CN117593664B - Waterproof surface extraction method and device integrating multi-mode remote sensing data - Google Patents
Waterproof surface extraction method and device integrating multi-mode remote sensing data Download PDFInfo
- Publication number
- CN117593664B CN117593664B CN202311690068.2A CN202311690068A CN117593664B CN 117593664 B CN117593664 B CN 117593664B CN 202311690068 A CN202311690068 A CN 202311690068A CN 117593664 B CN117593664 B CN 117593664B
- Authority
- CN
- China
- Prior art keywords
- remote sensing
- features
- characteristic
- building height
- index
- 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
- 238000000605 extraction Methods 0.000 title claims description 37
- 238000000034 method Methods 0.000 claims abstract description 66
- 230000010287 polarization Effects 0.000 claims abstract description 48
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims abstract description 35
- 238000001228 spectrum Methods 0.000 claims abstract description 31
- 238000000926 separation method Methods 0.000 claims description 34
- 239000002689 soil Substances 0.000 claims description 27
- 238000012549 training Methods 0.000 claims description 20
- 230000003595 spectral effect Effects 0.000 claims description 12
- 238000007637 random forest analysis Methods 0.000 claims description 11
- 238000000354 decomposition reaction Methods 0.000 claims description 6
- 230000004927 fusion Effects 0.000 claims description 3
- 230000000717 retained effect Effects 0.000 claims description 3
- 238000012706 support-vector machine Methods 0.000 claims description 3
- 238000002310 reflectometry Methods 0.000 description 12
- 239000000463 material Substances 0.000 description 11
- 230000000694 effects Effects 0.000 description 8
- 230000008569 process Effects 0.000 description 8
- 239000004567 concrete Substances 0.000 description 6
- 230000006870 function Effects 0.000 description 6
- 230000003287 optical effect Effects 0.000 description 5
- 238000012360 testing method Methods 0.000 description 5
- 239000010426 asphalt Substances 0.000 description 4
- 230000008901 benefit Effects 0.000 description 4
- 239000002184 metal Substances 0.000 description 4
- 238000002156 mixing Methods 0.000 description 4
- 238000007781 pre-processing Methods 0.000 description 4
- 238000011160 research Methods 0.000 description 4
- 230000007547 defect Effects 0.000 description 3
- 239000008186 active pharmaceutical agent Substances 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000013145 classification model Methods 0.000 description 2
- 239000000470 constituent Substances 0.000 description 2
- 238000012937 correction Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 239000002352 surface water Substances 0.000 description 2
- 101100532679 Caenorhabditis elegans scc-1 gene Proteins 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 235000019646 color tone Nutrition 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 239000011365 complex material Substances 0.000 description 1
- 238000013075 data extraction Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000007477 logistic regression Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 239000007769 metal material Substances 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000002203 pretreatment Methods 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 230000002441 reversible effect Effects 0.000 description 1
- 239000011435 rock Substances 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 238000012876 topography Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- -1 vegetation Substances 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/13—Satellite images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
- G06V10/806—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of extracted features
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A30/00—Adapting or protecting infrastructure or their operation
- Y02A30/60—Planning or developing urban green infrastructure
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Databases & Information Systems (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Astronomy & Astrophysics (AREA)
- Remote Sensing (AREA)
- Image Processing (AREA)
Abstract
The invention provides a method and a device for extracting a watertight surface fused with multi-mode remote sensing data, wherein the method comprises the following steps: acquiring remote sensing image data of a region to be analyzed; extracting building height characteristics, impermeable surface remote sensing index characteristics, spectrum characteristics and polarization characteristics from the remote sensing image data; and merging the extracted building height characteristics, the remote sensing index characteristics of the impervious surface, the spectrum characteristics and the polarization characteristics, and identifying the impervious surface of the area to be analyzed. The invention obviously reduces confusion of the impervious surface, bare land and the blocked short building and water body by fusing the multidimensional features, and the fused multi-feature has positive contribution to extracting impervious surface information of the complex urban area.
Description
Technical Field
The invention relates to the technical field of water resource research, in particular to a watertight surface extraction method and device integrating multi-mode remote sensing data.
Background
One significant change in urbanization is the abrupt expansion of the water impermeable surface. The impermeable surface (impervious surface area, ISA) refers to a natural or artificial surface type where surface water cannot penetrate into soil, and the natural impermeable surface mainly refers to impermeable rock, and the artificial impermeable surface is an artificial target such as a building roof, a road, a parking lot, and the like. The impermeable surface is an important index for measuring the degree of urban development and evaluating the quality of urban ecological environment. There are related studies that demonstrate that the spatial distribution of urban water-impermeable surfaces has a high degree of consistency with surface temperature. The impermeable material absorbs heat quickly, has small specific heat capacity, easily forms a high heat accumulator in cities to form urban heat island effect, and the rapid promotion of urban land form greatly reduces natural landscapes mainly comprising green vegetation, so that the urban heat island effect is aggravated to a certain extent. The rapid increase of the impermeable surface prevents the surface water from leaking down, so that the urban surface runoff is increased, urban flood disasters are easily caused when the urban surface runoff is in heavy rainy days, and the drainage of the urban government and the flood control pressure of the urban are increased. Meanwhile, as the urban watertight surface expands on a large scale to increase runoff, a large amount of non-point source pollution enters a river, and the non-point source pollution diffuses to the river and the lake along the surface runoff to deteriorate water quality, thereby threatening the health of human bodies.
In summary, the impervious surface information reflects not only land utilization and land-cover change (LUCC) caused by urbanization, but also changes in urban ecological environment caused by city LUCC. Therefore, how to extract the water-impermeable surface information timely and accurately is of great significance for monitoring the urban expansion degree and knowing the influence on the urban ecological environment.
At present, three main methods for extracting the water impermeable surface by using remote sensing technology at home and abroad are as follows:
The first type is a method for extracting a watertight surface based on optical remote sensing data, which generally refers to a method for extracting a watertight surface by using optical remote sensing data such as Landsat-5/7/8/9, sentinel-2, quickBird, plae iades and Wordview/3, etc., and common methods comprise a spectrum mixing decomposition method, a regression method, a spectrum index method, an object-oriented method, a deep learning method, etc., but the method based on the optical remote sensing data is difficult to solve the problems of 'homography and foreign matter homography'.
The second type is a method based on radar data extraction, namely, the method based on radar data is used for extracting spatial features such as textures, shapes, neighborhoods and the like to improve the recognition precision of the impermeable surface, the method based on radar data makes up the defect of optical data to a certain extent, but the precision of a model for extracting the impermeable surface by using only radar data is often limited, and a great challenge still exists in extracting the impermeable surface in a region with complex topography.
The third type is a waterproof surface extraction method based on multi-source remote sensing data, the multi-source remote sensing data comprise data of different sensors, different resolutions and different phases, compared with a single data source, the multi-source remote sensing data are more abundant in information, the defect of insufficient characteristics of the single data source is overcome, the precision is obviously improved, but the confusion between the waterproof surface and bare soil is always a main problem of identifying the waterproof surface based on the remote sensing data due to the influence of the geometric shape, the form, the constituent materials and other factors of the waterproof surface in different urban areas, the problems of poor universality, low extraction precision and the like of the method for removing the bare soil in many researches are solved, a large number of shadows exist in the high-resolution images, the radiation information of shadow areas is lost, the subsequent remote sensing image interpretation process is more complicated and difficult, and the classification result precision is lower.
It can be seen that the existing method still has certain drawbacks, at least including:
1) The bare soil is more in variety, the spectrum complexity is higher, and the influence of various types of bare soil on the extraction of the impermeable surface cannot be removed by the existing method.
2) Shadow areas caused by the shielding of low buildings by medium-high buildings are very similar to spectrum and polarization characteristics of water in images, and the separability is poor, so that the difficulty in extracting the shielded low building areas is increased.
Therefore, a water-impermeable surface extraction technique is needed to solve the above-mentioned problems.
Disclosure of Invention
One of the objectives of the present application is to provide a method and a device for extracting a watertight surface fused with multi-mode remote sensing data, so as to solve the problems set forth in the background art.
According to an embodiment of an aspect of the present invention, there is provided a method for extracting a watertight surface fused with multimodal remote sensing data, including:
acquiring remote sensing image data of a region to be analyzed;
Extracting building height characteristics, impermeable surface remote sensing index characteristics, spectrum characteristics and polarization characteristics from the remote sensing image data;
and merging the extracted building height characteristics, the remote sensing index characteristics of the impervious surface, the spectrum characteristics and the polarization characteristics, and identifying the impervious surface of the area to be analyzed.
Preferably, before the extracting the building height feature in the remote sensing image data, the method further includes:
Extracting a plurality of features identifying building height information;
acquiring contribution degrees of the plurality of features according to a preset method;
Reserving the characteristic of which the contribution degree reaches a specified value as the characteristic for identifying the building height information;
And training a building height feature extraction model by using the reserved features.
Preferably, the obtaining the contribution degrees of the plurality of features according to a preset method includes:
and adopting a random forest for the subsamples of the randomly extracted part, and reserving the characteristic that the contribution degree reaches the preset threshold importance as the characteristic for identifying the building height information.
Preferably, the step of training the building height feature extraction model using the retained features comprises:
And training the building height feature extraction model based on a support vector machine regression model SVR by using the reserved features.
Preferably, before extracting the remote sensing index feature of the impervious surface, the method further comprises:
Classifying different ground objects according to preset rules;
calculating the separation degree of the ground objects of different classifications on different feature dimensions;
And selecting a characteristic training impervious remote sensing index characteristic extraction model with the separation degree larger than 1 to extract the impervious remote sensing index characteristic by using the impervious remote sensing index characteristic extraction model.
Preferably, the step of calculating the degree of separation of the ground objects of different classifications in different feature dimensions includes:
The degree of separation is calculated using the formula
，
，
Wherein B represents the Papanicolaou distance between the samples of each class on a certain characteristic,AndThe average value of two ground objects is shown,AndThe standard deviation of two ground objects is represented, J represents the separation degree, ln is a logarithmic function, and e is an exponential function.
Preferably, the step of calculating the degree of separation of the ground objects of different classifications in different feature dimensions includes:
selecting two classified ground objects as a group, wherein the degree of distinction of the two ground objects in the same group in a single characteristic dimension is lower than a set value;
And calculating the separation degree of each group of ground objects in different characteristic dimensions.
Preferably, the feature of the degree of separation greater than 1 includes:
Normalized vegetation index features, building index features, bare soil index features, normalized water index features, polarization coherence coefficient band 1 features VV, polarization coherence coefficient band 2 features VH, and polarization decomposition-based anisotropy features.
Preferably, the step of merging the extracted building height features, the remote sensing index features of the impervious surface, the spectrum features and the polarization features to identify the impervious surface of the area to be analyzed includes:
Inputting the building height characteristics, the impermeable surface remote sensing index characteristics, the spectrum characteristics and the polarization characteristics into a trained random forest model;
and obtaining the watertight surface identification result of the area to be analyzed, which is output by the random forest model.
Another embodiment of the present application provides a device for extracting a watertight surface fused with multi-mode remote sensing data, including:
The first unit is used for acquiring remote sensing image data of the area to be analyzed;
The second unit is used for extracting building height characteristics, impermeable surface remote sensing index characteristics, spectrum characteristics and polarization characteristics from the remote sensing image data;
And the third unit is used for merging the extracted building height characteristics, the remote sensing index characteristics of the impervious surface, the spectrum characteristics and the polarization characteristics and identifying the impervious surface of the area to be analyzed.
The invention has the beneficial effects that:
The method and the device for extracting the impermeable surface fused with the multi-mode remote sensing data can be used for extracting the impermeable surface of the complex urban area, and the impermeable surface of the area to be analyzed is identified by extracting the building height characteristic, the impermeable surface remote sensing index characteristic, the spectrum characteristic and the polarization characteristic in the remote sensing image data and fusing the extracted building height characteristic, the impermeable surface remote sensing index characteristic, the spectrum characteristic and the polarization characteristic. The invention obviously reduces confusion of the impervious surface, bare land and the blocked short building and water body by fusing the multidimensional features, and the fused multi-feature has positive contribution to extracting impervious surface information of the complex urban area.
Those of ordinary skill in the art will realize that while the following detailed description proceeds with reference to the illustrative embodiments, the accompanying drawings, the application is not limited to these embodiments. Rather, the scope of the application is broad and is intended to be defined as only set forth in the accompanying claims.
Drawings
Other features, objects and advantages of the present application will become more apparent upon reading of the detailed description of non-limiting embodiments, made with reference to the accompanying drawings in which:
fig. 1 is a flowchart of a method for extracting a watertight surface fused with multi-mode remote sensing data according to an embodiment of the present invention.
FIG. 2 is a graph showing the comparison of calculated separation degrees of different feature combinations according to an embodiment of the present invention;
FIG. 3 is a graph of comparison of results of extraction of impervious surfaces according to various feature combination schemes in accordance with an embodiment of the invention;
FIG. 4 is a comparison graph of results of identifying different land types for different feature combination schemes in accordance with an embodiment of the invention;
FIG. 5 is a graph of extracted impervious surface accuracy versus different feature combination schemes in accordance with an embodiment of the present invention;
fig. 6 is a schematic structural diagram of a watertight surface extraction device integrating multi-mode remote sensing data according to an embodiment of the present invention.
Those of ordinary skill in the art will realize that while the following detailed description proceeds with reference to the illustrative embodiments, the accompanying drawings, the application is not limited to these embodiments. Rather, the scope of the application is broad and is intended to be defined as only set forth in the accompanying claims.
Description of the embodiments
The following description of the embodiments of the present invention will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are some, but not all embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The terms first, second and the like in the description and in the claims, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged, as appropriate, such that embodiments of the present invention may be implemented in sequences other than those illustrated or described herein, and that the objects identified by "first," "second," etc. are generally of a type, and are not limited to the number of objects, such as the first object may be one or more. Furthermore, the term "and/or" as used in the specification and claims to describe an association of associated objects means that there may be three relationships, e.g., a and/or B, may mean: a exists alone, A and B exist together, and B exists alone. The character "/" generally indicates that the context-dependent object is an "or" relationship. The term "plurality" in embodiments of the present invention means two or more, and other adjectives are similar.
Before discussing exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although a flowchart depicts operations as a sequential process, many of the operations can be performed in parallel, concurrently, or at the same time. Furthermore, the order of the operations may be rearranged. The process may be terminated when its operations are completed, but may have additional steps not included in the figures. The processes may correspond to methods, functions, procedures, subroutines, and the like.
The methods discussed below (some of which are illustrated by flowcharts) may be implemented by hardware, software, firmware, middleware, microcode, hardware description languages, or any combination thereof. When implemented in software, firmware, middleware or microcode, the program code or code segments to perform the necessary tasks may be stored in a machine or computer readable medium such as a storage medium. The processor(s) may perform the necessary tasks.
The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and/or "comprising," when used herein, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.
It should also be noted that, in some alternative implementations, the functions/acts noted may occur out of the order noted in the figures. For example, two figures shown in succession may in fact be executed substantially concurrently or the figures may sometimes be executed in the reverse order, depending upon the functionality/acts involved.
Fig. 1 is a flowchart showing the operation of a method for extracting a watertight surface fused with multi-mode remote sensing data according to the present invention, the method includes the following steps:
S10, acquiring remote sensing image data of a region to be analyzed;
s11, extracting building height characteristics, impermeable surface remote sensing index characteristics, spectrum characteristics and polarization characteristics from the remote sensing image data;
S12, merging the extracted building height features, the remote sensing index features of the impervious surface, the spectrum features and the polarization features, and identifying the impervious surface of the area to be analyzed.
In order to understand the above steps accurately, the following describes each step in further detail.
The remote sensing image data in step S10 may be remote sensing image data of the area to be analyzed acquired through a satellite. The area to be analyzed is the area of the water-impermeable surface to be extracted.
As described in step S11, the present invention combines building height characteristics, impervious remote sensing index characteristics, spectral characteristics and polarization characteristics in the process of extracting impervious surfaces. Wherein the spectral features include: band characteristics of B2-B12 (which refer to different bands of Sentinel-2 spectrum); the polarization features include two polarization bands VV and VH of Sentinel-1.
The invention constructs two new characteristic indexes for extracting the impervious surface, including the building height characteristic and the impervious surface remote sensing index characteristic, and the extraction methods of the two characteristics adopt the method designed by the invention, and the specific steps are as follows.
The building height features are extracted by the method designed by the invention, and the method comprises the following steps:
S1101, collecting a data set and preprocessing to obtain a training set and a testing set;
The invention is not particularly limited to data set collection, and in one embodiment the data sets collected include, but are not limited to: the Sentinel-1 dataset, the corresponding pre-processing includes: thermal noise removal, radiometric calibration, range-doppler geometry correction, decibelization, and the like; the Sentinel-2 dataset, the corresponding pre-processing includes: mosaic cutting, radiometric calibration, atmospheric correction, cloud removal pretreatment and the like; NPP-VIIRS (manufactured by the national marine atmospheric administration (NOAA) and the national aerospace agency (NASA)) and the corresponding pretreatments include: projection transformation of the data, annual average data synthesis, negative value removal, background value elimination, in our study, these negative values were reset to 0, resulting in a cloud-free average radiance value product with outliers (including fires and other brief lights) removed. Building height data sets, e.g., ATL03 and ATL08 products of ICESat-2 data 8, the corresponding pre-processing includes: format conversion and elevation conversion, then subtracting two parameters by utilizing absolute elevation and absolute photon height to obtain ground feature height, and screening out sample points falling in the outline range of the building to be used as building height sample points; building contour data obtained from various mapping software may also be included, including information on floors and the like.
Wherein 90% of the samples can be used as a training set and 10% of the samples can be used as a test set.
S1102, training a building height feature extraction model by using a training set;
It will be appreciated that there are a number of characteristic parameters for assessing building height, including: the method provided by the embodiment of the invention comprises the following specific steps of:
firstly, extracting a plurality of features for identifying building height information; comprising the following steps: a plurality of features in spectral/polarization, temporal and spatial dimensions;
Then, the contribution degrees of the plurality of features are obtained according to a preset method, and the features with the contribution degrees reaching a specified value are reserved as features for identifying building height information; for example, a Random Forest (RFR) is employed for sub-samples (500 decision trees in total, 1% of features per segmentation test) of randomly extracted portions (e.g., 10%), and features with a contribution degree exceeding a predetermined threshold (e.g., 1%o) importance are retained as features identifying building height information.
And finally, training a building height feature extraction model by using the reserved features. And training the building height feature extraction model based on a support vector machine regression model (SVR) when training the building height feature extraction model by using the reserved features.
And S1103, testing the building height feature extraction model by using a test set.
And S1104, inputting the building height feature extraction model based on the feature parameters extracted from the remote sensing image data, so as to obtain the building height of the area to be analyzed. The extracted characteristic parameters reach the characteristic that the contribution degree reserved when the building height characteristic extraction model is trained before reaches the specified value.
The method for extracting the remote sensing index features of the impervious surfaces is designed by the invention, and a model for extracting the remote sensing index features of the impervious surfaces is required to be trained, and the specific data set collection and training process is similar to the building height feature extraction model, and is different in the feature selection process, wherein one embodiment for selecting the model features for extracting the remote sensing index features of the impervious surfaces by training is as follows:
firstly, classifying different ground objects according to preset rules; for example, the following seven types of materials can be classified according to the composition of different features, including: metal materials, concrete, asphalt, vegetation, water, high-reflectivity bare soil and low-reflectivity bare soil.
Then, calculating the separation degree of the ground objects with different classifications on different feature dimensions;
Wherein the features include: spectral characteristics and radar index characteristics. The spectral features mainly include: normalized vegetation index (NDVI), building index (PII), bare soil index (BAI), and normalized water index (mNDWI); the radar characteristic index includes: based on the anisotropic characteristic of polarization decomposition and two polarization coherence coefficient bands (VV and VH), the characteristic coefficient is obtained by logistic regression analysis.
One embodiment of the invention uses SEaTH algorithm to calculate the degree of separation of the features, for example, using the following formula
，
，
Wherein B represents the Papanicolaou distance between the samples of each class on a certain characteristic,AndThe average value of two ground objects is shown,AndThe standard deviation of two ground objects is represented, J represents the separation degree, ln is a logarithmic function, and e is an exponential function.
It can be understood that the feature distinction degree of different constituent materials is different, in order to better extract the impermeable surface, the invention selects two classified features as one group, the distinction degree of the two features of the same group is lower than a set value on a single characteristic dimension (that is, the single characteristic dimension of the two classified features of the same group is easy to be confused, such as on spectral characteristics or polarization characteristics, and is not easy to be distinguished), and the separation degree of each group of features on different characteristic dimensions is calculated, so that the two features with low distinction degree can be better identified, and the impermeable surface can be extracted more accurately. As shown in fig. 2, a contrast plot of separation was calculated using different features for different feature combinations. The ordinate in fig. 2 shows different combinations of features, the abscissa shows different features, I/V represents the water impermeable surface/vegetation, I/W represents the water impermeable surface/body, C/DS represents the concrete/low reflectivity bare soil, P/DS represents asphalt/low reflectivity bare soil, and M/HS represents metal/high reflectivity bare soil. NDVI in abscissa represents normalized vegetation index, PII represents building index, BAI represents bare soil index, mNDWI represents normalized water index, coh1 represents polarization coherence coefficient band 1, coh2 represents polarization coherence coefficient band 2, and a represents anisotropic feature based on polarization decomposition. The separation degree is 0-2, and the greater the separation degree, the higher the separation property.
As can be seen from fig. 2, the feature separation degree of different features is not uniform in different feature sets. NDVI can better distinguish between impervious surfaces and vegetation, water, and concrete, and in this characteristic dimension the degree of distinction between metal and high reflectivity bare soil is low. The impervious surface and bare soil are relatively highly separated on the BAI features, but less distinguishable from bright bare soil and metal. mNDWI can separate the body of water from the impervious surface, the concrete and the bare soil, the bitumen and the bare soil. PII has higher separation degree of the impermeable surface and bare soil than the former three, but the overall separation effect is poor. In Coh, coh, and a feature dimensions, the impermeable surface of metal, asphalt, and concrete materials can be distinguished from bare soil. In general, the spectrum features have lower discrimination between bare soil and impermeable water, higher discrimination between vegetation, water and impermeable water, and higher discrimination between impermeable water and bare soil of different materials, so that the fusion spectrum features and radar index features form complementary advantages.
And then, selecting a characteristic training impervious remote sensing index characteristic extraction model with the separation degree larger than 1 to extract the impervious remote sensing index characteristic by using the impervious remote sensing index characteristic extraction model.
It can be seen that the features with a degree of separation greater than 1 include: normalized vegetation index features, building index features, bare soil index features, normalized water index features, polarization coherence coefficient band 1 features VV, polarization coherence coefficient band 2 features VH, and polarization decomposition-based anisotropy features.
The step of merging the extracted building height feature, the remote sensing index feature of the impervious surface, the spectrum feature and the polarization feature in the step S12, and the step of identifying the impervious surface of the area to be analyzed may include: inputting the building height characteristics, the impermeable surface remote sensing index characteristics, the spectrum characteristics and the polarization characteristics into a trained random forest model; and obtaining the watertight surface identification result of the area to be analyzed, which is output by the random forest model.
That is, the embodiment of the invention trains the model for extracting the impervious surface based on the random forest algorithm by utilizing the building height characteristic, the impervious surface remote sensing index characteristic, the spectrum characteristic and the polarization characteristic, and in order to verify the effect of extracting the impervious surface by applying the model, the invention designs classification experiments of six different characteristic combinations for six urban areas respectively, inputs different characteristics into a random classifier, and the specific information of the six schemes is shown in the table
NDBI in the above table represents building normalization index, LISI represents large scale impervious surface index feature, build height represents building height feature, VISWac represents impervious surface remote sensing index feature. Scheme VI is a corresponding scheme of the invention. The impervious surface information is extracted according to different characteristic combination schemes in the table, and as a result, for example, as shown in fig. 3, (a), (b), (c), (d), (e) and (f) in fig. 3 respectively represent different cities. Comparing six experimental schemes, it can be seen that in different urban areas, the effect of extracting the impermeable surface by using single data is poor, and by comparing the scheme VI corresponding to the scheme VI, the scheme of the invention can be seen to have the best effect, and by fusing multidimensional features, confusion between the impermeable surface and bare land and between the blocked short building and water body is obviously reduced, so that the invention fully proves that the fused multi-feature has positive contribution to extracting impermeable surface information of the complex urban area.
As shown in FIG. 4, the recognition results of different land types are compared, in FIG. 4, (a), (b), (c) and (d) respectively represent a central business area, a residential area, an industrial area and a highway, and different rows represent different cities. As can be seen from a comparison of FIG. 4, the central business zone water-impermeable surface includes a low-reflectivity water-impermeable surface and a high-reflectivity water-impermeable surface, and the materials are various and complex. The problem that a large number of building shadows are recognized as water bodies only by utilizing spectral information in dense areas of tall buildings is solved, because the low-reflectivity water-impermeable surface material presents darker color tones on remote sensing images, and a part of building shadows are caused by shielding of short buildings, so that the spectral characteristics and texture characteristics of the buildings in the area are very similar to those of the water bodies, and the buildings are easily mistakenly divided into the water bodies, for example, scheme I and scheme II all have different degrees of mistaken division in six research areas; the advantage of radar remote sensing data can be utilized to make up for the defect that the attribution of a shadow area in an optical remote sensing image is difficult to distinguish, the problem of underestimation of a watertight surface caused by building shadows is effectively avoided, a shielded building can be separated from a water body on the polarization characteristic, the problem that the building shadows are hardly recognized as the water body by extracting the watertight surface by utilizing the polarization characteristic in the scheme III, but the problem of mixing with other places is caused, and the underestimation of the watertight surface is often caused; the spectral features and the polarization features are fused, so that the separation degree of water, bare soil and the impermeable surface can be increased to a certain extent, the impermeable surfaces of different materials can be identified more easily, the underestimation of the impermeable surfaces is reduced, but a small amount of building shadows still exist to identify the problems as water, and the spectral information in the dense area of the impermeable surfaces is proved to play a larger role, for example, a scheme IV; the sensitivity of the fusion spectrum and night light features to the impervious surface of the high-reflectivity material is higher, but the recognition accuracy of the impervious surface mainly made of concrete and other materials is lower, and a situation that a small amount of water is mistakenly divided into water bodies exists, for example, a scheme V; the scheme VI has a good extraction effect on urban areas with complex materials, reduces the influence of building shadows to a great extent, and simultaneously keeps the building outline details of images, mainly because the blocked low buildings have height information, and the water body has no height information, so that the mixing of the two is reduced, and meanwhile, the false separation of high-reflectivity buildings and high-bright bare soil is reduced.
The method utilizes a confusion matrix method to carry out accuracy verification on the extraction results of the classification models of different feature combinations. And evaluating the respective classification performance of the six models by comparing and analyzing four indexes of the overall precision, kappa coefficient, misclassification rate and omission rate of the six classification models, so as to determine a characteristic combination model suitable for extracting the water impermeable surface information of the complex urban area. Each model classifies Overall Accuracy (OA), kappa coefficient, misclassification rate (CE) and misclassification rate (OE) as shown in FIG. 5, with different rows representing different cities. As can be seen from FIG. 5, the extraction effect of the schemes I and III is poor, the overall accuracy is about 70.00%, the overall accuracy of the scheme II is above 85.00% in each urban area, but the error division and leakage division rate is above 10.00%; the overall precision of the scheme III is about 70.00%; compared with the prior scheme, the scheme IV has the advantages that the precision is greatly improved, the average overall precision is 92.72 percent, and the average error rate and the omission rate are within 10 percent; the overall precision of the scheme V is above 90.00%, and the kappa coefficient is above 0.8500; the overall precision of the scheme VI in six research areas is over 95.00 percent, and compared with the scheme V, the method improves the overall precision by about 3.00 percent, reduces the error rate and the leakage rate by about 2.50 percent, and achieves the highest precision in six cities.
The embodiment of the invention also provides a device for extracting the impervious surface fused with the multi-mode remote sensing data, as shown in fig. 6, which is a schematic diagram of the structure of the device, and comprises:
A first unit 60, configured to acquire remote sensing image data of an area to be analyzed;
A second unit 61, configured to extract building height features, impervious surface remote sensing index features, spectrum features and polarization features from the remote sensing image data;
a third unit 62 for merging the extracted building height features, the remote sensing index features of the impervious surface, the spectral features and the polarization features to identify the impervious surface of the area to be analyzed.
With respect to the apparatus in the above-described embodiments, the specific manner in which the respective units perform the operations has been described in detail in the embodiments regarding the method.
In summary, according to the method and the device for extracting the impervious surface fused with the multi-mode remote sensing data, the building height feature, the impervious surface remote sensing index feature, the spectrum feature and the polarization feature in the remote sensing image data are extracted; and merging the extracted building height characteristics, the remote sensing index characteristics of the impervious surface, the spectrum characteristics and the polarization characteristics, and identifying the impervious surface of the area to be analyzed. The invention obviously reduces confusion of the impervious surface, bare land and the blocked short building and water body by fusing the multidimensional features, and fully proves that the fused multi-feature has positive contribution to extracting the impervious surface information of the complex urban area. In addition, the invention reduces the influence of building shadows to a great extent and simultaneously maintains the building outline details of the image, mainly because the blocked short building has height information, and the water body has no height information, thereby reducing the mixing of the two, and simultaneously reducing the miscibility of the high-reflectivity building and the high-brightness bare soil.
It is to be understood that the invention is not limited to the precise arrangements and instrumentalities shown in the drawings, which have been described above, and that various modifications and changes may be effected without departing from the scope thereof. The scope of the invention is limited only by the appended claims.
The foregoing description of the preferred embodiments of the invention is not intended to limit the invention to the precise form disclosed, and any such modifications, equivalents, and alternatives falling within the spirit and scope of the invention are intended to be included within the scope of the invention.
Claims (8)
1. The method for extracting the impervious surface fused with the multi-mode remote sensing data is characterized by comprising the following steps of:
acquiring remote sensing image data of a region to be analyzed;
Extracting building height characteristics, impermeable surface remote sensing index characteristics, spectrum characteristics and polarization characteristics from the remote sensing image data;
Before extracting the building height characteristics in the remote sensing image data, the method further comprises the following steps:
Extracting a plurality of features identifying building height information;
acquiring contribution degrees of the plurality of features according to a preset method;
Reserving the characteristic of which the contribution degree reaches a specified value as the characteristic for identifying the building height information;
Training a building height feature extraction model by using the reserved features;
before extracting the remote sensing index features of the impervious surface, the method further comprises the following steps:
Classifying different ground objects according to preset rules;
calculating the separation degree of the ground objects of different classifications on different feature dimensions;
selecting a characteristic training impervious remote sensing index characteristic extraction model with the separation degree larger than 1, and extracting the impervious remote sensing index characteristic by using the impervious remote sensing index characteristic extraction model;
and merging the extracted building height characteristics, the remote sensing index characteristics of the impervious surface, the spectrum characteristics and the polarization characteristics, and identifying the impervious surface of the area to be analyzed.
2. The method of claim 1, wherein the obtaining the contribution of the plurality of features according to a preset method comprises:
and adopting a random forest for the subsamples of the randomly extracted part, and reserving the characteristic that the contribution degree reaches the preset threshold importance as the characteristic for identifying the building height information.
3. The method of claim 1, wherein training the building height feature extraction model using the retained features comprises:
And training the building height feature extraction model based on a support vector machine regression model SVR by using the reserved features.
4. The method of claim 1, wherein the step of calculating the degree of separation of the differently classified features in different feature dimensions comprises:
The degree of separation is calculated using the formula
J＝2(1-e-B)
Wherein B represents the Pasteur distance between each class of samples on a certain characteristic, m 1 and m 1 represent the average value of two features, sigma 1 and sigma 2 represent the standard deviation of the two features, J represents the degree of separation, ln is a logarithmic function, and e is an exponential function.
5. The method of claim 1, wherein the step of calculating the degree of separation of the differently classified features in different feature dimensions comprises:
selecting two classified ground objects as a group, wherein the degree of distinction of the two ground objects in the same group in a single characteristic dimension is lower than a set value;
And calculating the separation degree of each group of ground objects in different characteristic dimensions.
6. The method of claim 1, wherein the feature of a degree of separation greater than 1 comprises:
Normalized vegetation index features, building index features, bare soil index features, normalized water index features, polarization coherence coefficient band 1 features VV, polarization coherence coefficient band 2 features VH, and polarization decomposition-based anisotropy features.
7. The method of claim 1, wherein the step of merging the extracted building height features, the water-impermeable surface remote sensing index features, the spectral features, and the polarization features to identify the water-impermeable surface of the area to be analyzed comprises:
Inputting the building height characteristics, the impermeable surface remote sensing index characteristics, the spectrum characteristics and the polarization characteristics into a trained random forest model;
and obtaining the watertight surface identification result of the area to be analyzed, which is output by the random forest model.
8. The utility model provides a waterproof face extraction element of fusion multimode remote sensing data which characterized in that includes:
The first unit is used for acquiring remote sensing image data of the area to be analyzed;
The second unit is used for extracting building height characteristics, impermeable surface remote sensing index characteristics, spectrum characteristics and polarization characteristics from the remote sensing image data;
Before extracting the building height characteristics in the remote sensing image data, the method further comprises the following steps:
Extracting a plurality of features identifying building height information;
acquiring contribution degrees of the plurality of features according to a preset method;
Reserving the characteristic of which the contribution degree reaches a specified value as the characteristic for identifying the building height information;
Training a building height feature extraction model by using the reserved features;
before extracting the remote sensing index characteristics of the impermeable surface, the method further comprises the following steps:
Classifying different ground objects according to preset rules;
calculating the separation degree of the ground objects of different classifications on different feature dimensions;
selecting a characteristic training impervious remote sensing index characteristic extraction model with the separation degree larger than 1, and extracting the impervious remote sensing index characteristic by using the impervious remote sensing index characteristic extraction model;
And the third unit is used for merging the extracted building height characteristics, the remote sensing index characteristics of the impervious surface, the spectrum characteristics and the polarization characteristics and identifying the impervious surface of the area to be analyzed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311690068.2A CN117593664B (en) | 2023-12-11 | 2023-12-11 | Waterproof surface extraction method and device integrating multi-mode remote sensing data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311690068.2A CN117593664B (en) | 2023-12-11 | 2023-12-11 | Waterproof surface extraction method and device integrating multi-mode remote sensing data |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117593664A CN117593664A (en) | 2024-02-23 |
| CN117593664B true CN117593664B (en) | 2024-07-05 |
Family
ID=89915096
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311690068.2A Active CN117593664B (en) | 2023-12-11 | 2023-12-11 | Waterproof surface extraction method and device integrating multi-mode remote sensing data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117593664B (en) |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111122449A (en) * | 2019-12-31 | 2020-05-08 | 北京师范大学 | Urban impervious surface remote sensing extraction method and system |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107609526A (en) * | 2017-09-21 | 2018-01-19 | 吉林大学 | Rule-based fine dimension city impervious surface rapid extracting method |
| CN109374564A (en) * | 2018-08-20 | 2019-02-22 | 广州地理研究所 | A kind of multi- source Remote Sensing Data data city impervious surface extracting method |
| CN109934770A (en) * | 2019-01-21 | 2019-06-25 | 广州地理研究所 | City impervious surface extracting method based on high resolution ratio satellite remote-sensing image |
| CN113705523B (en) * | 2021-09-06 | 2023-07-14 | 青岛星科瑞升信息科技有限公司 | Layered urban impervious surface extraction method based on optical and dual-polarized SAR fusion |
-
2023
- 2023-12-11 CN CN202311690068.2A patent/CN117593664B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111122449A (en) * | 2019-12-31 | 2020-05-08 | 北京师范大学 | Urban impervious surface remote sensing extraction method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117593664A (en) | 2024-02-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Robson et al. | Automated detection of rock glaciers using deep learning and object-based image analysis | |
| Hu et al. | An automatic approach for land-change detection and land updates based on integrated NDVI timing analysis and the CVAPS method with GEE support | |
| Wei et al. | Cloud detection for Landsat imagery by combining the random forest and superpixels extracted via energy-driven sampling segmentation approaches | |
| Sun et al. | Combinational build-up index (CBI) for effective impervious surface mapping in urban areas | |
| Shahi et al. | A novel spectral index to automatically extract road networks from WorldView-2 satellite imagery | |
| Dibs et al. | Fusion Landsat-8 thermal TIRS and OLI datasets for superior monitoring and change detection using remote sensing | |
| CN108830844B (en) | Facility vegetable extraction method based on multi-temporal high-resolution remote sensing image | |
| Worden et al. | Surface water detection in the Caucasus | |
| CN112989692A (en) | Lake eutrophication inversion method based on remote sensing data | |
| CN111104889A (en) | Water body remote sensing identification method based on U-net | |
| CN111339948B (en) | Automatic identification method for newly-added buildings of high-resolution remote sensing images | |
| CN105809140A (en) | Method and device for extracting surface water body information based on remote sensing model | |
| CN111582194A (en) | Multi-temporal high-resolution remote sensing image building extraction method based on multi-feature LSTM network | |
| CN111007013B (en) | Crop rotation fallow remote sensing monitoring method and device for northeast cold region | |
| Zhang et al. | Combinational biophysical composition index (CBCI) for effective mapping biophysical composition in urban areas | |
| S Bhagat | Use of remote sensing techniques for robust digital change detection of land: A review | |
| CN103839267A (en) | Building extracting method based on morphological building indexes | |
| CN111062368A (en) | City update region monitoring method based on Landsat time sequence remote sensing image | |
| Liu et al. | Ephemeral gully recognition and accuracy evaluation using deep learning in the hilly and gully region of the Loess Plateau in China | |
| CN116385867A (en) | Ecological land block monitoring, identifying and analyzing method, system, medium, equipment and terminal | |
| Colditz et al. | Detecting change areas in Mexico between 2005 and 2010 using 250 m MODIS images | |
| Choudhary et al. | Surface water area extraction by using water indices and DFPS method applied to satellites data | |
| CN117593664B (en) | Waterproof surface extraction method and device integrating multi-mode remote sensing data | |
| Weng | Remote sensing of urban biophysical environments | |
| Liu et al. | Water extraction on the hyperspectral images of gaofen-5 satellite using spectral indices |
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 |