CN114821334B - Roof capping illegal construction identification method based on regional positioning and local feature matching - Google Patents
Roof capping illegal construction identification method based on regional positioning and local feature matching Download PDFInfo
- Publication number
- CN114821334B CN114821334B CN202210540282.9A CN202210540282A CN114821334B CN 114821334 B CN114821334 B CN 114821334B CN 202210540282 A CN202210540282 A CN 202210540282A CN 114821334 B CN114821334 B CN 114821334B
- Authority
- CN
- China
- Prior art keywords
- building
- roof
- image
- gradient
- corner
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 38
- 238000010276 construction Methods 0.000 title abstract description 4
- 239000013598 vector Substances 0.000 claims abstract description 52
- 230000008859 change Effects 0.000 claims abstract description 32
- 238000011524 similarity measure Methods 0.000 claims abstract description 19
- 238000001514 detection method Methods 0.000 claims description 25
- 238000004364 calculation method Methods 0.000 claims description 16
- 230000011218 segmentation Effects 0.000 claims description 11
- 230000003595 spectral effect Effects 0.000 claims description 7
- 230000008569 process Effects 0.000 claims description 5
- 238000003708 edge detection Methods 0.000 claims description 4
- 238000007781 pre-processing Methods 0.000 claims description 4
- 230000009466 transformation Effects 0.000 claims description 4
- 238000000605 extraction Methods 0.000 abstract description 7
- 230000000694 effects Effects 0.000 abstract description 5
- 230000000875 corresponding effect Effects 0.000 description 5
- 238000012545 processing Methods 0.000 description 5
- 238000012544 monitoring process Methods 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 238000001228 spectrum Methods 0.000 description 3
- 230000006978 adaptation Effects 0.000 description 2
- 230000001419 dependent effect Effects 0.000 description 2
- 238000006073 displacement reaction Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 230000002596 correlated effect Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
- 238000005316 response function Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000035939 shock Effects 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
- 238000012549 training Methods 0.000 description 1
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
Abstract
The invention discloses a roof capping illegal construction identification method based on regional positioning and local feature matching, which comprises the steps of utilizing existing building base vector data to encode a building, and respectively positioning roof regions of the same building on front and rear images; carrying out local feature extraction on the located two-stage images and the same building roof area; based on the extracted characteristic information, respectively constructing front-stage and rear-stage orthographic image building top region characteristic vectors, and calculating similarity measure between the two characteristic vectors; and taking the calculated similarity measure as a change threshold value, extracting a changed building roof area, identifying a roof capped illegal building image and the like. The remarkable effects are as follows: the method can accurately detect the roof capped illegal buildings and provides powerful support for the administrative law enforcement of the illegal buildings.
Description
Technical Field
The invention relates to the technical field of remote sensing image processing, in particular to a roof capping illegal building identification method based on regional positioning and local feature matching.
Background
Common roof capping illegal buildings in cities have risks of exceeding the overall bearing capacity of the building and affecting the overall shock resistance of the building, are easy to generate secondary disasters, threaten the safety of the building and the life and property safety of residents of the whole building, and have to be strictly regulated. The existing monitoring work of the urban illegal buildings mainly relies on means of on-site inspection, manual visual comparison of multi-time-phase remote sensing images for inspecting suspected illegal buildings and the like, so that the problems of large investment of human resources, low efficiency and blind, messy and slow are extremely remarkable, and automatic detection and extraction technology of large-area suspected roof capping illegal buildings is urgently needed.
The existing automatic detection and extraction technology of the building generally detects the integral change of the building from the remote sensing image, including the new addition, disappearance and appearance change of the building, and the monitoring of illegal buildings focuses on extracting the new addition of the building, and no accurate method for identifying the capping illegal building on the high-resolution remote sensing image exists. Meanwhile, from the analysis granularity angle, the existing multi-temporal remote sensing image building change detection method is mainly divided into three types, namely scene-based, object-oriented and pixel-based: classifying scenes formed by combining multiple types of ground objects based on a scene change detection method, but not suitable for extracting single targets; the pixel-based change detection method mainly utilizes a deep learning algorithm to extract abstract features of complex ground objects in a multi-level manner and output change detection results end to end, but the method is too dependent on earlier training samples, and the effect is not ideal in a complex environment; the object-oriented method can be roughly divided into detection after segmentation and detection after classification according to different change detection strategies, wherein the detection method is used for respectively segmenting and extracting multi-temporal images or directly comparing the multi-temporal images after mutually overlapping and segmenting and extracting target objects, and the change detection precision is seriously dependent on segmentation results; the latter classifies the object-oriented on different simultaneous images, and then determines whether to change or not according to the comparison analysis of geometric shapes, contextual characteristics and the like.
The methods have limitations, and the problems of building contour shape and structure difference, shadow, shielding and the like of roof covering type illegal building detection and extraction based on high-resolution remote sensing images are more difficult compared with the detection of the change of other ground objects, so that the interference of geometric displacement, shadow shielding and other factors on the images caused by poor projection is particularly important to be reduced.
Disclosure of Invention
Aiming at the problem that the prior art does not have a method for pertinently extracting roof capped illegal buildings, the invention aims to provide a method for identifying roof capped illegal buildings based on region positioning and local feature matching, which is used for positioning the roof region of the building on the front and rear high-resolution remote sensing images through the existing building base vector data, and then combining the characteristic information of the local region to perform the change detection of the roof of the building, and automatically detecting and extracting the roof capped suspected illegal building target, thereby achieving the purpose of accurately detecting the roof capped illegal buildings.
In order to achieve the above purpose, the invention adopts the following technical scheme:
a roof capping illegal construction identification method based on regional positioning and local feature matching is characterized by comprising the following steps:
step 1: encoding the building by using the existing building base vector data, and respectively positioning the roof areas of the same building on the front and rear images;
step 2: carrying out local feature extraction on the two-stage images positioned in the step 1 and the same building roof area;
step 3: based on the feature information extracted in the step 2, respectively constructing front and rear orthophoto building top region feature vectors, and calculating similarity measure between the two feature vectors;
step 4: and (3) comparing the similarity measure calculated in the step (3) with a change threshold value, extracting a changed building roof area, and identifying a roof capping type illegal building image.
Further, the roof area positioning step of the building in the step 1 is as follows:
step 1.1: performing edge detection on the high-resolution image by using a Gaussian Laplace operator, and extracting an initial edge;
step 1.2: carrying out probability Hough transformation linear detection on the initial edge of the extracted image to obtain a fitted line segment set;
step 1.3: detecting corner points by using a Harris corner point detection algorithm based on the initial edge of the extracted image and the fitted line segment set, and taking the corner points as candidate point sets of building roof vector vertexes;
step 1.4: and encoding the base vector image spots of each building, and carrying out matching measure calculation in a search window through direction constraint and corner constraint so as to detect the position of the roof area of the corresponding building.
Further, the direction constraint in step 1.4 is the cosine similarity absolute value of the edge line and the contour vector line in the contour vector buffer of the building, and the corner constraint includes that the euclidean distance between the vertex S of the contour vector of the building and the corner D in the contour vector buffer is the smallest, and the matching degree of the gradient in k×k field of the vertex S, i.e. k×k rectangle at each corner of the contour buffer, and the contour corner gradient template.
Further, in the step 1.4, the matching policy calculation is a sum of weighted calculation after normalizing the direction constraint and the corner constraint respectively, and the calculation formula is as follows:
G (x，y) ＝w 1 I cos +w 2 I c ，
wherein ,G(x,y) Is the calculated value, w, of the matching degree when the center point of the profile falls on (x, y) 1 、w 2 Is the empirical weight, I cos For the average of the cosine similarity absolute values of all n edge lines and contour vector lines in the buffer,I c is the average value of the matching degree of the gradient value of the corner point and the gradient template of the corner point.
Further, the specific process of extracting the local features of the roof area of the same building on the two-phase images in the step 2 is as follows:
step 2.1: performing image preprocessing on the front and rear original images of the determined roof area;
step 2.2: performing super-pixel segmentation on the preprocessed image by adopting an SLIC algorithm;
step 2.3: and extracting the characteristics of the image after the super-pixel segmentation.
Further, the features extracted in step 2.3 include spectral features, texture features and directional gradient histograms.
Further, the step of obtaining the direction gradient histogram is as follows:
step A1: dividing the image into a plurality of small connected areas;
step A2: collecting gradient or edge direction histograms of all pixel points in all connected areas;
step A3: the obtained direction histogram is formed into a direction gradient histogram.
Further, the calculation formula of the similarity measure in step 3 is:
wherein ,is the post-image feature vector->And the earlier image feature vector->Similarity measure between, ∈>Is spectral feature->LBP characterization->Directional gradient characteristics->And combining the feature space vectors.
Further, in the step 4, the specific steps of extracting the changed building roof area by using the similarity measure as the change threshold value are as follows:
utilizing the similarity measure values in the step 3, and utilizing different similarity measure values as change threshold values to change the superpixel of each roof area;
by adjusting the appropriate change threshold, a changed roof area of the building is obtained, thereby identifying the roof capped illegal building.
The invention has the remarkable effects that:
1. the spatial information provided by the existing vector data and the characteristic information existing in the high-resolution remote sensing image data are fully utilized, the roof area of the building is rapidly positioned by an optimal matching method, the change condition of the area is identified by utilizing the characteristic information, the roof capped illegal building can be accurately detected, and a powerful support is provided for the administrative law enforcement of the illegal building;
2. the identification method of the roof capped illegal building can also be used for identifying other types of local area change detection, has universal applicability and can generate important market value;
3. the method realizes the positioning of the roof area of the building in the early and later high-resolution remote sensing images, can automatically identify and extract newly added roof capped illegal buildings in the area, and has important significance for developing dynamic monitoring and administrative law enforcement of the illegal buildings of the type.
Drawings
FIG. 1 is a flow chart of the method of the present invention;
fig. 2 is a schematic diagram of a contour vector buffer, corner constraints, and direction constraints for a building base vector contour.
Detailed Description
The following describes the embodiments and working principles of the present invention in further detail with reference to the drawings.
As shown in FIG. 1, the method for identifying the roof capping violations based on the regional positioning and the local feature matching comprises the following specific processes:
step 1: encoding the building by using the existing building base vector data, and respectively positioning the roof areas of the same building on the front and rear images;
because of the projection difference on the high-resolution remote sensing images of different time phases, the geometric displacement of different degrees exists between the vector data of the basic building and the high-resolution remote sensing images, so that the searching of the features of the corresponding building roofs on the two-phase images is very difficult. In order to eliminate the influence of projection difference on the change detection of a building and reduce the interference of factors such as shadow shielding, the invention provides a method for finding out a corresponding building roof area on a front-and-back high-resolution remote sensing image by taking building base vector data as assistance, fully utilizing the advantages of the two data and improving the detection precision and efficiency.
The positioning of roof areas of the same building on two-phase images through the existing building base vector information is specifically realized as follows:
step 1.1: performing edge detection on the high-resolution image by using a Gaussian Laplace operator, and extracting an initial edge;
the Gaussian Laplace operator (Log) is a Gaussian smoothing process of the image I (x, y) with the scale sigma to suppress noise interferenceThen find its Laplace second derivative to detect the edge +.>I.e. Log operator->Then, carrying out normalization processing on the Log operator detection result, and setting a threshold value threshold according to actual situation requirements to enable +.> The finally obtained binary image is the extracted initial edge.
Step 1.2: the initial edge of the extracted image is a point set, probability Hough transformation (HHPT) straight line detection is carried out on the extracted image to obtain a complete line segment, the initial edge obtained in the step 1.1 is mapped to a Hough space, a proper threshold value is set, part of interference straight lines are filtered, and the image space is mapped back to obtain a fitted line segment set L;
step 1.3: detecting corner points by using a Harris corner point detection algorithm based on the initial edge of the extracted image and the fitted line segment set, and taking the corner points as candidate point sets of building roof vector vertexes;
calculating the gradient value and the angular point response function value R of each pixel in the x and y directions in the initial edge image obtained in the step 1.1,λ 1 、λ 2 is the eigenvalue of the local eigenvalue matrix M. Extracting features by setting an appropriate R thresholdAnd the feature points are output by taking the largest feature point in each 5X 5 field as the corner point, so as to obtain a corner point set D.
Step 1.4: encoding each building base vector image spot, and carrying out matching measure calculation in a search window through direction constraint and corner constraint so as to detect the position of a corresponding building roof area, wherein the matching measure comprises the following steps:
as shown in fig. 2, a buffer region (shown in fig. 2 (b)) of a distance k is set around a building contour vector map (shown in fig. 2 (a)), and a vertex point set S of the contour vector is extracted, and then a center point (x) of the contour vector is set 0 ，y 0 ) The search window is established with the set distance as a radius for the center. In the search window, corner constraints (as shown in fig. 2 (c)) and direction constraints (as shown in fig. 2 (d)) of the contour vector are set to calculate a matching measure, and non-maximum suppression processing of the matching measure is performed to find the roof area position that best matches the building base vector.
wherein ：
the corner constraint comprises the minimum Euclidean distance between the building contour vector vertex S and the corner D in the contour vector buffer zone, and the matching degree of gradient in k x k rectangular in k x k field of the vertex S, namely each corner of the contour buffer zone, and the contour corner gradient template.
The direction constraint is the cosine similarity absolute value of the edge straight line L and the contour vector straight line in the contour vector buffer area of the building.
The matching measure calculation is the sum of weighted calculation after respectively normalizing the direction constraint and the angular point constraint. G% x ， y )＝w 1 I cos +w 2 I c ，G (x,y) Calculated as the matching degree when the center point of the profile falls on (x, y), w 1 、w 2 Is the empirical weight, I cos Is the average value of cosine similarity absolute values of all n edge straight lines L and contour vector straight lines in the buffer zoneI c Average of gradient value of corner and matching degree of corner gradient templateValues.
The candidate edges and the corner points of the roof area can be extracted through the method, and then the roof area positioning of the building is realized through the matching algorithm in the step 1.4, so that the candidate edges and the corner points are used as a basis for identifying the roof covered illegal building target from the roof area. Step 1.2 is further processing of step 1.1, step 1.1 is a set of extracted points, and step 1.2 fits the set of points to a line segment as an edge of a shape; step 1.3 is to extract the corner points, i.e. the vertices of the roof area vector, based on the set of points obtained in step 1.1, which are constrained by the edge line segments formed in step 1.2, so the order of the three methods is thus determined.
Step 2: and (3) carrying out local feature extraction on the two-stage images positioned in the step (1) and the same building roof area, wherein the specific process is as follows:
step 2.1: performing image preprocessing on the front and rear original images of the determined roof area;
before the image of the roof area of the building obtained by positioning is segmented, the original image is subjected to proper pretreatment, including graying, histogram equalization, filtering treatment and the like, in order to improve the image quality and facilitate the image processing.
Step 2.2: performing super-pixel segmentation on the preprocessed image by adopting an SLIC algorithm;
the image obtained by the preprocessing is used as an image to be subjected to target segmentation, and a SLIC (simple linear iterative clustering) algorithm with a good edge contour retaining effect is used.
Taking the number k=n/256 of the super pixels, wherein N is the total number of pixels of the roof area on the image, 256 is the number of pixels contained in the segmented super pixels, and then the distance (step length) between adjacent cluster centers is approximately s=sqrt (N/K), and sampling initialization clustering is carried out in the step length of S. And calculating gradient values of all pixel points in a 3*3 neighborhood of the clustering center, and moving the clustering center to a position with the minimum gradient in the neighborhood to obtain a new clustering center, wherein the pixel points in the field of each clustering center obtain the same-class labels. The expected superpixel is an area approximating sxs, limiting the size of the search area to reduce the number of distance calculations, accelerating algorithm convergenceSearching in a region 2S multiplied by 2S around the super pixel center, calculating the distance D from each pixel point in the region to the clustering center,d c 、d s respectively the color distance and the space distance between pixels, N c 、N s The maximum color distance and the maximum spatial distance within the class, respectively, the distance D taking into account both color similarity and spatial proximity. d, d c ＝|l j -l i |，l j 、l i Respectively gray values of two pixels; />x and y are respectively the positions of the horizontal and vertical coordinates of the pixel point; at the same time, to simplify the calculation, make N s ＝S，N c M, m is a constant, the relative importance of color similarity and spatial proximity can be set by adjusting the size of m, where m=10. Each pixel point is searched by a plurality of cluster centers, and the cluster center corresponding to the minimum value of D is taken as the category to which the pixel point belongs. The steps are iterated continuously until the clustering center of each pixel point is not changed any more, and the clustering of each pixel is the segmented super pixel.
Meanwhile, in order to enhance comparability and computational consistency between the front and rear remote sensing image super pixels, a final segmentation result is obtained through superposition analysis.
Step 2.3: and extracting the characteristics of the image after the super-pixel segmentation.
The high-resolution remote sensing images have various ground objects, contain complex and various information, and can possibly cause target extraction errors due to insufficient description of single objects or areas based on spectral features, so that the image objects are described by combining structural information such as spectra, textures, direction gradient Histograms (HOGs) and the like.
The spectrum characteristic is the combination of the normalized value of the gray average of all the pixel channels in the object and the HSV color characteristic. For visible light three-band image wherein ,respectively representing RGB average values of all pixels in the ith object; />The HSV mean values of all pixels in the ith object, respectively. The HSV color features describe an image from three dimensions of hue (H), saturation (S) and brightness (V), and the hue value H, the saturation value S and the brightness value V of each pixel are obtained by converting the image from an RGB space to an HSV space, so that the overall brightness degree of the image is reflected.
The texture features are Local Binary Pattern (LBP) texture description operator processed LBP features, the local texture features of the image are represented by comparing the relationship between a central pixel and surrounding pixels, the method has the remarkable advantages of rotation invariance, gray invariance and the like, the calculation is simple, the efficiency is high, and the local texture features can be extracted well.
The HOG features describe the appearance and shape of local objects by gradient or edge direction density distribution, and features are formed by calculating and counting gradient direction histograms of local areas of the image. First, dividing the image into small connected areas (cell units); then collecting a gradient or edge direction histogram of each pixel point in the cell unit; finally, the histograms are combined to form the feature descriptor. Since HOG is operated on a local square cell of the image, it remains well invariant to both geometric and optical deformations of the image.
The three features comprise the most important three aspects of the image, namely color, texture and structural features, so that the accuracy of the identification result can be effectively ensured.
Step 3: based on the feature information extracted in the step 2, respectively constructing feature vectors of super-pixel areas with homogeneous pixels at the tops of buildings in the front-stage orthographic images and the rear-stage orthographic images, and calculating similarity measure between the two feature vectors;
from step 2, spectral features can be obtainedLBP characterization->HOG characteristics->Combining three feature information vectors of the front and rear orthophoto local area into a feature space vector +.> The range of the correlation coefficient r between the image characteristic space vectors in the front and rear stages is calculated, wherein the range of r is (-1, 1), 1 is basically uncorrelated, and 1 is almost correlated.
Step 4: and (3) comparing the similarity measure calculated in the step (3) with a change threshold value, extracting a changed building roof area, and identifying a roof capping type illegal building image.
And (3) observing the change condition of each super-pixel region under the condition of different change thresholds by using the similarity measure value r in the step (3). By adjusting the proper change threshold, the changed roof area of the building is obtained, so that the roof capping illegal building is identified.
The adjustment mode of the change threshold value is as follows: and judging whether the super pixel area is correctly changed when r is larger than the change threshold value through visual interpretation, and obtaining the value with the best judging effect, namely the change threshold value with proper r value.
The method comprises the steps of combining existing vector data with edge detection, hough linear transformation and Harris corner detection, calculating a matching measure by using direction constraint and corner constraint, and accurately positioning a roof area of a building; and detecting the newly added roof capping illegal building through characteristic matching of super-pixel segmentation and local characteristics such as spectrum characteristics, local Binary Pattern (LBP) characteristics, direction gradient Histogram (HOG) characteristics and the like. The method realizes the positioning of the roof area of the building in the early and later high-resolution remote sensing images, can automatically identify and extract newly added roof capped illegal buildings in the area, and has important significance for developing dynamic monitoring and administrative law enforcement of the illegal buildings of the type.
The technical scheme provided by the invention is described in detail. The principles and embodiments of the present invention have been described herein with reference to specific examples, the description of which is intended only to facilitate an understanding of the method of the present invention and its core ideas. It should be noted that it will be apparent to those skilled in the art that various modifications and adaptations of the invention can be made without departing from the principles of the invention and these modifications and adaptations are intended to be within the scope of the invention as defined in the following claims.
Claims (5)
1. A roof capping illegal building identification method based on regional positioning and local feature matching is characterized by comprising the following steps:
step 1: encoding the building by using the existing building base vector data, and respectively positioning the roof areas of the same building on the front and rear images;
the roof area positioning step of the building in the step 1 is as follows:
step 1.1: performing edge detection on the high-resolution image by using a Gaussian Laplace operator, and extracting an initial edge;
step 1.2: carrying out probability Hough transformation linear detection on the initial edge of the extracted image to obtain a fitted line segment set;
step 1.3: detecting corner points by using a Harris corner point detection algorithm based on the initial edge of the extracted image and the fitted line segment set, and taking the corner points as candidate point sets of building roof vector vertexes;
step 1.4: encoding the base vector image spots of each building, and carrying out matching measure calculation in a search window through direction constraint and angular point constraint so as to detect the position of the roof area of the corresponding building;
and (2) in the step (1.4), the matching measure calculation is a weighted calculation sum after respectively normalizing the direction constraint and the angular point constraint, and the calculation formula is as follows:
G (x，y) ＝w 1 I cos +w 2 I c ，
wherein ,G(x,y) Is the calculated value, w, of the matching degree when the center point of the profile falls on (x, y) 1 、w 2 Is the empirical weight, I cos For the average of the cosine similarity absolute values of all n edge lines and contour vector lines in the buffer,I c the average value of the gradient value of the corner point and the matching degree of the gradient template of the corner point;
step 2: extracting local features of the same building roof region of the two-stage images positioned in the step 1, wherein the local features comprise spectral features, texture features and a direction gradient histogram, and the spectral features are the combination of normalized values of gray level averages of all channels of all pixels in an object and HSV color features; the texture features are Local Binary Pattern (LBP) texture description operator processed LBP features using a rotation invariant pattern; the direction gradient histogram describes the appearance and shape of a local target by using the gradient or the direction density distribution of the edge, and the characteristic is formed by calculating and counting the gradient direction histogram of the local area of the image;
step 3: based on the feature information extracted in the step 2, respectively constructing front and rear orthophoto building top region feature vectors, and calculating similarity measure between the two feature vectors;
the calculation formula of the similarity measure in the step 3 is as follows:
wherein ,is the post-image feature vector->And the earlier image feature vector->Similarity measure between, ∈>Is spectral feature->LBP characterization->Directional gradient characteristics->Combining the feature space vectors;
step 4: and (3) comparing the similarity measure calculated in the step (3) with a change threshold value, extracting a changed building roof area, and identifying a roof capping type illegal building image.
2. The method for identifying the roof capping violations based on regional positioning and local feature matching according to claim 1, wherein the method comprises the following steps: the direction constraint in step 1.4 is the cosine similarity absolute value of the edge line and the contour vector line in the contour vector buffer of the building, the corner constraint comprises the minimum euclidean distance between the vertex S of the contour vector of the building and the corner D in the contour vector buffer, and the matching degree of the gradient in the k×k field of the vertex S, i.e. the k×k rectangle at each corner of the contour buffer, and the contour corner gradient template.
3. The method for identifying the roof capping violations based on regional positioning and local feature matching according to claim 1, wherein the method comprises the following steps: in the step 2, the specific process of extracting the local features of the roof area of the same building on the two-stage images is as follows:
step 2.1: performing image preprocessing on the front and rear original images of the determined roof area;
step 2.2: performing super-pixel segmentation on the preprocessed image by adopting an SLIC algorithm;
step 2.3: and extracting the characteristics of the image after the super-pixel segmentation.
4. The method for identifying the roof capping violations based on regional positioning and local feature matching according to claim 1, wherein the method comprises the following steps: the directional gradient histogram is obtained by the following steps:
step A1: dividing the image into a plurality of small connected areas;
step A2: collecting gradient or edge direction histograms of all pixel points in all connected areas;
step A3: the obtained direction histogram is formed into a direction gradient histogram.
5. The method for identifying the roof capping violations based on regional positioning and local feature matching according to claim 1, wherein the method comprises the following steps: in the step 4, the specific steps of extracting the changed building roof area by using the similarity measure as the change threshold value are as follows:
utilizing the similarity measure values in the step 3, and utilizing different similarity measure values as change threshold values to change the superpixel of each roof area;
and obtaining the changed roof area of the building through the adjusted change threshold value, thereby identifying the roof capping type illegal building.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210540282.9A CN114821334B (en) | 2022-05-17 | 2022-05-17 | Roof capping illegal construction identification method based on regional positioning and local feature matching |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210540282.9A CN114821334B (en) | 2022-05-17 | 2022-05-17 | Roof capping illegal construction identification method based on regional positioning and local feature matching |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114821334A CN114821334A (en) | 2022-07-29 |
| CN114821334B true CN114821334B (en) | 2023-08-04 |
Family
ID=82514488
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210540282.9A Active CN114821334B (en) | 2022-05-17 | 2022-05-17 | Roof capping illegal construction identification method based on regional positioning and local feature matching |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114821334B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2278553A1 (en) * | 2008-04-23 | 2011-01-26 | PASCO Corporation | Building roof outline recognizing device, building roof outline recognizing method, and building roof outline recognizing program |
| KR101380329B1 (en) * | 2013-02-08 | 2014-04-02 | (주)나노디지텍 | Method for detecting change of image |
| CN107092877A (en) * | 2017-04-12 | 2017-08-25 | 武汉大学 | Remote sensing image roof contour extracting method based on basement bottom of the building vector |
| CN109919944A (en) * | 2018-12-29 | 2019-06-21 | 武汉大学 | A kind of joint super-pixel figure of complex scene building variation detection cuts optimization method |
| CN112862774A (en) * | 2021-02-02 | 2021-05-28 | 重庆市地理信息和遥感应用中心 | Accurate segmentation method for remote sensing image building |
-
2022
- 2022-05-17 CN CN202210540282.9A patent/CN114821334B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2278553A1 (en) * | 2008-04-23 | 2011-01-26 | PASCO Corporation | Building roof outline recognizing device, building roof outline recognizing method, and building roof outline recognizing program |
| KR101380329B1 (en) * | 2013-02-08 | 2014-04-02 | (주)나노디지텍 | Method for detecting change of image |
| CN107092877A (en) * | 2017-04-12 | 2017-08-25 | 武汉大学 | Remote sensing image roof contour extracting method based on basement bottom of the building vector |
| CN109919944A (en) * | 2018-12-29 | 2019-06-21 | 武汉大学 | A kind of joint super-pixel figure of complex scene building variation detection cuts optimization method |
| CN112862774A (en) * | 2021-02-02 | 2021-05-28 | 重庆市地理信息和遥感应用中心 | Accurate segmentation method for remote sensing image building |
Non-Patent Citations (1)
| Title |
|---|
| 国土资源典型要素变化遥感智能监测关键技术及应用;丁忆 等;地理信息世界;第1-2部分 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114821334A (en) | 2022-07-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107610114B (en) | optical satellite remote sensing image cloud and snow fog detection method based on support vector machine | |
| CN106650731B (en) | Robust license plate and vehicle logo recognition method | |
| CN109409355B (en) | Novel transformer nameplate identification method and device | |
| CN107992856B (en) | High-resolution remote sensing building shadow detection method under urban scene | |
| CN113963222B (en) | High-resolution remote sensing image change detection method based on multi-strategy combination | |
| CN110659649A (en) | Image processing and character recognition algorithm based on near infrared light imaging | |
| CN110084587B (en) | Automatic dinner plate settlement method based on edge context | |
| CN111047616B (en) | Remote sensing image landslide target constraint active contour feature extraction method | |
| CN110929739B (en) | Automatic impervious surface range remote sensing iterative extraction method | |
| CN114821334B (en) | Roof capping illegal construction identification method based on regional positioning and local feature matching | |
| CN110458064B (en) | Low-altitude target detection and identification method combining data driving type and knowledge driving type | |
| CN110276270B (en) | High-resolution remote sensing image building area extraction method | |
| CN109829511B (en) | Texture classification-based method for detecting cloud layer area in downward-looking infrared image | |
| CN111539330A (en) | Transformer substation digital display instrument identification method based on double-SVM multi-classifier | |
| CN111476233A (en) | License plate number positioning method and device | |
| CN114758139B (en) | Method for detecting accumulated water in foundation pit | |
| CN110956083A (en) | Bohai sea ice drift remote sensing detection method based on high-resolution four-signal optical satellite | |
| Han et al. | Accurate and robust vanishing point detection method in unstructured road scenes | |
| CN115512310A (en) | Vehicle type recognition method and system based on face features under video monitoring | |
| CN114862883A (en) | Target edge extraction method, image segmentation method and system | |
| CN115019306A (en) | Embedding box label batch identification method and system based on deep learning and machine vision | |
| CN113642430A (en) | High-precision visual positioning method and system for underground parking lot based on VGG + NetVLAD | |
| Huang et al. | Classification of very high spatial resolution imagery based on the fusion of edge and multispectral information | |
| CN113723314A (en) | Sugarcane stem node identification method based on YOLOv3 algorithm | |
| Liu et al. | Identification of Damaged Building Regions from High-Resolution Images Using Superpixel-Based Gradient and Autocorrelation Analysis |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |