CN114677600A - Illegal construction detection method, illegal construction detection system, computer equipment and storage medium - Google Patents
Illegal construction detection method, illegal construction detection system, computer equipment and storage medium Download PDFInfo
- Publication number
- CN114677600A CN114677600A CN202210168689.3A CN202210168689A CN114677600A CN 114677600 A CN114677600 A CN 114677600A CN 202210168689 A CN202210168689 A CN 202210168689A CN 114677600 A CN114677600 A CN 114677600A
- Authority
- CN
- China
- Prior art keywords
- target
- orthophoto map
- image
- building
- orthophoto
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Processing Or Creating Images (AREA)
- Image Processing (AREA)
Abstract
The invention discloses an illegal construction detection method, which comprises the following steps: acquiring an aerial orthographic image set of a target area; carrying out normalization processing on the target orthophoto map; identifying a top image of a target building in the target orthophoto map, and calculating coordinates of the target building; extracting a reference area corresponding to the target building according to the coordinates of the target building in the reference aerial orthographic image set; and comparing the top image of the target building with the corresponding reference area to judge whether the target building is illegal. The invention also discloses a system for detecting the illegal construction, computer equipment and a computer readable storage medium. The invention can effectively realize the illegal construction detection, and has small operand and high detection speed.
Description
Technical Field
The present invention relates to the field of image recognition technologies, and in particular, to an illegal construction detection method, an illegal construction detection system, a computer device, and a computer-readable storage medium.
Background
With the rapid development of urban infrastructure, residential houses, ecological environment and other construction, the continuous enhancement of urban comprehensive service capability and competitiveness becomes an important and urgent foundation for building a modern city. When the city is developed at a high speed and expands day by day, a large number of illegal buildings are derived, the planning order of the city is disturbed, the continuous development of the city is prevented, and therefore the illegal building behaviors of the city are powerfully supervised.
At present, the commonly used illegal construction detection method is mainly carried out by manpower, but illegal construction inspection of a city can be completed only by manual inspection for months, time and labor are wasted, and illegal construction and follow-up are difficult to find in time.
In addition, artificial intelligence technology is introduced into some illegal building detection. The method comprises the steps of shooting a region to be detected at a plurality of angles, and reconstructing point cloud data to construct an accurate three-dimensional model, so that illegal construction detection is realized. However, the three-dimensional modeling has a large amount of calculation, needs support of a large amount of point cloud data, is high in cost, is a small-area modeling, and can be realized only by operating a general machine for one or two days, so that the actual requirements of users cannot be met, and the three-dimensional modeling is not beneficial to practical application.
Disclosure of Invention
The technical problem to be solved by the present invention is to provide an illegal building detection method, an illegal building detection system, a computer device and a computer readable storage medium, which can effectively implement illegal building detection, and have small computation amount and high detection speed.
In order to solve the above technical problem, the present invention provides an illegal building detection method, which comprises: acquiring an aerial photography orthographic image set of a target area, wherein the aerial photography orthographic image set comprises a reference aerial photography orthographic image set and a target aerial photography orthographic image set, the reference aerial photography orthographic image set comprises at least one reference orthographic image, and the target aerial photography orthographic image set comprises at least one target orthographic image; normalizing the target orthographic projection image; identifying a top image of a target building in the target orthophoto map, and calculating coordinates of the target building; extracting a reference area corresponding to the target building according to the coordinates of the target building in the reference aerial orthographic image set; and comparing the top image of the target building with the corresponding reference area to judge whether the target building is illegal.
As an improvement of the above, the step of normalizing the target orthophoto map includes: carrying out scaling processing on the target ortho-image map so as to enable actual distances represented by pixel points in the target ortho-image map and the reference ortho-image map on the same ground surface to be consistent; and performing rotation processing on the target orthophoto map so as to enable the direction of the same ground surface in the target orthophoto map and the reference orthophoto map to be consistent.
As an improvement of the above solution, the step of scaling the target orthophoto map includes: calculating the target size of the target orthophoto map according to the picture size of the reference orthophoto map, the shooting height of the reference orthophoto map and the shooting height of the target orthophoto map; and scaling the target orthophoto map according to the target size.
As an improvement of the above, the step of performing rotation processing on the target orthophoto map includes: acquiring a yaw angle of acquisition equipment corresponding to the target orthophoto map; and rotating the target orthophoto map according to the yaw angle.
As an improvement of the above solution, the step of identifying the top image of the target building in the target orthophoto map and calculating the coordinates of the target building includes: acquiring the central coordinate of the target orthophoto map; identifying a top image of a target building in the target orthophoto map through deep learning; acquiring the actual distance represented by the pixel points in the normalized target orthophoto map; extracting the relative position of the target building and the central coordinate in the normalized target orthophoto map; and calculating the coordinates of the target building according to the actual distance, the relative position and the center coordinates.
As an improvement of the above, the step of extracting the reference area corresponding to the target building from the coordinates of the target building includes: acquiring the central coordinate of the reference orthophoto map; extracting a reference orthophoto map closest to the coordinates of the target building according to the central coordinates of the reference orthophoto map; locating a reference point in accordance with the coordinates of the target building in the extracted reference orthophoto map; and cutting out a reference area in a preset range by taking the reference point as a center in the extracted reference orthographic projection image.
As an improvement of the above aspect, the method of acquiring the center coordinates of the target orthophoto map or the reference orthophoto map includes: and generating the central coordinate of the target orthophoto map or the reference orthophoto map according to the longitude and latitude of the acquisition equipment corresponding to the target orthophoto map or the reference orthophoto map.
As an improvement of the above solution, the step of comparing the top image of the target building with the corresponding reference area to determine whether the target building is illegal comprises: matching the top image of the target building with the corresponding reference area in a characteristic point matching mode; judging whether a reference building top image similar to the top image of the target building exists in the reference area or not; if so, indicating that the target building has no illegal construction; and if not, indicating that the target building is illegal.
Correspondingly, the invention also provides an illegal construction detection system, which comprises: the device comprises an acquisition module, a display module and a display module, wherein the acquisition module is used for acquiring an aerial orthographic image set of a target area, the aerial orthographic image set comprises a reference aerial orthographic image set and a target aerial orthographic image set, the reference aerial orthographic image set comprises at least one reference orthographic image, and the target aerial orthographic image set comprises at least one target orthographic image; the normalizing module is used for carrying out normalization processing on the target orthographic projection image; the identification module is used for identifying the top image of a target building in the target orthophoto map and calculating the coordinates of the target building; the reference module is used for extracting a reference area corresponding to the target building according to the coordinates of the target building in the reference aerial orthographic image set; and the comparison module is used for comparing the top image of the target building with the corresponding reference area so as to judge whether the target building is illegal.
Correspondingly, the invention further provides a computer device, which comprises a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above violation detection method when executing the computer program.
Accordingly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when being executed by a processor, performs the steps of the above-mentioned violation detection method.
The implementation of the invention has the following beneficial effects:
the top images of all the target buildings in the target orthophoto map are efficiently and accurately determined by identifying the top images of the target buildings, so that the effective positioning of the target buildings is realized by calculating the coordinates of the target buildings; meanwhile, the top image of the target building is accurately compared with the reference area through the targeted construction of the reference area, so that unnecessary operation consumption can be effectively reduced, the comparison speed and precision can be improved, and the cost is low.
Furthermore, after the target ortho-image is subjected to scaling and rotation processing, the reference ortho-image and the target ortho-image can adopt the same height and direction standards, so that the subsequent comparison and identification are facilitated; in addition, through the feature point algorithm, the feature points in the top image of the target building and the feature points in the reference region can be extracted for comparison, and the accuracy is high.
Drawings
FIG. 1 is a flow chart of an embodiment of a violation detection method of the present invention;
FIG. 2 is a schematic diagram of a fixed-point shooting lattice of the UAV of the present invention;
FIG. 3 is a flowchart of an embodiment of the present invention for normalizing a target orthophoto map;
FIG. 4 is a schematic view of the flight path of the drone of the present invention;
FIG. 5 is two orthographic views of a subject taken in succession in accordance with the present invention;
FIG. 6 is a schematic view of a rotation of a target orthophotomap of the present invention;
FIG. 7 is a flowchart of an embodiment of the present invention for identifying the top image of a target structure in a target orthophoto map and calculating coordinates of the target structure;
FIG. 8 is a top image of a target building identified in the present invention;
FIG. 9 is a schematic diagram of the center coordinates and coordinates of a target building in accordance with the present invention;
FIG. 10 is a flowchart of an embodiment of the present invention for extracting a reference area corresponding to a target building according to coordinates of the target building;
FIG. 11 is a schematic illustration of a reference region in the present invention;
FIG. 12 is a flowchart illustrating an embodiment of comparing a top image of a target building with a corresponding reference area to determine whether the target building is illegal;
FIG. 13 is a schematic illustration of the absence of buildings in violation of the present invention;
FIG. 14 is a schematic illustration of the violation of a target structure in accordance with the present invention;
FIG. 15 is a schematic diagram of the configuration of the violation detection system of the present invention.
Detailed Description
To make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the accompanying drawings.
Referring to fig. 1, fig. 1 shows a flowchart of an embodiment of a method for violation detection, which includes:
s101, acquiring an aerial orthographic image set of the target area.
The aerial photography orthographic image set comprises a reference aerial photography orthographic image set and a target aerial photography orthographic image set, wherein the reference aerial photography orthographic image set comprises at least one reference orthophoto map, and the target aerial photography orthographic image set comprises at least one target orthophoto map.
The reference ortho-image and the target ortho-image are both ortho-images, and are remote sensing images with an ortho-projection property, namely pictures for vertically looking at the ground in the air; the reference aerial orthographic image set is a set of all reference orthographic images for scanning type orthographic image shooting on all places in the target area for the first time; the target orthophoto image refers to an image obtained by scanning type orthophoto image shooting on a certain place at the Nth time (N is a positive integer greater than 1), and the target aerial orthophoto image set refers to a set of all reference orthophoto images obtained by scanning type orthophoto image shooting on all places in the target area at the Nth time; that is, the reference orthophoto map and the target orthophoto map are images of the same location at different times.
It should be noted that, the original remote sensing image has different degrees of distortion and distortion due to the influence of the internal state change (optical system distortion, scanning system nonlinearity, etc.), the external state (e.g. attitude change) and the surface condition (e.g. earth curvature, topography fluctuation) of the sensor during imaging, and for the geometric processing of the remote sensing image, not only the spatial information needs to be extracted (e.g. drawing contour lines), but also the gray scale of the image needs to be resampled according to the correct geometric relationship to form a new orthographic image. In the invention, the orthoscopic image is mainly manufactured by special equipment, such as a flat ground center projection type aerial photo, and an optical mechanical type corrector can be used.
According to the invention, the shooting reference aerial photography orthographic image set and the target aerial photography orthographic image set both adopt the same photographic element, the corresponding focal length, effective pixels, lens size and the like are the same, and the pitch angle is vertical to the ground during shooting. During shooting, can install photographic elements on unmanned aerial vehicle to carry out the multiple spot through unmanned aerial vehicle and take photo by plane, with the whole earth's surface image of covering the target area.
As shown in fig. 2, the unmanned aerial vehicle performs fixed-point shooting in a dot matrix manner, and each point in fig. 2 represents a position where the unmanned aerial vehicle shoots an ortho-image at high altitude in the location.
S102, normalization processing is carried out on the target orthophoto map.
Since the flying height and direction of the unmanned aerial vehicle may be different during the shooting process of the reference orthophoto map and the target orthophoto map, the target orthophoto map needs to be normalized by using the reference orthophoto map as a standard. Specifically, the normalization process may include, but is not limited to, coordinate centering, x-shear i ng normalization, scaling normalization, or rotation normalization.
S103, identifying the top image of the target building in the target orthophoto map, and calculating the coordinates of the target building.
The coordinate value of the target building is the longitude and latitude of a certain target building in the target orthophoto map.
And S104, extracting a reference area corresponding to the target building according to the coordinates of the target building in the reference aerial orthographic image set.
And S105, comparing the top image of the target building with the corresponding reference area to judge whether the target building is illegal.
Specifically, if the overhead image of the target building exists in the corresponding reference area, it indicates that there is no building violation in the target building, and if the overhead image of the target building does not exist in the corresponding reference area, it indicates that there is a building violation in the target building.
Therefore, the top image of the target building is identified, and the targeted reference area is constructed, so that the top image of the target building is accurately compared with the reference area, the precision is high, the cost is low, the calculation amount is greatly reduced, and the detection speed is improved.
Furthermore, because each target orthophoto map has a plurality of target buildings, in practical application, the target buildings can be sequentially/simultaneously extracted and compared with the reference area according to the preset rule, and the flexibility is strong.
Referring to fig. 3, fig. 3 shows specific steps of normalizing the target orthophoto map, including:
s201, scaling processing is carried out on the target orthophoto map.
Because the flying heights of the two aerial shots of the unmanned aerial vehicle are possibly different, the target orthophoto map needs to be scaled by taking the reference orthophoto map as a standard, so that the actual distances represented by the pixel points in the target orthophoto map and the reference orthophoto map on the same ground surface are consistent.
Specifically, the step of scaling the target orthophoto map includes:
(1) and calculating the target size of the target orthophoto map according to the picture size of the reference orthophoto map, the shooting height of the reference orthophoto map and the shooting height of the target orthophoto map.
Preferably, the picture size is a picture width.
In practical applications, the target size of the target orthophotograph can be calculated according to the formula Wi dthB ═ Wi dthA × He i lightb/He i lighta, where Wi dthA is the picture width of the reference orthophotograph, He i lighta is the shooting height of the reference orthophotograph (i.e., the flight height of the drone), Wi dthB is the scaled picture width of the target orthophotograph, and He i lightb is the shooting height of the target orthophotograph (i.e., the flight height of the drone).
(2) And scaling the target orthophoto map according to the target size.
For example, if the actual distance represented by one pixel point in the reference ortho-image is 20cm and the actual distance represented by one pixel point in the reference ortho-image is 30cm, the actual distance represented by one pixel point in the target ortho-image can be 20cm after the target ortho-image is amplified, so that the actual distances represented by the pixel points in the target ortho-image and the reference ortho-image on the same ground surface are both 20 cm.
S202, a rotation process is performed on the target orthophoto map.
As shown in fig. 4, the flight path of the drone is mostly snakelike to-and-fro flying, so that the direction of the photographed target orthophoto image changes along with the flight direction of the drone, which is not convenient for coordinate transformation and comparison and identification, and therefore, the target orthophoto image needs to be rotated by using the reference orthophoto image as a standard, so that the target orthophoto image and the reference orthophoto image have the same direction on the same ground surface.
Specifically, the step of performing rotation processing on the target orthophoto map includes:
(1) acquiring a yaw angle of acquisition equipment corresponding to a target orthophoto map;
it should be noted that the acquisition device in the present invention refers to an unmanned aerial vehicle equipped with a camera device.
(2) And rotating the target orthophoto map according to the yaw angle.
As shown in fig. 5, the buildings in the two orthographic images are rotated by about 90 ° due to the steering of the drone. Therefore, the target orthophoto map is rotated by acquiring the yaw angle of the unmanned aerial vehicle, so that all the target orthophoto maps can be rotated to a uniform angle, such as north, south and north.
As shown in FIG. 6, the rotated area A is the void left after the object orthophoto image is twisted, and the area B is the lost image after the object orthophoto image is twisted.
Therefore, after the target orthographic image is subjected to scaling and rotation processing, the reference orthographic image and the target orthographic image can adopt the same height and direction standards, and subsequent comparison and identification are facilitated.
Referring to fig. 7, fig. 7 shows the specific steps of identifying the top image of the target building in the orthophoto map and calculating the coordinates of the target building, including:
S301, acquiring the central coordinate of the target orthophoto map;
since the target orthophoto image is an orthophoto image, the longitude and latitude of the unmanned aerial vehicle during shooting can be regarded as the center coordinates of the target orthophoto image as the orthophoto image. In actual operation, the central coordinate of the target orthophoto map can be generated according to the longitude and latitude of the acquisition equipment corresponding to the target orthophoto map; that is, the center coordinates may be represented by latitude and longitude.
S302, identifying a top image of a target building in the target orthophoto map through deep learning;
as shown in fig. 8, the top image of the target building in the target orthophoto map can be automatically recognized by inputting the target orthophoto map into a deep learning model trained in advance (see the dashed box in fig. 8).
Preferably, top parts of target buildings with too small area or large repetition rate can be deleted properly to improve recognition efficiency and reduce repetition rate.
S303, acquiring the actual distance represented by the pixel points in the normalized target orthophoto map;
s304, extracting the relative position of the target building and the central coordinate in the normalized target orthophoto map;
In practical application, coordinate axes can be constructed according to the position and the center coordinates of the target building, so that the relative position can be decomposed into X-axis displacement and Y-axis displacement.
And S305, calculating the coordinates of the target building according to the actual distance, the relative position and the center coordinates.
As shown in fig. 9, M points are central coordinates of the target orthophoto map, N points are position points of the target building in the target orthophoto map, and coordinates of the target building can be calculated through geometric calculation according to an X-axis displacement (i.e., a width difference of pixel points) and a Y-axis displacement (a height difference of pixel points) between the target building and the central coordinates and an actual distance represented by the pixel points. Preferably, the coordinates may be represented by longitude and latitude.
For example, if the actual distance represented by the pixel point is 20cm, the width between M and N is 6 pixel points, and the height is 3 pixel points, it can be known that the actual width between M and N is 120cm (20cm × 6 is 120cm), and the actual height is 60cm (20cm × 3 is 60cm), so that the coordinates of the target building can be calculated from the center coordinates of M.
Therefore, the top images of all target buildings in the target orthophoto map can be efficiently and accurately identified through the deep learning technology, and effective positioning of the target buildings is achieved through calculating the coordinates of the target buildings.
Referring to fig. 10, fig. 10 shows a specific step of extracting a reference area corresponding to a target building according to coordinates of the target building, including:
s401, acquiring a central coordinate of a reference orthophoto map;
since the reference orthophoto map is an orthophoto image, the longitude and latitude of the unmanned aerial vehicle during shooting can be regarded as the center coordinates of the reference orthophoto map as an orthophoto image. In actual operation, the central coordinate of the reference ortho-image map can be generated according to the longitude and latitude of the acquisition equipment corresponding to the reference ortho-image map; that is, the center coordinates may be represented by latitude and longitude.
S402, extracting a reference orthophoto map closest to the coordinates of the target building according to the central coordinates of the reference orthophoto map;
it should be noted that the reference ortho-image maps may be prestored in the database, and the database also records the center coordinates of each reference ortho-image map, so that the reference ortho-image map closest to the coordinates of the target building can be quickly extracted by searching the database.
S403, positioning a reference point consistent with the coordinates of the target building in the extracted reference orthophoto map;
And S404, cutting out a reference area in a preset range by taking the reference point as a center in the extracted reference orthophoto map.
Since subsequent comparisons consume a large amount of computation power, the relevant region can be narrowed from the reference orthographic image to a smaller range (i.e., the reference region) in order to reduce unnecessary consumption and increase speed.
For example, a square reference area with a side length of 4cm is cut out with the reference point as the center of the square (see fig. 11); for another example, a circular reference region having a side length of 5cm is cut out with the reference point as the center of the circle.
Therefore, through the targeted construction of the reference region, unnecessary operation consumption can be effectively reduced, and the comparison speed can be improved.
Referring to fig. 12, fig. 12 shows specific steps of comparing the top image of the target building with the corresponding reference area to determine whether the target building is illegal, including:
s501, matching the top image of the target building with the corresponding reference area in a characteristic point matching mode;
s502, judging whether a reference building top image similar to the top image of the target building exists in the reference area;
s503, if yes, the target building is not established illegally;
S504, if the judgment result is no, the target building is illegal.
Through a feature point algorithm, feature points in the top image of the target building and feature points in a reference area can be extracted; then, matching the characteristic points in the top image with the characteristic points in the reference area; at this time, if a corresponding target building is matched, it is considered that the target building is also present in the reference orthographic projection image, and therefore the target building is not created (see fig. 13). If the corresponding target building cannot be matched, it is determined that the target building does not exist in the reference orthophoto map, and therefore the target building is illegal (see fig. 14).
Further, the matching proportion of the characteristic points can be manually set or adaptively adjusted; the self-adaptation means that if the building which is identified at this time is actually not built after manual judgment, the requirement of the matching proportion of the feature points can be reduced; if the building not established at this time is not identified, but the establishment point is found manually, the threshold value can be increased.
Referring to fig. 15, fig. 15 shows a specific structure of the violation detection system 100 of the present invention, which comprises:
the acquiring module 1 is used for acquiring an aerial orthographic image set of the target area. The aerial orthographic image set comprises a reference aerial orthographic image set and a target aerial orthographic image set, the reference aerial orthographic image set comprises at least one reference orthophoto map, and the target aerial orthographic image set comprises at least one target orthophoto map.
The normalizing module 2 is used for normalizing the target orthographic image; since the flying height and direction of the unmanned aerial vehicle may be different in the shooting process of the reference orthophoto map and the target orthophoto map, the target orthophoto map needs to be normalized by using the reference orthophoto map as a standard. Specifically, the normalization process may include, but is not limited to, coordinate centering, x-shear i ng normalization, scaling normalization, or rotation normalization.
The identification module 3 is used for identifying the top image of the target building in the target orthophoto map and calculating the coordinates of the target building; specifically, the coordinate value of the target building is the longitude and latitude of a certain target building in the target orthophoto map.
The reference module 4 is used for extracting a reference area corresponding to the target building according to the coordinates of the target building in the reference aerial orthographic image set;
and the comparison module 5 is used for comparing the top image of the target building with the corresponding reference area so as to judge whether the target building is illegal.
Therefore, the top image of the target building is identified, and the targeted reference area is constructed, so that the top image of the target building is accurately compared with the reference area, the precision is high, the cost is low, the calculation amount is greatly reduced, and the detection speed is improved.
The following respectively describes the normalization module 2, the identification module 3, the reference module 4 and the comparison module 5 in detail:
one, one module
The normalizing module 2 comprises a scaling unit and a rotating unit, and specifically:
and the scaling unit is used for scaling the target orthographic projection image. Because the flying heights of the unmanned aerial vehicle in the two aerial shots may be different, the target orthophoto map needs to be scaled by using the reference orthophoto map as a standard, so that the actual distances represented by the pixel points in the target orthophoto map and the reference orthophoto map on the same ground surface are consistent.
Specifically, the scaling unit calculates the target size of the target orthophoto map according to the picture size of the reference orthophoto map, the shooting height of the reference orthophoto map and the shooting height of the target orthophoto map, and scales the target orthophoto map according to the target size. Preferably, the picture size is a picture width.
In practical applications, the target size of the target orthophotomap may be calculated according to the formula Wi dthB — width ha × height b/height a, where width ha is the picture width of the reference orthophotomap, Hei _ right a is the shooting height of the reference orthophotomap (i.e., the flight height of the drone), width b is the scaled picture width of the target orthophotomap, and height b is the shooting height of the target orthophotomap (i.e., the flight height of the drone).
For example, if the actual distance represented by one pixel point in the reference ortho-image is 20cm and the actual distance represented by one pixel point in the reference ortho-image is 30cm, the actual distance represented by one pixel point in the target ortho-image can be 20cm after the target ortho-image is amplified, so that the actual distances represented by the pixel points in the target ortho-image and the reference ortho-image on the same ground surface are both 20 cm.
And the rotating unit is used for rotating the target orthographic projection image. As shown in fig. 4, the flight path of the drone is mostly snakelike to-and-fro flying, so that the direction of the photographed target orthophoto image changes along with the flight direction of the drone, which is not convenient for coordinate transformation and comparison and identification, and therefore, the target orthophoto image needs to be rotated by using the reference orthophoto image as a standard, so that the target orthophoto image and the reference orthophoto image have the same direction on the same ground surface.
Specifically, the rotating unit first obtains a yaw angle of the acquisition device corresponding to the target orthophoto map, and then rotates the target orthophoto map according to the yaw angle. It should be noted that the acquisition device in the present invention refers to an unmanned aerial vehicle equipped with a camera device.
As shown in fig. 5, the buildings in the two orthographic images are rotated by about 90 ° due to the steering of the drone. Therefore, the target orthophoto map is rotated by acquiring the yaw angle of the unmanned aerial vehicle, so that all the target orthophoto maps can be rotated to a uniform angle, such as north, south and north. As shown in FIG. 6, the rotated area A is the void left after the object orthophoto image is twisted, and the area B is the lost image after the object orthophoto image is twisted.
Therefore, after the target ortho-image is subjected to scaling and rotation processing, the reference ortho-image and the target ortho-image can adopt the same height and direction standards, and subsequent comparison and identification are facilitated.
Second, identification module
The identification module 3 comprises a target center acquisition unit, a top image identification unit, an actual distance acquisition unit, a relative position acquisition unit and a coordinate calculation unit, and specifically comprises:
the target center acquisition unit is used for acquiring the center coordinates of the target orthophoto map; because the target orthophoto map is an orthophoto, the longitude and latitude of the unmanned aerial vehicle during shooting can be regarded as the central coordinate of the target orthophoto map which is an orthophoto. In actual operation, the central coordinate of the target orthophoto map can be generated according to the longitude and latitude of the acquisition equipment corresponding to the target orthophoto map; that is, the center coordinates may be represented by latitude and longitude.
The top image identification unit is used for identifying the top image of the target building in the target orthophoto map through deep learning; as shown in fig. 8, the top image of the target building in the target orthophoto map can be automatically identified by inputting the target orthophoto map into a deep learning model trained in advance (see the dashed box in fig. 8). Preferably, top parts of target buildings with too small area or large repetition rate can be deleted properly to improve recognition efficiency and reduce repetition rate.
The actual distance acquisition unit is used for acquiring the actual distance represented by the pixel points in the normalized target orthophoto map;
the relative position acquisition unit is used for extracting the relative position of the target building and the central coordinate in the normalized target orthophoto map; in practical application, coordinate axes can be constructed according to the position and the center coordinates of the target building, so that the relative position can be decomposed into X-axis displacement and Y-axis displacement.
And the coordinate calculation unit is used for calculating the coordinates of the target building according to the actual distance, the relative position and the center coordinates. As shown in fig. 9, M points are central coordinates of the target orthophoto map, N points are position points of the target building in the target orthophoto map, and coordinates of the target building can be calculated through geometric calculation according to an X-axis displacement (i.e., a width difference of pixel points) and a Y-axis displacement (a height difference of pixel points) between the target building and the central coordinates and an actual distance represented by the pixel points. Preferably, the coordinates may be represented by longitude and latitude. For example, if the actual distance represented by the pixel point is 20cm, the width between M and N is 6 pixel points, and the height is 3 pixel points, it can be known that the actual width between M and N is 120cm (20cm × 6 is 120cm), and the actual height is 60cm (20cm × 3 is 60cm), so that the coordinates of the target building can be calculated from the center coordinates of M.
Therefore, the top images of all target buildings in the target orthophoto map can be efficiently and accurately identified through the deep learning technology, and effective positioning of the target buildings is achieved through calculating the coordinates of the target buildings.
Three, benchmark module
The reference module 4 comprises a reference center acquisition unit, a reference image extraction unit, a reference point positioning unit and a reference domain cutting unit, and specifically comprises:
a reference center acquisition unit for acquiring a center coordinate of a reference orthophoto map; since the reference orthophoto map is an orthophoto image, the longitude and latitude of the unmanned aerial vehicle during shooting can be regarded as the center coordinates of the reference orthophoto map as an orthophoto image. In actual operation, the central coordinate of the reference ortho-image map can be generated according to the longitude and latitude of the acquisition equipment corresponding to the reference ortho-image map; that is, the center coordinates may be represented by latitude and longitude.
A reference image extracting unit for extracting a reference orthophoto map closest to the coordinates of the target building according to the center coordinates of the reference orthophoto map; it should be noted that the reference ortho-image maps may be prestored in the database, and the database also records the center coordinates of each reference ortho-image map, so that the reference ortho-image map closest to the coordinates of the target building can be quickly extracted by searching the database.
A reference point positioning unit for positioning a reference point consistent with the coordinates of the target building in the extracted reference orthophoto map;
and the reference domain cutting unit is used for cutting a reference domain in a preset range by taking the reference point as the center in the extracted reference orthophoto map. Since subsequent comparisons consume a large amount of computational power, the region of interest can be narrowed from the reference orthographic image to a smaller region (i.e., the reference region) in order to reduce unnecessary consumption and increase speed. For example, a square reference area with a side length of 4cm is cut out with the reference point as the center of the square (see fig. 11); for another example, a circular reference region having a side length of 5cm is cut out with the reference point as the center of the circle.
Therefore, through the targeted construction of the reference region, unnecessary operation consumption can be effectively reduced, and the comparison speed can be improved.
Fourth, compare the module
The comparison module 5 comprises a feature matching unit and a judging unit, specifically:
the characteristic matching unit is used for matching the top image of the target building with the corresponding reference area in a characteristic point matching mode;
a determination unit configured to determine whether a reference building top image similar to a top image of the target building exists in the reference area; if yes, the target building is not established; and if not, indicating that the target building is illegal.
Through a feature point algorithm, feature points in the top image of the target building and feature points in a reference area can be extracted; then, matching the characteristic points in the top image with the characteristic points in the reference area; at this time, if a corresponding target building is matched, it is considered that the target building is also present in the reference orthographic projection image, and therefore the target building is not created (see fig. 13). If the corresponding target building cannot be matched, it is determined that the target building does not exist in the reference orthophoto map, and therefore the target building is illegal (see fig. 14).
Further, the matching proportion of the characteristic points can be manually set or adaptively adjusted; the self-adaptation means that if the building which is identified at this time is actually not built after manual judgment, the requirement of the matching proportion of the feature points can be reduced; if the building not established at this time is not identified, but the establishment point is found manually, the threshold value can be increased.
In conclusion, the top images of all the target buildings in the target orthophoto map are efficiently and accurately determined by identifying the top images of the target buildings, so that the effective positioning of the target buildings is realized by calculating the coordinates of the target buildings; meanwhile, the top image of the target building is accurately compared with the reference area through the targeted construction of the reference area, so that unnecessary operation consumption can be effectively reduced, the comparison speed and precision can be improved, and the cost is low. Furthermore, after the target ortho-image is subjected to scaling and rotation processing, the reference ortho-image and the target ortho-image can adopt the same height and direction standards, so that the subsequent comparison and identification are facilitated; in addition, through the feature point algorithm, the feature points in the top image of the target building and the feature points in the reference region can be extracted for comparison, and the accuracy is high.
Correspondingly, the invention also discloses computer equipment which comprises a memory and a processor, wherein the memory stores a computer program, and the processor realizes the steps of the violation detection method when executing the computer program. Meanwhile, the invention also discloses a computer readable storage medium, on which a computer program is stored, wherein the computer program, when being executed by a processor, realizes the steps of the above violation detection method.
While the foregoing is directed to the preferred embodiment of the present invention, it will be understood by those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the invention.
Claims (11)
1. A method of detection of an illegal build, comprising:
acquiring an aerial photography orthographic image set of a target area, wherein the aerial photography orthographic image set comprises a reference aerial photography orthographic image set and a target aerial photography orthographic image set, the reference aerial photography orthographic image set comprises at least one reference orthographic image, and the target aerial photography orthographic image set comprises at least one target orthographic image;
carrying out normalization processing on the target orthophoto map;
Identifying a top image of a target building in the target orthophoto map, and calculating coordinates of the target building;
extracting a reference area corresponding to the target building according to the coordinates of the target building in the reference aerial orthographic image set;
and comparing the top image of the target building with the corresponding reference area to judge whether the target building is illegal.
2. The method of claim 1, wherein normalizing the target orthophoto map comprises:
carrying out scaling processing on the target ortho-image map so as to enable actual distances represented by pixel points in the target ortho-image map and the reference ortho-image map on the same ground surface to be consistent;
and performing rotation processing on the target orthophoto map so as to enable the direction of the same ground surface in the target orthophoto map and the reference orthophoto map to be consistent.
3. The method of claim 2, wherein the step of scaling the target orthophoto map comprises:
calculating the target size of the target orthophoto map according to the picture size of the reference orthophoto map, the shooting height of the reference orthophoto map and the shooting height of the target orthophoto map;
And scaling the target orthophoto map according to the target size.
4. The method of claim 2, wherein the step of rotating the target orthophoto map comprises:
acquiring a yaw angle of acquisition equipment corresponding to the target orthophoto map;
and rotating the target orthophoto map according to the yaw angle.
5. The method of claim 1, wherein the step of identifying the overhead image of the target structure in the orthophoto map and calculating the coordinates of the target structure comprises:
acquiring the central coordinate of the target orthophoto map;
identifying a top image of a target building in the target orthophoto map through deep learning;
acquiring the actual distance represented by the pixel points in the normalized target orthophoto map;
extracting the relative position of the target building and the central coordinate in the normalized target orthophoto map;
and calculating the coordinates of the target building according to the actual distance, the relative position and the center coordinates.
6. The method of claim 1, wherein the step of extracting the reference area corresponding to the target building from the coordinates of the target building comprises:
Acquiring the central coordinate of the reference orthophoto map;
extracting a reference orthophoto map closest to the coordinates of the target building according to the central coordinates of the reference orthophoto map;
locating a reference point consistent with the coordinates of the target building in the extracted reference orthographic projection image;
and cutting out a reference area in a preset range by taking the reference point as a center in the extracted reference orthographic projection image.
7. The method of claim 5 or 6, wherein the method of obtaining the center coordinates of the target or reference orthophoto map comprises: and generating the central coordinate of the target orthophoto map or the reference orthophoto map according to the longitude and latitude of the acquisition equipment corresponding to the target orthophoto map or the reference orthophoto map.
8. The illegal construction detection method of claim 1, wherein the step of comparing the top image of the target building with the corresponding reference area to determine whether the target building is illegal comprises:
matching the top image of the target building with the corresponding reference area in a characteristic point matching mode;
Judging whether a reference building top image similar to the top image of the target building exists in the reference area or not;
if so, indicating that the target building has no illegal construction;
and if not, indicating that the target building is illegal.
9. A breach detection system, comprising:
the system comprises an acquisition module, a processing module and a display module, wherein the acquisition module is used for acquiring an aerial orthographic image set of a target area, the aerial orthographic image set comprises a reference aerial orthographic image set and a target aerial orthographic image set, the reference aerial orthographic image set comprises at least one reference orthographic image, and the target aerial orthographic image set comprises at least one target orthographic image;
the normalizing module is used for carrying out normalization processing on the target orthophoto map;
the identification module is used for identifying the top image of the target building in the target orthophoto map and calculating the coordinates of the target building;
the reference module is used for extracting a reference area corresponding to the target building according to the coordinates of the target building in the reference aerial orthographic image set;
and the comparison module is used for comparing the top image of the target building with the corresponding reference area so as to judge whether the target building is illegal.
10. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor, when executing the computer program, implements the steps of the method of any of claims 1 to 8.
11. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210168689.3A CN114677600A (en) | 2022-02-23 | 2022-02-23 | Illegal construction detection method, illegal construction detection system, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210168689.3A CN114677600A (en) | 2022-02-23 | 2022-02-23 | Illegal construction detection method, illegal construction detection system, computer equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114677600A true CN114677600A (en) | 2022-06-28 |
Family
ID=82072525
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210168689.3A Pending CN114677600A (en) | 2022-02-23 | 2022-02-23 | Illegal construction detection method, illegal construction detection system, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114677600A (en) |
-
2022
- 2022-02-23 CN CN202210168689.3A patent/CN114677600A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110070615B (en) | Multi-camera cooperation-based panoramic vision SLAM method | |
| US10086955B2 (en) | Pattern-based camera pose estimation system | |
| CN108810473B (en) | Method and system for realizing GPS mapping camera picture coordinate on mobile platform | |
| US10451403B2 (en) | Structure-based camera pose estimation system | |
| KR102200299B1 (en) | A system implementing management solution of road facility based on 3D-VR multi-sensor system and a method thereof | |
| US9858669B2 (en) | Optimized camera pose estimation system | |
| CN109871739B (en) | Automatic target detection and space positioning method for mobile station based on YOLO-SIOCTL | |
| CN114004977A (en) | Aerial photography data target positioning method and system based on deep learning | |
| EP4068210A1 (en) | System and method for automated estimation of 3d orientation of a physical asset | |
| Cosido et al. | Hybridization of convergent photogrammetry, computer vision, and artificial intelligence for digital documentation of cultural heritage-a case study: the magdalena palace | |
| CN113920263A (en) | Map construction method, map construction device, map construction equipment and storage medium | |
| CN112700498A (en) | Wind driven generator blade tip positioning method and system based on deep learning | |
| CN113034347B (en) | Oblique photography image processing method, device, processing equipment and storage medium | |
| CN114777768A (en) | High-precision positioning method and system for satellite rejection environment and electronic equipment | |
| CN116594419A (en) | Routing inspection route planning method and device, electronic equipment and storage medium | |
| CN114677600A (en) | Illegal construction detection method, illegal construction detection system, computer equipment and storage medium | |
| CN115631240B (en) | Visual positioning data processing method for large-scale scene | |
| CN116152783B (en) | Target element labeling data acquisition method, computer equipment and storage medium | |
| CN116518981B (en) | Aircraft visual navigation method based on deep learning matching and Kalman filtering | |
| CN112183171B (en) | Method and device for building beacon map based on visual beacon | |
| US20220222909A1 (en) | Systems and Methods for Adjusting Model Locations and Scales Using Point Clouds | |
| Choi et al. | Identifying CCTV Surveillance Coverage Using MMS-Acquired Point Cloud and CCTV Images | |
| CN117635865A (en) | Map true value generation method, device, equipment and storage medium | |
| CN116580152A (en) | Three-dimensional modeling method and device for large-resolution oblique photographic image | |
| CN115457129A (en) | Aircraft positioning method and device, electronic equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |