An illegal building inspection method based on unmanned aerial vehicle remote sensing images

Abstract

The application discloses a kind of illegal construction inspection methods based on unmanned aerial vehicle remote sensing image, comprising: using unmanned aerial vehicle to carry out data acquisition to existing building, obtain building image and building coordinates;The building image is carried out edge extraction, and extraction image is obtained;Based on the extraction image and the building coordinates model is constructed;Get historical planning model, compare the historical planning model with the building coordinates model, obtain comparison result;Based on the comparison result, the building coordinates model is marked with illegal construction, and illegal construction inspection result is obtained.The application fuses unmanned aerial vehicle application and spatial data interaction technology, empowers city management, realizes city fine management, promotes the upgrading of traditional management mode.

Description

Technical Field

[0001] This invention belongs to the field of architectural planning, and in particular relates to a method for inspecting illegal buildings based on UAV remote sensing images. Background Technology

[0002] As cities continue to expand, urban management demands greater precision and intelligence. Traditional manual management and enforcement methods have revealed shortcomings in urban oversight, highlighting the growing need for cost reduction, efficiency improvement, and digital operations. Integrating drone applications with spatial data interaction technology to empower urban management, achieve refined urban management, and promote the upgrading of traditional management methods has become a crucial aspect of current drone-based urban management applications.

[0003] Currently, due to the continuous expansion of urban areas, illegal construction and the addition of floors to buildings occur frequently. However, the addition of floors to buildings cannot be effectively identified by manual observation. At the same time, the current methods of detecting illegal constructions are mainly based on manual patrols, which are time-consuming, labor-intensive, and inefficient. Summary of the Invention

[0004] The purpose of this invention is to provide a method for inspecting illegal buildings based on UAV remote sensing images, so as to solve the problems existing in the prior art.

[0005] To achieve the above objectives, the present invention provides a method for inspecting illegal buildings based on UAV remote sensing images, comprising:

[0006] Drones were used to collect data on existing buildings, obtaining building images and coordinates;

[0007] The building image is then subjected to edge extraction to obtain the extracted image;

[0008] A building coordinate model is constructed based on the extracted image and the building coordinates;

[0009] Obtain the historical planning model, compare the historical planning model with the building coordinate model, and obtain the comparison results;

[0010] Based on the comparison results, illegal buildings are marked on the building coordinate model to obtain the inspection results of illegal buildings.

[0011] Preferably, the process of obtaining the building image and building coordinates includes:

[0012] Based on street planning, drone patrol routes are set;

[0013] The drone collects building images based on the patrol route;

[0014] The origin of the coordinate system is set based on the patrol route of the UAV. During the patrol, the UAV sets the coordinates of the street buildings to obtain the building coordinates.

[0015] Preferably, the process of acquiring and extracting the image includes:

[0016] Edge detection is performed on the building image based on the direction and magnitude of the image to obtain the edge location image;

[0017] The edge location image is divided into edge regions and noise regions;

[0018] A spatial domain method is set up, and the noise region is denoised based on the spatial domain method to obtain a denoised region;

[0019] The edge region and the denoised region are integrated to obtain the extracted image.

[0020] Preferably, the process of obtaining the denoised region includes:

[0021] Select a filter, and perform a convolution operation on the noise region based on the filter to obtain a first denoised region;

[0022] The first denoised region is subjected to bilateral filtering to obtain the second denoised region;

[0023] The second denoised region is obtained by performing pixel continuity detection and outputting the result.

[0024] Preferably, the process of constructing the building coordinate model includes:

[0025] The extracted image is acquired, and a three-dimensional pose detection is performed on the extracted image to obtain the three-dimensional pose of the image;

[0026] Based on the building coordinates, the image's three-dimensional pose is analyzed for real-space height, and coordinate processing is performed to generate the building's three-dimensional coordinates.

[0027] By setting an allowable error, the three-dimensional coordinates of the building are calculated to obtain the building coordinate model.

[0028] Preferably, the process of obtaining the comparison results includes:

[0029] The historical planning model is obtained by establishing a model based on historical planning information.

[0030] The historical planning model and the architectural coordinate model are compared to obtain the comparison results.

[0031] Preferably, the process of obtaining the historical planning model includes:

[0032] Obtain historical planning information, calculate the historical planning information, and obtain a first planning model;

[0033] The first planning model is divided based on the origin of the coordinate system to obtain the second planning model;

[0034] The second planning model is calibrated to generate the historical planning model.

[0035] Preferably, the process of obtaining the inspection results of illegal buildings includes:

[0036] The similarity results are calculated by performing a similarity calculation on the comparison results to obtain the similarity results;

[0037] A similarity threshold is set, and illegal buildings are marked on the building coordinate model based on the similarity results to obtain the inspection results of illegal buildings. Among them, non-illegal buildings are marked in green, and illegal buildings are marked in red.

[0038] The technical effects of this invention are as follows:

[0039] 1. This invention uses the construction of a three-dimensional city model for the inspection of illegal buildings, which can promptly and effectively detect and deal with the construction of illegal buildings;

[0040] 2. This invention improves the efficiency of inspecting illegal buildings by using drones for patrols;

[0041] 3. This invention integrates drone applications with spatial data interaction technology to empower urban management, achieve refined urban management, and promote the upgrading of traditional management methods. Attached Figure Description

[0042] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0043] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0044] Figure 1 This is a schematic diagram of the illegal building inspection method in an embodiment of the present invention. Detailed Implementation

[0045] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0046] Example 1

[0047] like Figure 1 As shown, this embodiment provides a method for inspecting illegal buildings based on UAV remote sensing images, including:

[0048] Drones were used to collect data on existing buildings, obtaining building images and coordinates;

[0049] The building image is then subjected to edge extraction to obtain the extracted image;

[0050] A building coordinate model is constructed based on the extracted image and the building coordinates;

[0051] Obtain the historical planning model, compare the historical planning model with the building coordinate model, and obtain the comparison results;

[0052] Based on the comparison results, illegal buildings are marked on the building coordinate model to obtain the inspection results of illegal buildings.

[0053] Further optimization of the scheme includes the following process for obtaining building images and coordinates:

[0054] The process involves acquiring 3D point cloud data of the target street, calibrating feature point coordinates and initial camera coordinates from the 3D point cloud data, determining the horizontal projection direction vector between the initial camera coordinates and feature point coordinates in the horizontal plane based on the feature point coordinates and the initial camera coordinates, determining the drone's nose direction based on the horizontal projection direction vector, determining the horizontal unit vector of the projection direction based on the nose direction, correcting the projection direction vector by inferring the horizontal unit vector of the projection direction from the nose direction, and calculating the coordinates of feasible photo points based on the 3D unit vector calculated from the horizontal projection direction vector, the feature point coordinates, and the preset spatial distance.

[0055] The patrol routes for drones, based on street planning, mainly include:

[0056] (1) UAV waypoint planning: set once and call repeatedly.

[0057] (2) Each waypoint can be set with individual commands, and commands such as taking pictures, recording videos, hovering, and adjusting the drone's orientation can be set automatically.

[0058] (3) Additional waypoints can be set up for key areas such as entrances and exits, parking lots, fire lanes and other important facilities to conduct key inspections, record photos or video data and archive them.

[0059] The drone collects building images based on the patrol route;

[0060] The origin of the coordinate system is set based on the patrol route of the UAV. During the patrol, the UAV sets the coordinates of the street buildings to obtain the building coordinates.

[0061] Further optimization of the scheme, the process of acquiring and extracting the image includes:

[0062] Edge detection is performed on the building image based on the direction and magnitude of the image to obtain the edge location image;

[0063] An image includes the target, background, and noise. The most common way to extract the target object is to use an automated method to extract a threshold T. The image data is then divided into two parts by T, which is called binarization. Binarization separates the image into the target object and the background.

[0064] The edge location image is divided into edge regions and noise regions;

[0065] A spatial domain method is set up, and the noise region is denoised based on the spatial domain method to obtain a denoised region;

[0066] The edge region and the denoised region are integrated to obtain the extracted image.

[0067] Further optimization of the scheme, the process of obtaining the denoised region includes:

[0068] Select a filter, and perform a convolution operation on the noise region based on the filter to obtain a first denoised region;

[0069] The high-frequency components of an image mainly consist of its edges and details. Therefore, high-pass filtering can enhance these high-frequency signals, making blurry images clearer. Common high-pass filters include ideal high-pass filters, Butterworth high-pass filters, exponential high-pass filters, and trapezoidal high-pass filters. This application uses a Butterworth high-pass filter.

[0070] The values ​​in a Gaussian kernel follow a Gaussian distribution, with the largest value at the center and other values ​​decreasing in magnitude with increasing distance from the center. Convolving an image with a Gaussian kernel makes the image more blurred (smooth), and the degree of blurring is determined by the standard deviation of the Gaussian kernel; the larger the standard deviation, the greater the smoothness. Gaussian filtering can effectively remove Gaussian noise from images.

[0071] The specific algorithm is as follows:

[0072]

[0073]

[0074] The Gaussian function's range is between (0,1), meaning the sum of the nine values ​​in the Gaussian kernel should equal 1. This 1 is then distributed across the nine pixels, giving the middle pixels a higher percentage. In essence, Gaussian filtering applies a weighted average to the grayscale values ​​of the pixels covered by the Gaussian kernel, with the middle pixels receiving a higher weight and the surrounding pixels receiving a lower weight.

[0075] The first denoised region is subjected to bilateral filtering to obtain the second denoised region;

[0076] The specific algorithm for obtaining the second denoised region is as follows:

[0077]

[0078]

[0079] The second denoised region is obtained by performing pixel continuity detection and outputting the result.

[0080] Further optimization of the scheme includes the following process for constructing the building coordinate model:

[0081] The extracted image is acquired, and a three-dimensional pose detection is performed on the extracted image to obtain the three-dimensional pose of the image;

[0082] Based on the building coordinates, the image's three-dimensional pose is analyzed for real-space height, and coordinate processing is performed to generate the building's three-dimensional coordinates.

[0083] In this embodiment, two unit vectors i and j, oriented in the same direction as the x-axis and y-axis respectively, are taken as a basis. a is any vector in the Cartesian coordinate system, and vector OP = a is constructed with the origin O as the starting point. According to the fundamental theorem of planar vectors, there exists one and only one pair of real numbers (x, y) such that a = vector OP = x. i +y j Therefore, the real number pair (x, y) is called the coordinates of vector a, denoted as a = (x, y).

[0084] By setting an allowable error, the three-dimensional coordinates of the building are calculated to obtain the building coordinate model.

[0085] To further optimize the scheme, the process of obtaining the comparison results includes:

[0086] The historical planning model is obtained by establishing a model based on historical planning information.

[0087] The historical planning model and the architectural coordinate model are compared to obtain the comparison results.

[0088] The process of further optimizing the scheme and obtaining the historical planning model includes:

[0089] Obtain historical planning information, calculate the historical planning information, and obtain a first planning model;

[0090] The first planning model is divided based on the origin of the coordinate system to obtain the second planning model;

[0091] The second planning model is calibrated to generate the historical planning model.

[0092] This embodiment determines the first position coordinates of the target entity object in the pixel coordinate system from the 3D pose detection results of the target entity object in the image frame to be processed; when the target entity object is detected to be outside the spatial reference plane, the first spatial plane image belonging to the spatial reference plane and the second spatial plane image belonging to the target spatial plane are obtained from the image frame to be processed; the first spatial plane image is used as the spatial plane texture of the lower bounding spatial plane of the building image, the second spatial plane image is used as the spatial plane texture of the upper bounding spatial plane of the building image, and the real spatial height between the target spatial plane and the spatial reference plane is used as the spatial reference height between the upper bounding spatial plane and the lower bounding spatial plane to construct a 3D spatial model of the building image mapped in virtual space; according to the first position coordinates of the target entity object in the pixel coordinate system, the upper intersection point coordinates and the lower intersection point coordinates are obtained from the 3D spatial model respectively; coordinate transformation processing is performed on the upper intersection point coordinates and the lower intersection point coordinates between the pixel coordinate system and the virtual space world coordinate system respectively.

[0093] Further optimization of the scheme, the process of obtaining the inspection results of illegal buildings includes:

[0094] The similarity results are calculated by performing a similarity calculation on the comparison results to obtain the similarity results;

[0095] A similarity threshold is set, and illegal buildings are marked on the building coordinate model based on the similarity results to obtain the inspection results of illegal buildings. Among them, non-illegal buildings are marked in green, and illegal buildings are marked in red.

[0096] The above description is merely a preferred embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1.A method for illegal construction inspection based on unmanned aerial vehicle remote sensing images, characterized in that, The method comprises the following steps: Data collection on existing buildings using a drone to obtain building images and building coordinates, including: setting a patrol route for the drone based on street planning; Building image collection by the drone based on the patrol route to obtain the building images; Setting a coordinate origin based on the patrol route of the drone, and setting coordinates for street buildings during the patrol process to obtain the building coordinates; Edge extraction of the building images to obtain an extracted image, including: edge detection of the building images based on the direction and amplitude of the images to obtain an edge position image; Dividing the edge position image to obtain an edge region and a noise region; Setting a spatial domain method, and denoising the noise region based on the spatial domain method to obtain a denoised region; Integrating the edge region and the denoised region to obtain the extracted image; The process of obtaining the denoised region comprises: Selecting a filter, and performing convolution operation on the noise region based on the filter to obtain a first denoised region; Bilateral filtering the first denoised region to obtain a second denoised region; Performing pixel continuity detection on the second denoised region and then outputting to obtain the denoised region; Building a building coordinate model based on the extracted image and the building coordinates, including: obtaining the extracted image, performing three-dimensional pose detection on the extracted image to obtain an image three-dimensional pose; Performing real space height analysis on the image three-dimensional pose based on the building coordinates, and simultaneously performing coordinate processing to generate a building three-dimensional coordinate; Setting an allowable error, and performing coordinate calculation on the building three-dimensional coordinate to obtain the building coordinate model; Obtaining a historical planning model, comparing the historical planning model with the building coordinate model to obtain a comparison result, including: establishing a model based on historical planning information to obtain the historical planning model; The process of obtaining the historical planning model comprises: Obtaining historical planning information, and performing calculation on the historical planning information to obtain a first planning model; Dividing the first planning model based on a coordinate origin to obtain a second planning model; Performing model correction on the second planning model to generate the historical planning model; Model comparison between the historical planning model and the building coordinate model to obtain the comparison result; Based on the comparison result, marking illegal buildings on the building coordinate model to obtain an illegal building inspection result, including: similarity calculation on the comparison result to obtain a similarity result; Setting a similarity threshold, marking illegal buildings on the building coordinate model based on the similarity result to obtain the illegal building inspection result, wherein non-illegal buildings are marked in green and illegal buildings are marked in red.

Images