching, a remote sensing image of the target bridge is obtained by using a drone according to the recorded spatial position information of the image, so as to obtain basic information of the target bridge, wherein the basic information is used to display basic construction data of the target bridge; The lane load of the target bridge is estimated based on the basic information of the target bridge according to the bridge design standard.

2. The method for estimating target bridge lane load based on remote sensing image detection according to claim 1 is characterized in that: The remote sensing image set of the area to be studied is collected to extract the characteristic information of the bridge and its standard load vehicle, including: Collect the actual bridge construction environment in the area to be studied and obtain the corresponding remote sensing images; The remote sensing images are preprocessed and cropped according to visual interpretation to obtain a remote sensing image set of the area to be studied, so as to extract characteristic information of the bridges and standard load vehicles in the area to be studied.

3. The method for estimating target bridge lane load based on remote sensing image detection according to claim 2 is characterized in that: Based on the obtained bridge and standard load vehicle feature information, a bridge training sample set and a standard load vehicle image training sample set of the area to be studied are obtained according to manual visual interpretation, including: Based on the bridge feature information and manual visual interpretation, typical images of bridges in the study area containing information on the number of lanes are obtained as a bridge training sample set; Based on the characteristic information of standard load vehicles, manual visual interpretation is performed and combined with prior knowledge and vehicle management data, images of standard load vehicles of different grades in the study area are obtained as standard load vehicle image training sample sets, wherein the standard load vehicles include Class A standard load vehicles and Class B standard load vehicles.

4. The method for estimating target bridge lane load based on remote sensing image detection according to claim 3 is characterized in that: The step of training the deep learning model based on the training sample set to construct a recognition model includes: Based on the U-Net model, the training sample set is used as input data to train and optimize the model to build a recognition model.

5. The method for estimating target bridge lane load based on remote sensing image detection according to claim 3 or 4, characterized in that: The preliminary identification and matching of the corresponding grades of the beam bridges and standard trucks in the area to be studied based on the identification model to determine the specific grade information of the beam bridges in the area to be studied and record the image space position information thereof includes: Determine whether there is a beam bridge in the image to be studied. If so, verify the time attribute of the identified beam bridge and determine the time range of bridge construction to match the calculation formula in the bridge design standard determined within the corresponding time. Otherwise, discard the image. On the basis of determining whether there is a beam bridge, further determine whether there is a Class A standard truck and / or a Class B standard truck in the image to be studied, and if not, discard the image; If yes, if a beam bridge and a Class A standard truck appear in the image at the same time, the bridge is determined to be a Class A bridge; If a beam bridge, a Class A standard truck, and a Class B standard truck appear in the image at the same time, the bridge is determined to be a Class A bridge; If a beam bridge and a Class B standard truck appear in the image at the same time, the bridge is determined to be a Class B bridge; The spatial position information of the beam bridge after the time attribute is determined and verified is recorded.

6. The method for estimating target bridge lane load based on remote sensing image detection according to claim 5 is characterized in that: The determining whether there is a beam bridge in the image to be studied, and if so, verifying the time attribute of the identified beam bridge, and determining the time range of bridge construction to match the calculation formula in the bridge design standard determined within the corresponding time, specifically includes: Based on the image to be studied, first confirm whether there is a beam bridge. If not, continue to select the image to be studied at a specific time interval according to the needs to confirm whether there is a beam bridge. If yes, then obtaining the image time of the image to be studied; Based on the acquired image time, the time range of the beam bridge during design or construction is determined to match the calculation formula in the bridge design standard determined within the corresponding time, wherein the bridge design standard is continuously updated and optimized over time, and each set of bridge design standards has different provisions for standard trucks.

7. The method for estimating target bridge lane load based on remote sensing image detection according to claim 5 is characterized in that: The target bridge obtained based on the preliminary matching obtains a remote sensing image of the target bridge by using a drone according to the recorded spatial position information of the image to obtain basic information of the target bridge, including: Obtain a two-dimensional side view image of the target bridge in the area to be studied and perform image processing; Based on the scaling ratio of the UAV remote sensing image, the clear span of the target bridge is calculated according to the distance in the image. Meanwhile, the number of lanes and piers is obtained through manual visual interpretation to obtain the basic information of the target bridge, wherein the basic information includes the classification of bridges, clear span of bridges, number of piers, and number of lanes in the area to be studied.

8. The method for estimating target bridge lane load based on remote sensing image detection according to claim 7 is characterized in that: The estimating the lane load of the target bridge based on the basic information of the target bridge according to the bridge design standard includes: The basic information of the target bridge obtained is brought into the bridge lane load calculation formula in the corresponding bridge design standard to estimate the target bridge lane load. The calculation formula is expressed as: total vehicle load of target bridge = number of lanes × single lane load × load coefficient × number of spans.

9. A system for estimating target bridge lane load based on remote sensing image detection, characterized in that: include: An acquisition module is used to collect a remote sensing image set of the area to be studied to extract characteristic information of the bridge and its standard load vehicle; A training sample module is used to obtain a bridge training sample set and a standard truck image training sample set in the area to be studied based on the obtained characteristic information of the bridge and the standard truck according to manual visual interpretation; A model building module is used to train the deep learning model based on the training sample set to build a recognition model; A matching module, used for preliminarily identifying and matching the corresponding grades of the beam bridges in the study area and the standard trucks in the study area based on the recognition model, so as to determine the specific grade information of the beam bridges in the study area and record the image space position information thereof; A span extraction module is used to obtain a remote sensing image of the target bridge obtained by preliminary matching through a drone according to the recorded spatial position information of the image, so as to obtain basic information of the target bridge, wherein the basic information is used to display basic construction data of the target bridge; The load estimation module is used to estimate the lane load of the target bridge based on the basic information of the target bridge according to the bridge design standard.

10. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 8 are implemented.

Images