A remote sensing image detection method for building changes in airport clear space protection areas

Abstract

The present invention discloses a remote sensing image detection method for building changes in an airport clearance protection zone, comprising the following steps: based on the optical remote sensing images of the previous and later periods, respectively obtaining the morphological building index and morphological shadow index of the previous and later periods; obtaining elevation information based on the SAR remote sensing image pairs of the previous and later periods, and obtaining the image incoherence index based on the SAR remote sensing images of the previous and later periods; performing projection difference correction on the morphological building index and morphological shadow index of the previous and later periods based on the elevation information obtained in the previous and later periods; constructing a decision forest, inputting the corrected morphological building index, morphological shadow index, elevation information and image incoherence index of the previous and later periods into the decision forest, and obtaining identification information of building changes. The present invention realizes the description of building features in high-resolution remote sensing images, realizes intelligent identification and change detection of buildings by using machine learning algorithms, and meets the actual needs of monitoring and inspection of building changes in airport clearance protection zones.

Description

Technical Field

[0001] The invention belongs to the technical field of remote sensing detection, and in particular relates to a remote sensing image detection method for building changes in an airport clearance protection zone. Background Art

[0002] In recent years, with the development of aviation, aerospace, satellite and other technologies, remote sensing change detection technology has become an effective means of monitoring changes on the earth's surface, and it is also one of the research hotspots in the field of remote sensing. Remote sensing change detection technology is a technology that discovers surface changes by analyzing the previous period data and the later period remote sensing images of the same area. After years of development, remote sensing image change detection technology has become increasingly mature in both theory and technology, and has been widely used in building change detection. Change detection algorithms for buildings usually include two categories: one is the direct comparison method, and the other is the post-classification comparison method. The main idea of ​​the direct comparison method is to obtain the content of the changes based on the spectrum, texture, geometric shape and other features of the building in the remote sensing image through direct comparison. However, due to the shooting process of the remote sensing image, external factors often affect the imaging quality of the image, which leads to the low accuracy of the detection results of most algorithms at present, and the false alarm rate of the change information is very high, making it difficult to apply the specific algorithm to practice. The main idea of ​​the post-classification comparison method is to first classify the ground objects in the previous and next phase images, and then compare the building targets in them to find changes. However, the change detection accuracy of this type of method is greatly affected by the classification accuracy, and the difference in shooting angles of buildings in two phases will also have a great impact on the detection results.

[0003] In addition, in remote sensing images of different phases, the types of objects that change are diverse. It is impractical to use only the same rule to detect all types of changes in the image. Through a comprehensive analysis of domestic and foreign research, the current building change detection technology is developing in the following two directions: (1) Change detection technology tends to make comprehensive use of multiple types of source data. The image features of different types of source data can complement each other and thus improve the change detection effect. (2) The specific methods of change detection have evolved from a single detection method to the direction of multiple methods integrating each other. Summary of the invention

[0004] In order to solve the above technical problems, the present invention proposes a remote sensing image detection method for building changes in airport clearance protection zone, which realizes the intelligent, rapid and accurate extraction of building change information in response to the actual needs of building change monitoring in airport clearance protection zone monitoring and inspection.

[0005] To achieve the above object, the present invention provides a remote sensing image detection method for building changes in an airport clearance protection zone, comprising:

[0006] Acquire an early optical remote sensing image and an early SAR remote sensing image pair, as well as a late optical remote sensing image and a late SAR remote sensing image pair, wherein the early period and the late period are respectively the start time and the end time of the time period of the building changes in the airport airspace protection zone to be detected;

[0007] Based on the early optical remote sensing image, an early morphological building index and an early morphological shadow index are obtained, and based on the late optical remote sensing image, a late morphological building index and a late morphological shadow index are obtained;

[0008] Based on the previous SAR remote sensing image pair and the later SAR remote sensing image pair, obtaining elevation information, and based on the previous SAR remote sensing image and the later SAR remote sensing image, obtaining an image decoherence index;

[0009] Based on the elevation information, projection difference correction is performed on the early morphological building index, the late morphological building index, the early morphological shadow index, and the late morphological shadow index;

[0010] Constructing a decision forest;

[0011] The corrected early morphological building index, early morphological shadow index, late morphological building index, late morphological shadow index and image incoherence index are input into the decision forest to obtain identification information of building changes.

[0012] Optionally, the step of obtaining the preliminary morphological building index includes:

[0013] Performing multi-scale segmentation on the previous optical remote sensing image;

[0014] Obtaining the maximum value of the brightness of each pixel in each band in the segmented early optical remote sensing image as the image brightness value in the segmented early optical remote sensing image;

[0015] Performing a white hat transformation on the segmented early optical remote sensing image based on the image brightness value;

[0016] Based on the result of the white hat transformation, the preliminary morphological building index is obtained.

[0017] Optionally, the step of obtaining a later morphological building index comprises:

[0018] Performing multi-scale segmentation on the late optical remote sensing image;

[0019] Obtaining the maximum value of the brightness of each pixel in each band in the segmented late optical remote sensing image as the image brightness value in the segmented late optical remote sensing image;

[0020] Performing a white hat transformation on the segmented late optical remote sensing image based on the image brightness value;

[0021] Based on the result of the white hat transformation, the late morphological architectural index is obtained.

[0022] Optionally, the step of obtaining a preliminary morphological shadow index includes:

[0023] Performing multi-scale segmentation on the previous optical remote sensing image;

[0024] Obtaining the maximum value of the brightness of each pixel in each band in the segmented early optical remote sensing image as the image brightness value in the segmented early optical remote sensing image;

[0025] Performing a black hat transformation on the segmented early optical remote sensing image based on the image brightness value;

[0026] Based on the result of the black hat transformation, the early morphological shadow index is obtained.

[0027] Optionally, the step of obtaining a late morphological shadow index comprises:

[0028] Performing multi-scale segmentation on the late optical remote sensing image;

[0029] Obtaining the maximum value of the brightness of each pixel in each band in the segmented late optical remote sensing image as the image brightness value in the segmented late optical remote sensing image;

[0030] Performing a black hat transformation on the segmented late optical remote sensing image based on the image brightness value;

[0031] Based on the result of the black hat transformation, the late morphological shadow index is obtained.

[0032] Optionally, the elevation information includes early elevation information and late elevation information:

[0033] Performing interference processing on the previous SAR remote sensing image pair to obtain the previous elevation information;

[0034] The late SAR remote sensing image pair is subjected to interference processing to obtain the late elevation information.

[0035] Optionally, the identification information of the building change is: newly added buildings and reduced buildings.

[0036] Optionally, the judgment conditions for adding new buildings and reducing buildings are:

[0037] (1) determining whether the image decoherence index is less than a coherence threshold, and if so, proceeding to the next step;

[0038] (2) respectively comparing the early and late morphological shadow indices after the projection difference correction with the shadow index threshold; if the early morphological shadow index is greater than the shadow index threshold, the early shadow area is extracted; if the late morphological shadow index is greater than the shadow index threshold, the late shadow area is extracted;

[0039] (3) Using double filtering method to identify and extract buildings, the early morphological building index and shadow index, and the late morphological building index and shadow index after projection difference correction are used respectively;

[0040] (4) If a building is identified as a non-building in the previous period but is later identified as a building, it shall be classified as a newly added building; if a building is identified as a building in the previous period but is later identified as a non-building, it shall be classified as a reduced building;

[0041] (5) If a building is identified in the early stage and later stage, the elevation information of the early stage and the elevation information of the later stage are compared. If the elevation increases, it is classified as a newly added building; if the elevation decreases, it is classified as a reduced building.

[0042] Compared with the prior art, the present invention has the following advantages and technical effects:

[0043] Aiming at the actual demand for monitoring building changes in the monitoring and inspection of the airport clearance protection zone, the present invention is based on object-oriented image analysis technology. On the basis of fully analyzing the image features of buildings in high-resolution remote sensing images, mathematical morphological operations are used to realize the description of building features in high-resolution remote sensing images. Based on the early optical remote sensing images, the early morphological building index and the early morphological shadow index are obtained, and based on the late optical remote sensing images, the late morphological building index and the late morphological shadow index are obtained; based on the early SAR remote sensing image pair and the late SAR remote sensing image pair, the elevation information is obtained, and based on the early SAR remote sensing image and the late SAR remote sensing image, the image decoherence index is obtained; based on the elevation information, the early morphological building index, the late morphological building index, the early morphological shadow index, and the late morphological shadow index are corrected for projection differences; based on the above indexes after the projection difference correction, a machine learning algorithm is used to realize intelligent recognition and change detection of buildings, and intelligent, fast and accurate extraction of building change information is realized. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0045] Figure 1 A schematic flow chart of a remote sensing image detection method for building changes in an airport clearance protection zone according to a first embodiment of the present invention;

[0046] Figure 2 A schematic diagram of a multi-scale hierarchical network structure of an image object according to the first embodiment of the present invention;

[0047] Figure 3 This is a schematic diagram of the random forest structure of the first embodiment of the present invention;

[0048] Figure 4 This is a schematic diagram of remote sensing image data according to the first embodiment of the present invention;

[0049] Figure 5 This is the newly added building recognition result of the area in the first embodiment of the present invention. DETAILED DESCRIPTION

[0050] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0051] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0052] Embodiment 1

[0053] like Figure 1 As shown, this embodiment provides a remote sensing image detection method for building changes in an airport clearance protection zone, including:

[0054] Acquire an early optical remote sensing image and an early SAR remote sensing image pair, as well as a late optical remote sensing image and a late SAR remote sensing image pair, wherein the early period and the late period are respectively the start time and the end time of the time period of the building changes in the airport airspace protection zone to be detected;

[0055] Based on the early optical remote sensing image, an early morphological building index and an early morphological shadow index are obtained, and based on the late optical remote sensing image, a late morphological building index and a late morphological shadow index are obtained;

[0056] Based on the previous SAR remote sensing image pair and the later SAR remote sensing image pair, obtaining elevation information, and based on the previous SAR remote sensing image and the later SAR remote sensing image, obtaining an image decoherence index;

[0057] In this embodiment, an interference pair is formed based on two SAR remote sensing images with a relatively short time interval (such as a satellite revisit period of 11 days) to obtain early building elevation information, an interference pair is formed based on two SAR remote sensing images with a relatively short actual time interval to obtain late building elevation information, and an interference pair is formed based on any early SAR remote sensing image and any late SAR remote sensing image with a long time interval (such as half a year) to obtain image decoherence index;

[0058] Based on the elevation information, projection difference correction is performed on the early morphological building index, the late morphological building index, the early morphological shadow index, and the late morphological shadow index;

[0059] Constructing a decision forest;

[0060] The corrected early morphological building index, early morphological shadow index, late morphological building index, late morphological shadow index and image incoherence index are input into the decision forest to obtain identification information of building changes.

[0061] Furthermore, the steps of obtaining the preliminary morphological building index include:

[0062] (1) Perform multi-scale segmentation on previous optical remote sensing images;

[0063] (2) obtaining the maximum value of the brightness of each pixel in each band in the segmented early optical remote sensing image as the image brightness value in the segmented early optical remote sensing image;

[0064] (3) Perform white hat transformation on the segmented early optical remote sensing image based on the image brightness value;

[0065] (4) Based on the results of white hat transformation, the preliminary morphological architectural index is obtained.

[0066] The steps to obtain the post-morphological architectural index include:

[0067] (1) Perform multi-scale segmentation on late-stage optical remote sensing images;

[0068] (2) obtaining the maximum value of the brightness of each pixel in each band in the segmented later optical remote sensing image as the image brightness value in the segmented later optical remote sensing image;

[0069] (3) Perform white hat transformation on the segmented optical remote sensing image based on the image brightness value;

[0070] (4) Based on the results of white hat transformation, the post-morphological architectural index is obtained.

[0071] The steps to obtain the preliminary morphological shadow index include:

[0072] Perform multi-scale segmentation on previous optical remote sensing images;

[0073] Obtain the maximum value of the brightness of each pixel in each band in the segmented early optical remote sensing image as the image brightness value in the segmented early optical remote sensing image;

[0074] Perform black hat transformation on the segmented early optical remote sensing image based on the image brightness value;

[0075] Based on the results of black hat transformation, the preliminary morphological shadow index is obtained.

[0076] The steps to obtain the post-morphological shadow index include:

[0077] Perform multi-scale segmentation on late-stage optical remote sensing images;

[0078] Obtain the maximum value of the brightness of each pixel in each band in the segmented later optical remote sensing image as the image brightness value in the segmented later optical remote sensing image;

[0079] Perform black hat transformation on the segmented post-optical remote sensing image based on the image brightness value;

[0080] Based on the results of black hat transformation, the post-morphological shadow index is obtained.

[0081] Furthermore, the basis and key of object-based image processing and analysis technology lies in the construction of image objects, which is usually achieved by image segmentation technology. Image segmentation divides the image into several non-overlapping sub-regions with certain consistent attributes according to image features, and these sub-regions can be regarded as image processing and analysis primitives, namely image objects.

[0082] Fractal Net Evolution Approach (FNEA), also known as Multi-Resolution Segmentation (MRS) algorithm, is an image segmentation algorithm based on object thinking. The algorithm believes that images are composed of image objects with semantic information and their relationships. The algorithm idea is to adjust the scale parameters and perform bottom-up iterative segmentation from the pixel layer to form a multi-scale hierarchical network structure of image objects, such as Figure 2 shown.

[0083] The process of image segmentation using fractal network evolution algorithm is essentially a process of region growth, and the basis for region merging is the minimum increase in heterogeneity.

[0084] Furthermore, common features of buildings include spectral features, texture features, and geometric features. The roof materials of buildings in urban areas are generally asphalt, cement, metal, etc. The roof reflectivity is high, and the spectral characteristics in remote sensing images are characterized by high brightness. The image characteristics of buildings that are locally irregular but overall regular are remote sensing image texture features. Buildings and their shadows usually exist together. The brightness of buildings is higher than that of shadows, forming special texture features of local light and dark contrast in remote sensing images. As a typical man-made feature, buildings are usually a combination of different regular geometric bodies and are mainly polygonal, with very obvious geometric features.

[0085] On the basis of fully considering the spectrum, texture, shape and other characteristics of buildings in high-resolution remote sensing images, the present invention uses Morphological Building Index (MBI) and Morphological Shadow Index (MSI) to describe the building characteristics of high-resolution remote sensing images.

[0086] (1) Construction method of the Morphological Building Index (MBI);

[0087] In high-resolution remote sensing images, buildings have the characteristic of high local brightness due to shadows. Therefore, the morphological building index uses the maximum brightness of each pixel in each band as the brightness value of the image, that is,

[0088]

[0089] Where B(x, y) is the image brightness value of the pixel at (x, y) in the image, (x, y) represents the pixel position, and I k (x, y) represents the brightness value of the kth band of the pixel at the position (x, y) in the remote sensing image, k represents the number of bands of the remote sensing image, k∈[1, K], K is the total number of bands of the remote sensing image.

[0090] In order to extract the brighter spots in high-resolution remote sensing images, namely the possible building spots, white hat transformation is performed on the remote sensing images during the calculation of the morphological building index.

[0091]

[0092] Where WTH(d, s) is the result of white hat transformation, d and s represent the direction and scale of the structure element of white hat transformation, B is the image brightness, Represents the result of the morphological opening operation on the brightness image B using a linear structure element with direction d and size s.

[0093] By calculating the difference between the white-hat transformation results of adjacent scales, differential morphological profiles (DMP) are constructed to describe buildings of different scales in the same image.

[0094] DMP WTH (d,s)=WTH(d,s+Δs)-WTH(d,s) (3)

[0095] Where DMP WTH (d, s) is the differential morphological feature, d, s represent the direction and scale of the structure element of the white hat transformation, respectively, and Δs is the adjacent scale interval of the white hat transformation.

[0096] The morphological building index (MBI) is the weighted sum of differential morphological features.

[0097]

[0098] Where DMP WTH (d, s) is the differential morphological feature, d, s represent the direction and scale of the white hat transformation structure element, respectively, and D, S represent the total number of directions and the total number of scales of the white hat transformation structure element, respectively.

[0099] (2) Construction method of morphological shadow index (MSI);

[0100] The morphological shadow index uses the maximum brightness of each pixel in each band as the brightness value of the image, that is,

[0101]

[0102] Where B(x, y) is the image brightness value of the pixel at (x, y) in the image, (x, y) represents the pixel position, and I k (x, y) represents the brightness value of the kth band of the pixel at the position (x, y) in the remote sensing image, k represents the number of bands of the remote sensing image, k∈[1, K], K is the total number of bands of the remote sensing image.

[0103] By performing morphological black hat transformation on remote sensing images, the areas with lower brightness values ​​in remote sensing images, namely possible shadow areas, are extracted.

[0104]

[0105] Where BTH(d, s) is the result of black hat transformation, d and s represent the direction and scale of the structure element of white hat transformation, B is the image brightness, Represents the result of the morphological closing operation on the brightness image B using a linear structure element with direction d and size s.

[0106] By calculating the difference between the black hat transformation results of adjacent scales, differential morphological profiles (DMP) are constructed to describe the shadows of different scales in the same image.

[0107] DMP BTH (d,s)=BTH(d,s+Δs)-BTH(d,s) (7)

[0108] Where DMP BTH (d, s) is the differential morphological feature, d, s represent the direction and scale of the black hat transformation structure element respectively, and Δs is the adjacent scale interval of the black hat transformation.

[0109] The morphological shadow index (MSI) is the weighted sum of the differential morphological features.

[0110]

[0111] Where DMP BTH (d, s) is the differential morphological feature, d, s represent the direction and scale of the white hat transformation structure element, respectively, and D, S represent the total number of directions and the total number of scales of the white hat transformation structure element, respectively.

[0112] Furthermore, the method for correcting the projection difference is:

[0113] Acquire buildings whose image decoherence index is greater than a preset threshold in the early SAR remote sensing images and the later SAR remote sensing images;

[0114] Based on the buildings, projection difference correction is performed on the early morphological building index, the late morphological building index, the early morphological shadow index and the late morphological shadow index according to the elevation information.

[0115] The elevation information includes early elevation information and late elevation information;

[0116] Performing interference processing on the previous SAR remote sensing image pair to obtain the previous elevation information;

[0117] The late SAR remote sensing image pair is subjected to interference processing to obtain the late elevation information.

[0118] The identification information of building changes is obtained as: new buildings and reduced buildings.

[0119] The criteria for adding new buildings and reducing buildings are as follows:

[0120] (1) determining whether the image decoherence index is less than a coherence threshold (0.3-0.5); if so, proceeding to the next step;

[0121] (2) respectively comparing the early and late morphological shadow indices after the projection difference correction with the shadow index threshold. If the early morphological shadow index is greater than the shadow index threshold (1-2), the early shadow area is extracted. If the late morphological shadow index is greater than the shadow index threshold, the late shadow area is extracted.

[0122] (3) Double filtering method is used to identify and extract buildings for the early morphological building index and shadow index after projection difference correction, and the late morphological building index and shadow index; taking the early data as an example, if the morphological building index is greater than the first building index threshold (2.5-5) and its distance from the shadow area is less than the first distance index threshold (25-45); or the morphological building index is greater than the second building index threshold (1-2.5) and less than the first building index threshold (2.5-5), and its distance from the shadow area is less than the second distance index threshold (5-15), then it is identified as a building, otherwise it is identified as a non-building. Similar processing is performed on the late data;

[0123] (4) If a building is identified as a non-building in the previous period but is later identified as a building, it shall be classified as a newly added building; if a building is identified as a building in the previous period but is later identified as a non-building, it shall be classified as a reduced building;

[0124] (5) If a building is identified in the early stage and later stage, the elevation information of the early stage and the elevation information of the later stage are compared. If the elevation increases, it is classified as a newly added building; if the elevation decreases, it is classified as a reduced building.

[0125] Furthermore, the decision tree is an instance-based inductive learning algorithm. The algorithm recursively classifies the given data samples according to the feature space of the instance, forms a tree-structured classification rule, and constructs a mapping relationship between attribute values ​​and classification results. The decision tree structure includes leaf nodes, non-leaf nodes, and edges connecting the nodes. Leaf nodes correspond to classification results, non-leaf nodes represent the test of a certain feature, and each branch represents the test result of the feature in a certain value range.

[0126] The construction of a decision tree is to perform feature selection measurement and determine the topological structure between each feature, including two stages: construction and pruning. The construction of a decision tree is a top-down, divide-and-conquer process, which is essentially a greedy algorithm.

[0127] At present, the commonly used decision tree implementation algorithms include ID3, C4.5 and Classification and Regression Tree (CART). This project uses the CART algorithm for change decision analysis. As a non-parametric, nonlinear data mining and classification prediction algorithm, it uses the Gini coefficient and variance as measurement indicators to select the best test variables, build a binary decision tree model, and use post-pruning technology to achieve decision tree optimization.

[0128] Random forest is an integrated classifier based on decision trees. The random forest algorithm in the present invention selects subsamples from a given sample set according to the bagging strategy to construct different decision trees and form a forest. Each decision tree in the forest will independently classify the input data and obtain the classification result. The final classification result is obtained by voting on the results of all decision trees, such as Figure 3 shown.

[0129] Furthermore, the present invention takes the Tianjin Binhai Airport airspace protection zone as an example, and combines a remote sensing image detection method for building changes in the airport airspace protection zone proposed by the present invention to realize the identification information of building changes in the area.

[0130] The area selected in the embodiment of the present invention is located within the Tianjin Binhai Airport airspace protection zone, on the south side of Tianjin Binhai Airport, and within the administrative scope of Jinnan District. The area to be identified is near the location of the real estate development project "Jingrui Hanlin" and is a residential community under construction.

[0131] like Figure 4 As shown, the embodiment of the present invention uses WorldView2 satellite remote sensing images from the third quarter of 2017 and GF-2 satellite remote sensing images from 2019 for experiments, wherein the spatial resolution of WorldView2 is 0.5 meters and the spatial resolution of GF-3 is 0.8 meters.

[0132] The recognition results of the remote sensing image detection method for building changes in the airport clearance protection zone proposed by the present invention are as follows: Figure 5 From the third quarter of 2017 (September 2017) to the second quarter of 2019 (June 2019), a total of 125 new buildings were added in the experimental area, corresponding to the four newly built residential areas of Luneng Taishan No. 7 Area B, Yanlord Haiheyuan South Garden, Agile Yubinfu, and Jingrui Hanlin.

[0133] The above is only a preferred specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A remote sensing image detection method for building changes in an airport clear space protection zone, characterized in that: include: Acquire an early optical remote sensing image and an early SAR remote sensing image pair, as well as a late optical remote sensing image and a late SAR remote sensing image pair, wherein the early period and the late period are respectively the start time and the end time of the time period of the building changes in the airport airspace protection zone to be detected; Based on the early optical remote sensing image, an early morphological building index and an early morphological shadow index are obtained, and based on the late optical remote sensing image, a late morphological building index and a late morphological shadow index are obtained; Based on the previous SAR remote sensing image pair and the later SAR remote sensing image pair, obtaining elevation information, and based on the previous SAR remote sensing image and the later SAR remote sensing image, obtaining an image decoherence index; Based on the elevation information, projection difference correction is performed on the early morphological building index, the late morphological building index, the early morphological shadow index, and the late morphological shadow index; Constructing a decision forest; Inputting the corrected early morphological building index, early morphological shadow index, late morphological building index, late morphological shadow index and image decoherence index into the decision forest to obtain identification information of building changes; The steps of obtaining the early morphological building index and the late morphological building index include: Perform multi-scale segmentation on early and late optical remote sensing images respectively; Obtain the maximum value of the brightness of each pixel in each band in the early and late optical remote sensing images after segmentation as the image brightness value in the early and late optical remote sensing images after segmentation; Based on the image brightness value, white hat transformation is performed on the segmented early and late optical remote sensing images respectively; Based on the results of white hat transformation, the early morphological building index and the late morphological building index are obtained; The steps of obtaining the early morphological shadow index and the late morphological shadow index include: Perform multi-scale segmentation on early and late optical remote sensing images respectively; Obtain the maximum value of the brightness of each pixel in each band in the early and late optical remote sensing images after segmentation as the image brightness value in the early and late optical remote sensing images after segmentation; Based on the image brightness value, black hat transformation is performed on the segmented early and late optical remote sensing images respectively; Based on the results of black hat transformation, the early morphological shadow index and the late morphological shadow index are obtained.

2. The remote sensing image detection method for building changes in an airport clear space protection zone according to claim 1 is characterized in that: The elevation information includes early elevation information and late elevation information: Performing interference processing on the previous SAR remote sensing image pair to obtain the previous elevation information; The late SAR remote sensing image pair is subjected to interference processing to obtain the late elevation information.

3. The remote sensing image detection method for building changes in an airport clear space protection zone according to claim 1 is characterized in that: The identification information of the building change is: newly added buildings and reduced buildings.

4. The remote sensing image detection method for building changes in an airport clear space protection zone according to claim 3 is characterized in that: The judgment conditions for adding new buildings and reducing buildings are: (1) determining whether the image decoherence index is less than a coherence threshold, and if so, proceeding to the next step; (2) respectively comparing the early and late morphological shadow indices after the projection difference correction with the shadow index threshold; if the early morphological shadow index is greater than the shadow index threshold, the early shadow area is extracted; if the late morphological shadow index is greater than the shadow index threshold, the late shadow area is extracted; (3) Using double filtering method to identify and extract buildings, the early morphological building index and shadow index, and the late morphological building index and shadow index after projection difference correction are used respectively; (4) If a building is identified as a non-building in the early stage but as a building in the later stage, it shall be classified as a newly added building; If it is identified as a building in the previous period and as a non-building in the later period, it shall be classified as a reduced building; (5) If it is identified as a building in the early stage and a building in the later stage, a comparison is made based on the elevation information in the early stage and the elevation information in the later stage. If the elevation increases, it is classified as a newly added building; If the elevation decreases, it is classified as reducing buildings.

Images