Building surveying and mapping method, device, equipment, storage medium and product

Abstract

The invention discloses a building surveying and mapping method, device and equipment, a storage medium and a product, and relates to the technical field of image processing, and the method comprises the steps: obtaining a multi-view original observation image based on an aerial image; inputting the multi-view original observation image into a target building segmentation model, and performing feature extraction and semantic segmentation on the multi-view original observation image through a target surveying and mapping model to obtain a target building segmentation result; and extracting contour information of the target building based on the building segmentation result, and surveying and mapping the target building according to the contour information of the target building. By directly carrying out feature extraction and semantic segmentation on the multi-view original observation image, the original observation data in the photogrammetry project can be utilized more comprehensively, and the building surveying and mapping accuracy and surveying and mapping efficiency are improved. The target building contour information is extracted based on the building segmentation result, the accuracy of the edge contour in building surveying and mapping is improved, and the accuracy of building surveying and mapping is further improved.

Description

Technical Field

[0001] The present application relates to the field of image processing technology, and in particular to building surveying and mapping methods, devices, equipment, storage media and products. Background Art

[0002] In the field of photogrammetry, building mapping is of great significance in many applications such as urban planning and disaster management. Traditional image-based building mapping methods mostly rely on digital orthophoto maps (DOM) or true digital orthophoto maps (TDOM), but these signals are usually produced by multi-view images through photogrammetry. Relying only on single-view information for building mapping will ignore the valuable information obtained from redundant observations.

[0003] At present, building mapping suffers from the problem of inaccurate projection of image space to ground feature space due to the quality limitation of the model used (such as noise). Therefore, how to improve the accuracy of building mapping is a problem that needs to be solved urgently. Summary of the invention

[0004] The main purpose of this application is to provide a building surveying and mapping method, device, equipment, storage medium and product, aiming to solve the technical problem of low accuracy in building surveying and mapping.

[0005] To achieve the above objectives, the present application proposes a building surveying and mapping method, the method comprising:

[0006] Obtaining a multi-view original observation image based on the aerial image, wherein the multi-view original observation image is used to provide building observation information at different angles;

[0007] Inputting the multi-view original observation image into a target building segmentation model, performing feature extraction and semantic segmentation on the multi-view original observation image through the target building segmentation model, and obtaining a target building segmentation result;

[0008] The target building contour information is extracted based on the target building segmentation result, and the target building is surveyed and mapped according to the target building contour information.

[0009] In one embodiment, the target building segmentation model includes an image space and a ground object space, and the step of inputting the multi-view original observation image into the target building segmentation model, performing feature extraction and semantic segmentation on the multi-view original observation image through the target building segmentation model, and obtaining a target building segmentation result includes:

[0010] In the image space, the multi-view original observation image is input into the target building segmentation model, and an initial building segmentation result is generated through a feature encoder and a full convolution head structure;

[0011] In the object space, based on the initial contour information and the initial building segmentation result, the space features of the object are acquired through piecewise affine projection and feature fusion;

[0012] Based on the initial contour information and the initial building segmentation result, the spatial features of the ground object are optimized through a segmentation head structure to obtain a target building segmentation result.

[0013] In one embodiment, the step of inputting the multi-view original observation image into the target building segmentation model and generating an initial building segmentation result through a feature encoder and a full convolution head includes:

[0014] In the image space, splicing the multi-view original observation image with the corresponding digital surface model data to obtain splicing features;

[0015] Inputting the concatenated features into a feature encoder to generate a multi-view semantic feature map;

[0016] Based on the multi-view semantic feature map, an initial building segmentation result is generated through a full convolution head structure, and initial contour information is extracted according to the initial building segmentation result.

[0017] In one embodiment, the step of acquiring spatial features of ground objects through piecewise affine projection and feature fusion based on the initial contour information and the initial building segmentation result comprises:

[0018] Projecting the initial building segmentation result and the corresponding initial contour information into the ground object space through piecewise affine projection to obtain the projected initial building segmentation result and the orthophoto of the ground object space;

[0019] In the feature space, the projected initial building segmentation result, the orthophoto of the feature space and the digital surface model data are fused to obtain the feature space characteristics.

[0020] In one embodiment, the step of optimizing the spatial features of the ground object through a segmentation head structure based on the initial contour information and the initial building segmentation result to obtain the target building segmentation result includes:

[0021] The projected initial building segmentation result, the orthophoto of the feature space and the digital surface model data are fused to obtain a fused feature map;

[0022] Performing element-level operations on the fused feature map to obtain enhanced fused features;

[0023] The enhanced fusion features are input into the segmentation head structure to obtain the target building segmentation result.

[0024] In one embodiment, before the step of inputting the multi-view original observation image into the target building segmentation model, the step includes:

[0025] Input the multi-view original observation data set into the initial building segmentation model for model training, train the image space separately until convergence, and obtain the first intermediate building segmentation model;

[0026] Based on the first intermediate building segmentation model, integrating the ground feature space for end-to-end training to obtain a second intermediate building segmentation model;

[0027] The target building segmentation model is obtained by optimizing the second intermediate building segmentation model through a loss function, wherein the loss function is used to optimize the segmentation accuracy of the image space and the ground object space segmentation tasks.

[0028] In addition, to achieve the above-mentioned purpose, the present application also proposes a building surveying and mapping device, the building surveying and mapping device comprising:

[0029] An image acquisition module, used to obtain multi-view original observation images based on the aerial images, wherein the multi-view original observation images are used to provide building observation information at different angles;

[0030] A target segmentation module is used to input the multi-view original observation image into a target building segmentation model, and perform feature extraction and semantic segmentation on the multi-view original observation image through the target building segmentation model to obtain a target building segmentation result;

[0031] The target mapping module is used to extract the target building contour information based on the target building segmentation result, and perform target building mapping according to the target building contour information.

[0032] In addition, to achieve the above objectives, the present application also proposes a building surveying and mapping device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the building surveying and mapping method described above.

[0033] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the building surveying and mapping method described above are implemented.

[0034] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, the steps of the building surveying and mapping method described above are implemented.

[0035] One or more technical solutions proposed in this application have at least the following technical effects:

[0036] Based on the aerial images, the original multi-view observation images are obtained, and the original multi-view observation images are used to provide building observation information from different angles; the original multi-view observation images are input into the target building segmentation model, and the multi-view original observation images are subjected to feature extraction and semantic segmentation through the target mapping model to obtain the target building segmentation result; the target building contour information is extracted based on the building segmentation result, and the target building is mapped based on the target building contour information. By introducing the target building segmentation model to directly extract features and semantically segment the original multi-view observation images, the original observation data in the photogrammetry project can be more fully utilized, thereby improving the accuracy and mapping efficiency of building mapping. At the same time, the target building contour information is extracted based on the building segmentation result, which improves the accuracy of the edge contour in building mapping, and further improves the accuracy and robustness of building mapping. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0038] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0039] Figure 1 A schematic diagram of the flow chart of the first embodiment of the building surveying and mapping method of the present application;

[0040] Figure 2 A schematic diagram of the flow chart of the second embodiment of the building surveying and mapping method of the present application;

[0041] Figure 3 A schematic diagram of the flow chart of the third embodiment of the building surveying and mapping method of the present application;

[0042] Figure 4 This is a schematic diagram of the module structure of the building surveying and mapping device according to an embodiment of the present application;

[0043] Figure 5 Schematic diagram of the device structure of the hardware operating environment involved in the building surveying and mapping method in the embodiment of the present application.

[0044] The purpose, features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0045] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0046] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0047] In the field of photogrammetry, building mapping is of great significance in many applications such as urban planning and disaster management. Most traditional image-based building mapping methods rely on DOM or TDOM, but these signals are usually produced by multi-view images through photogrammetry. Relying only on single-view information for building mapping will ignore the valuable information obtained from redundant observations. Although deep learning technology has become mainstream in building mapping, there are still some problems with current learning methods. For example, when using multi-view information, many methods either do not directly input the original multi-view images into the prediction model, or combine multi-view data but do not achieve end-to-end learning, resulting in the inability to fully and effectively utilize multi-view information. In addition, when projecting multi-view image features into the object space, commonly used methods such as projection based on collinear equations will affect the accurate projection of image space semantic features due to errors in DSM (Digital Surface Model), thereby affecting the building mapping effect.

[0048] The present application provides a solution, which is to obtain multi-view original observation images based on aerial images, and the multi-view original observation images are used to provide building observation information from different angles; the multi-view original observation images are input into the target building segmentation model, and the multi-view original observation images are subjected to feature extraction and semantic segmentation through the target mapping model to obtain the target building segmentation result; the target building contour information is extracted based on the building segmentation result, and the target building is mapped based on the target building contour information. By introducing the target building segmentation model to directly perform feature extraction and semantic segmentation on the multi-view original observation images, the original observation data in the photogrammetry project can be more comprehensively utilized, thereby improving the accuracy and mapping efficiency of building mapping. At the same time, the target building contour information is extracted based on the building segmentation result, which improves the accuracy of the edge contour in building mapping, and further improves the accuracy and robustness of building mapping.

[0049] Based on this, the present application embodiment provides a building surveying method, referring to Figure 1 , Figure 1This is a flow chart of the first embodiment of the building surveying and mapping method of the present application.

[0050] In this embodiment, the building surveying method includes steps S10 to S30:

[0051] Step S10, obtaining a multi-view original observation image based on the aerial image.

[0052] It should be noted that the multi-view original observation images are used to provide building observation information from different angles. Aerial images can be ground images taken from aerial platforms (such as airplanes, drones, etc.), which are the data source for building surveying and mapping. Building observation information can be understood as visual information about the building obtained from multi-view images, such as the building's geometric shape, facade features, materials, etc.

[0053] Step S20, inputting the multi-view original observation images into the target building segmentation model, performing feature extraction and semantic segmentation on the multi-view original observation images through the target building segmentation model, and obtaining the target building segmentation result.

[0054] It should be noted that the target building segmentation model can include image space and object space. Image space can be used to process building features (such as contours and textures) in two-dimensional images, and object space can be combined with geographic information (such as latitude and longitude, terrain) to perform more accurate analysis and segmentation of buildings. Exemplarily, feature extraction can be used for further building recognition and segmentation, semantic segmentation can distinguish between building pixels and background pixels, and the target building segmentation result can be understood as the image result obtained after processing by the target building segmentation model.

[0055] Step S30, extracting target building contour information based on the target building segmentation result, and performing target building mapping according to the target building contour information.

[0056] Exemplarily, the target building contour information can be extracted from the pixel boundaries of the building in the segmentation result, which can be understood as the geometric information of the outer boundary of the building, including contour lines and shapes. Furthermore, the building can be accurately measured and drawn based on the target building contour information.

[0057] In this embodiment, by introducing the target building segmentation model to directly perform feature extraction and semantic segmentation on the multi-view original observation image, the original observation data in the photogrammetry project can be more comprehensively utilized, thereby improving the accuracy and efficiency of building mapping. At the same time, the target building contour information is extracted based on the building segmentation result, which improves the accuracy of the edge contour in building mapping and further improves the accuracy and robustness of building mapping.

[0058] Reference Figure 2, Figure 2 This is a flow chart of the second embodiment of the building surveying method of the present application. Figure 1 The first embodiment shown provides a second embodiment of the building surveying and mapping method of the present application.

[0059] In the second embodiment, the step S20 includes:

[0060] Step S201: In the image space, the multi-view original observation image is input into the target building segmentation model, and the initial building segmentation result is generated through the feature encoder and the full convolution head structure.

[0061] It should be noted that step S201 includes: in the image space, splicing the multi-view original observation image with the corresponding digital surface model data to obtain splicing features; inputting the splicing features into the feature encoder to generate a multi-view semantic feature map; based on the multi-view semantic feature map, generating an initial building segmentation result through a full convolution head structure, and extracting initial contour information based on the initial building segmentation result.

[0062] It should be noted that the Digital Surface Model (DSM) is a three-dimensional model used to represent the surface of terrain and the surface of objects. It can be generated by aerial images and other methods, and contains the height information of objects such as buildings. The splicing feature can be understood as combining the original observation image of multiple views with the digital surface model data in a specific dimension to form an input data containing more information. The feature encoder can be a neural network component for extracting the semantic features of the input data (such as edges, shapes, textures, etc.) and generating a high-dimensional semantic feature map. Correspondingly, the multi-view semantic feature map can be understood as the feature representation extracted by the encoder from the multi-view image, which integrates the visual information of the building from different angles. The full convolutional head structure can be a model based on a fully convolutional network (FCN, Fully Convolutional Network) for converting the feature map into a segmentation result. The initial building segmentation result can be understood as the result map of the preliminary segmentation, which shows the distinction between the building area and the background area in the image, and the corresponding building boundary information can be the initial contour information.

[0063] Step S202 , in the object space, based on the initial contour information and the initial building segmentation result, the spatial features of the object are obtained through segmented affine projection and feature fusion.

[0064] It should be noted that step S202 includes: projecting the initial building segmentation result and the corresponding initial contour information to the terrain space through segmented affine projection, obtaining the projected initial building segmentation result and the orthophoto of the terrain space; in the terrain space, fusing the projected initial building segmentation result, the orthophoto of the terrain space and the digital surface model data to obtain the terrain space characteristics.

[0065] Exemplarily, the piecewise affine projection can map the initial segmentation results and contour information in the two-dimensional image to the object space and correct it according to the terrain and DSM data. The orthophoto of the object space can eliminate the tilt or distortion and accurately reflect the position of the object in the real geographic space. The spatial features of the object can be represented as fused feature data, which combines the information of the building segmentation results, orthophotos and DSM data.

[0066] Step S203, based on the initial contour information and the initial building segmentation result, the spatial characteristics of the ground object are optimized through the segmentation head structure to obtain the target building segmentation result.

[0067] It should be noted that step S203 includes: fusing the projected initial building segmentation result, the orthophoto of the feature space, and the digital surface model data to obtain a fused feature map; performing element-level operations on the fused feature map to obtain enhanced fused features; and inputting the enhanced fused features into the segmentation head structure to obtain the target building segmentation result.

[0068] Exemplarily, semantic segmentation modules can be set in image space and object space respectively, and firstly processed in image space, each multi-view image is cropped and its corresponding projected DSM is spliced ​​and input into a feature encoder (CNN or self-attention structure can be used) to generate a multi-view semantic feature map, and then the building segmentation result is generated and the contour is extracted through a full convolution head structure. In object space, the TDOM image is cropped and spliced ​​with the DSM data and then input into a network structure aligned with the image space to generate an object space semantic feature map, and at the same time, the multi-view feature map of the image space is projected into the object space using a segmented mapping table, and the features are fused through element averaging and addition operations, and finally input into the segmentation head to obtain the final building mapping result.

[0069] In this embodiment, by combining the multi-view original observation image and the digital surface model, the two-dimensional image and the three-dimensional height information are organically integrated, making the representation of building features more comprehensive and accurate. By using segmented affine projection, high-precision correction from image space to ground object space is achieved. In particular, for complex terrain or high-rise buildings, terrain correction is performed by combining DSM data, effectively reducing errors such as tilt and distortion. The integration of multi-view features and the ground object space correction by segmented affine projection significantly improve the robustness of the method, which can adapt to various terrains, complex building forms or diversified data sources.

[0070] Reference Figure 3 , Figure 3 This is a flow chart of the third embodiment of the building surveying method of the present application. Figure 2 The second embodiment shown provides a third embodiment of the building surveying and mapping method of the present application.

[0071] In the third embodiment, before step S20, the following steps are included:

[0072] Step S11, inputting the multi-view original observation data set into the initial building segmentation model for model training, training the image space alone until convergence, and obtaining a first intermediate building segmentation model.

[0073] It should be noted that the first intermediate building segmentation model can be understood as a preliminary model obtained after training using only image space data. The multi-view original observation dataset is obtained based on the multi-view original observation images. Exemplarily, an aerial photogrammetry dataset containing original observation images (resolution of about 0.3m, 17310×11310 pixels, taken by a UCX camera at an average altitude of 3500m, with an overlap percentage of 65% and 30% within and between flight strips, respectively) and ground feature space data products (DSM, TDOM, ground sampling distance 0.3m) is constructed. ContextCapture software is used for aerial triangulation and dense matching, and TerraScan software is used to extract ground points to generate DEM. The dataset is divided into 30 sub-areas, 24 of which are used for training and verification, and 6 are used for testing. 3D polygons are obtained as real data by skilled personnel annotating the building outlines. It should be noted that when generating labels, the label map can be generated by rasterizing 3D building polygons in the object space, and the labels can be generated in the image space by projecting the 3D polygons onto the multi-view original image, and the occlusion-aware labeling method can be used to reduce the projection error.

[0074] Step S12: Based on the first intermediate building segmentation model, the ground feature space is integrated to perform end-to-end training to obtain a second intermediate building segmentation model.

[0075] It should be noted that the second intermediate building segmentation model can be understood as a segmentation model obtained through end-to-end training after integrating the spatial information of the ground object. Exemplarily, based on steps S11 to S12, a method of pre-calculating the segmentation transformation parameters and constructing a mapping table can be adopted to avoid repeated calculation of projection parameters during training, and a phased training strategy is implemented, first training the image space stage until convergence, and then integrating the ground object space workflow for end-to-end training.

[0076] Step S13, optimizing the second intermediate building segmentation model through the loss function to obtain the target building segmentation model.

[0077] It should be noted that the loss function is used to optimize the segmentation accuracy of image space and object space segmentation tasks. Exemplarily, the loss function uses a combination of dice loss and focal loss, and is applied to image and object space segmentation tasks at the same time. The overall performance of building segmentation can be evaluated using indicators such as intersection over union (IoU), F1 score (F1-score), precision (Precision) and recall (Recal), and the accuracy of building contours can be evaluated using contour-based indicators (calculating IoU, F1-score, Precision and Recal by performing mathematical morphological operations on generated contours and real polygons and taking the average value).

[0078] In this embodiment, the model training efficiency is improved by pre-calculating the segmented mapping table and the phased training strategy. The end-to-end learning framework enables the multi-view information to be effectively integrated in the entire surveying and mapping process, avoiding information loss in the intermediate links. The mapping performance is successfully improved through adaptive reordering and efficient fusion of multi-source features.

[0079] It should be noted that the above examples are only used to understand the present application and do not constitute a limitation on the building surveying and mapping method of the present application. More simple transformations based on this technical concept are all within the scope of protection of the present application.

[0080] This application also provides a building surveying and mapping device, please refer to Figure 4 , the building surveying and mapping device comprises:

[0081] An image acquisition module 10 is used to obtain a multi-view original observation image based on the aerial image, wherein the multi-view original observation image is used to provide building observation information at different angles;

[0082] The target segmentation module 20 is used to input the multi-view original observation image into the target building segmentation model, and perform feature extraction and semantic segmentation on the multi-view original observation image through the target building segmentation model to obtain a target building segmentation result;

[0083] The target mapping module 30 is used to extract the target building contour information based on the target building segmentation result, and perform target building mapping according to the target building contour information.

[0084] The building surveying device provided by the present application adopts the building surveying method in the above embodiment, which can solve the technical problem of low building surveying accuracy. Compared with the prior art, the beneficial effects of the building surveying device provided by the present application are the same as the beneficial effects of the building surveying method provided by the above embodiment, and other technical features in the building surveying device are the same as the features disclosed in the above embodiment method, which will not be repeated here.

[0085] The present application provides a building surveying and mapping device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the building surveying and mapping method in the above-mentioned embodiment 1.

[0086] Reference below Figure 5 , which shows a schematic diagram of the structure of a building surveying and mapping device suitable for implementing the embodiment of the present application. The building surveying and mapping device in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 5 The building surveying and mapping device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0087] like Figure 5As shown, the building surveying and mapping device may include a processing device 1001 (e.g., a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM: Random Access Memory) 1004. In RAM1004, various programs and data required for the operation of the building surveying and mapping device are also stored. The processing device 1001, ROM1002, and RAM1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the building surveying device to communicate with other devices wirelessly or by wire to exchange data. Figure 5 A building surveying device having various systems is shown, but it should be understood that it is not required to implement or have all of the systems shown. More or fewer systems may be implemented or have instead.

[0088] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0089] The building surveying and mapping device provided by the present application adopts the building surveying and mapping method in the above embodiment, which can solve the technical problem of low building surveying and mapping accuracy. Compared with the prior art, the beneficial effects of the building surveying and mapping device provided by the present application are the same as the beneficial effects of the building surveying and mapping method provided by the above embodiment, and the other technical features in the building surveying and mapping device are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.

[0090] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0091] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0092] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, and the computer-readable program instructions are used to execute the building surveying and mapping method in the above-mentioned embodiment.

[0093] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination of the above.

[0094] The computer-readable storage medium may be included in the building surveying and mapping device; or may exist independently without being assembled into the building surveying and mapping device.

[0095] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by the building surveying and mapping device, the building surveying and mapping device: obtains a multi-view original observation image based on the aerial image, and the multi-view original observation image is used to provide building observation information at different angles; inputs the multi-view original observation image into a target building segmentation model, and performs feature extraction and semantic segmentation on the multi-view original observation image through the target building segmentation model to obtain a target building segmentation result; extracts the target building contour information based on the target building segmentation result, and performs target building surveying and mapping according to the target building contour information.

[0096] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0097] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0098] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0099] The readable storage medium provided in the present application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned building surveying and mapping method, and can solve the technical problem of low building surveying and mapping accuracy. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in the present application are the same as the beneficial effects of the building surveying and mapping method provided in the above-mentioned embodiment, and will not be described in detail here.

[0100] The present application also provides a computer program product, including a computer program, which implements the steps of the above-mentioned building surveying and mapping method when executed by a processor.

[0101] The computer program product provided in this application can solve the technical problem of low accuracy in building surveying and mapping. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as the beneficial effects of the building surveying and mapping method provided in the above embodiment, which will not be repeated here.

[0102] The above descriptions are only some embodiments of the present application, and are not intended to limit the patent scope of the present application. All equivalent structural changes made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect applications in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A building surveying method, characterized in that: The method includes: Obtaining a multi-view original observation image based on the aerial image, wherein the multi-view original observation image is used to provide building observation information at different angles; Inputting the multi-view original observation image into a target building segmentation model, performing feature extraction and semantic segmentation on the multi-view original observation image through the target building segmentation model, and obtaining a target building segmentation result; The target building contour information is extracted based on the target building segmentation result, and the target building is surveyed and mapped according to the target building contour information.

2. The method according to claim 1, characterized in that The target building segmentation model includes an image space and a ground object space. The step of inputting the multi-view original observation image into the target building segmentation model, performing feature extraction and semantic segmentation on the multi-view original observation image through the target building segmentation model, and obtaining a target building segmentation result includes: In the image space, the multi-view original observation image is input into the target building segmentation model, and an initial building segmentation result is generated through a feature encoder and a full convolution head structure; In the object space, based on the initial contour information and the initial building segmentation result, the space features of the object are acquired through piecewise affine projection and feature fusion; Based on the initial contour information and the initial building segmentation result, the spatial features of the ground object are optimized through a segmentation head structure to obtain a target building segmentation result.

3. The method according to claim 2, characterized in that The step of inputting the multi-view original observation image into the target building segmentation model and generating an initial building segmentation result through a feature encoder and a full convolution head comprises: In the image space, splicing the multi-view original observation image with the corresponding digital surface model data to obtain splicing features; Inputting the concatenated features into a feature encoder to generate a multi-view semantic feature map; Based on the multi-view semantic feature map, an initial building segmentation result is generated through a full convolution head structure, and initial contour information is extracted according to the initial building segmentation result.

4. The method according to claim 3, characterized in that The step of acquiring the spatial features of the ground object by piecewise affine projection and feature fusion based on the initial contour information and the initial building segmentation result comprises: Projecting the initial building segmentation result and the corresponding initial contour information into the ground object space through piecewise affine projection to obtain the projected initial building segmentation result and the orthophoto of the ground object space; In the feature space, the projected initial building segmentation result, the orthophoto of the feature space and the digital surface model data are fused to obtain the feature space characteristics.

5. The method according to claim 4, characterized in that The step of optimizing the spatial features of the ground object through a segmentation head structure based on the initial contour information and the initial building segmentation result to obtain the target building segmentation result comprises: The projected initial building segmentation result, the orthophoto of the feature space and the digital surface model data are fused to obtain a fused feature map; Performing element-level operations on the fused feature map to obtain enhanced fused features; The enhanced fusion features are input into the segmentation head structure to obtain the target building segmentation result.

6. The method according to any one of claims 1 to 5, characterized in that: Before the step of inputting the multi-view original observation image into the target building segmentation model, the method includes: Input the multi-view original observation data set into the initial building segmentation model for model training, train the image space separately until convergence, and obtain the first intermediate building segmentation model; Based on the first intermediate building segmentation model, integrating the ground feature space for end-to-end training to obtain a second intermediate building segmentation model; The target building segmentation model is obtained by optimizing the second intermediate building segmentation model through a loss function, wherein the loss function is used to optimize the segmentation accuracy of the image space and the ground object space segmentation tasks.

7. A building surveying and mapping device, characterized in that: The device comprises: An image acquisition module, used to obtain multi-view original observation images based on the aerial images, wherein the multi-view original observation images are used to provide building observation information at different angles; A target segmentation module is used to input the multi-view original observation image into a target building segmentation model, and perform feature extraction and semantic segmentation on the multi-view original observation image through the target building segmentation model to obtain a target building segmentation result; The target mapping module is used to extract the target building contour information based on the target building segmentation result, and perform target building mapping according to the target building contour information.

8. A building surveying and mapping device, characterized in that: The device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the building surveying method according to any one of claims 1 to 6.

9. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the building surveying method according to any one of claims 1 to 6 are implemented.

10. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the steps of the building surveying method according to any one of claims 1 to 6 are implemented.

Images