A multi-spectral remote sensing image water body extraction method for multi-service system integration

Abstract

The present invention discloses a method for extracting water bodies from multispectral remote sensing images for the integration of multi-service systems, which includes the following steps: S101) obtaining multispectral remote sensing image data of the target area; S102) preprocessing the multispectral remote sensing image data; S103) selecting typical features of water bodies and non-water bodies for analysis and constructing spectral curves of different features; S104) fusing the normalized difference vegetation index and the process infrared index to construct a visible-infrared fusion index model; S105) inputting the preprocessed remote sensing image and extracting water body pixels through the visible-infrared fusion water index algorithm model. The present invention can solve the problem that water bodies and building shadows in remote sensing images are easily confused, quickly grasp the changes in surface water, meet the requirements of underpass rainstorm waterlogging simulation and analysis, urban waterlogging deduction, water pollution source tracing, etc., and realize the precise management and efficient governance of urban water resources.

Description

Technical Field

[0001] The invention relates to a remote sensing image water body extraction method, in particular to a multi-spectral remote sensing image water body extraction method oriented to multi-service system fusion. Background Art

[0002] Water plays an indispensable role in the ecosystem. It is also a key factor in economic and social development and is of great significance to regional ecological security. At the same time, changes in the water resources environment are closely related to climate change. Drastic changes in the water resources environment will have a serious impact on normal human production and life. At the same time, with the rapid development of cities, dynamic monitoring of urban water security is of great significance to the normal operation of cities. The rapid development of cities across the country has increased the demand for water resources. How to better achieve the goal of five water governance, namely, sewage treatment, flood prevention, drainage, water supply, and water conservation, and achieve precise governance of water resources and environment is one of the important problems that government departments at all levels face and need to solve urgently.

[0003] For example, the Water Affairs Bureau, Emergency Bureau, Ecological Bureau, Agriculture and Rural Bureau and other units all need real-time urban water data to meet the calculation requirements of business service plug-ins such as underpass rainstorm waterlogging simulation and analysis plug-in, flood and drought disaster warning plug-in, pump station drainage analysis plug-in, urban waterlogging simulation plug-in, water pollution source tracing business plug-in, water body potential pollution source identification plug-in, water environment assessment plug-in, water environment quality monitoring and early warning plug-in, water environment sensitive point pre-research and prejudgment plug-in, etc., to truly serve the work of urban intelligent management such as waterlogging control, flood and drought disaster warning, and water ecological environment protection. However, the real-time performance of the existing water data of various units is insufficient, and the water body and the shadow of the house are easily confused in the current water body extraction algorithm based on remote sensing technology. The main reason is that different types of water bodies have significant differences in shape, texture, and color due to different regional environments, and are subject to different interferences. Due to its strong absorption characteristics of electromagnetic waves, the overall reflectivity of water bodies is low. The methods of completing water body extraction based on remote sensing images are summarized and summarized, and it is found that the existing water body extraction methods can be classified into three types: manual visual interpretation, machine learning method, and threshold method.

[0004] Manual visual interpretation is the most primitive water extraction method, which is often easily affected by image resolution, regional environmental conditions, and the quality of the extraction personnel during the extraction process. This method cannot meet the requirements of water extraction in a large area, and it also takes too much time during the extraction process, which cannot meet the timeliness requirements in flood prevention and disaster reduction, ecological protection, and water environment monitoring.

[0005] The machine learning method is based on a large number of training samples. By annotating the features of various training samples and using various classifiers to build a learning model with multiple hidden layers, through the supervised training of a large number of training samples, machine learning can independently complete the learning of multi-level and multi-scale features of images, and comprehensively coordinate the detailed information features such as image color, edges, and textures. However, the machine learning method is often easily limited by training samples. It is difficult to obtain a large number of training samples for different water body types in different remote sensing images. In the case of insufficient training samples, the extraction results often cannot meet the accuracy requirements.

[0006] The threshold method is based on the fact that the reflection characteristics of water body information in the electromagnetic spectrum are different from those of other non-water ground objects. After selecting the characteristic band for operation, water body extraction is achieved through a threshold. The methods for completing water body extraction using a threshold mainly include the single-band method and the multi-band method. The single-band method is based on the fact that water bodies present different characteristics in images of different wavelengths. By analyzing the remote sensing images of each band and establishing logical judgment rules, water body information can be extracted, that is: water body threshold segmentation based on a single band. The most commonly used method in the multi-band method is the index method. The index method is based on the fact that different ground objects have different reflectivities in different bands. By combining different bands for operation, the characteristic values of the target ground object are highlighted, that is: through multi-band combined operation, water body information can be extracted. Based on this principle, a variety of water body extraction models have been proposed, such as the Normalized Difference Water Index (NDWI). This index can better suppress vegetation information, but it is more affected by thin clouds and shadows. To improve this problem, some scholars have used the short-wave near-infrared band instead of the near-infrared band to improve NDWI and proposed the Modified Normalized Difference Water Index (MNDWI). The MNDWI index can highlight water body information, but it is also easy to confuse water bodies with shadows. In addition, to solve the above problems, some scholars have successfully constructed a multi-band water index (MBWI). MBWI uses a total of 5 bands and has tested the effectiveness of MBWI in many areas with complex environments. MBWI can better suppress non-water body information, but one of its disadvantages is that it is not suitable for extracting small water bodies. At the same time, when introducing too many bands, it also leads to the superposition of errors between different bands.

[0007] More importantly, the ground objects in urban areas are complex, and there are often different types of non-water ground objects such as buildings, vegetation, and bare land. At the same time, due to the complexity of the city, the spectral characteristics of these non-water ground objects vary significantly in different regions. This also leads to many situations where existing spectral indices such as NDWI and MNDWI misjudge building shadow pixels as water bodies when extracting water bodies in urban areas, resulting in over-extraction of water bodies and unable to meet the needs of urban refined management and precise governance. Summary of the Invention

[0008] The technical problem to be solved by the present invention is to provide a multi-spectral remote sensing image water body extraction method for multi-service system integration, which can solve the problem that water bodies and building shadows in remote sensing images are easily confused, quickly grasp the changes in surface water, meet the needs of underpass rainstorm waterlogging simulation and analysis, urban waterlogging deduction, water pollution tracing, etc., and achieve precise management and efficient governance of urban water resources.

[0009] The technical solution adopted by the present invention to solve the above technical problem is to provide a multi-spectral remote sensing image water body extraction method for multi-service system integration, including the following steps: S101) Obtain multi-spectral remote sensing image data of the target area; S102) Preprocess the multi-spectral remote sensing image data; S103) Select typical ground objects of water bodies and non-water bodies for analysis and construct spectral curves of different ground objects; S104) Fusion of the normalized difference vegetation index and the process infrared index to construct a visible-infrared fusion index model; S105) Input the preprocessed remote sensing image, and extract water body pixels through the visible-infrared fusion water body index algorithm model.

[0010] Further, the multi-spectral remote sensing image data has a red band, a near-infrared band, and a short-wave infrared band. The band range of the near-infrared band is 750nm - 910nm, the band range of the red band is 590nm - 700nm, and the band range of the short-wave infrared band is 1550nm - 1800nm.

[0011] Further, the step S102 includes: S1021) Use the absolute radiometric calibration coefficient released by the satellite sensor to eliminate the error of the sensor itself; S1022) Perform atmospheric correction on the apparent reflectance formed after radiometric correction to eliminate the influence of various molecules in the atmosphere on the ground object reflection; S1023) Perform orthorectification on the geometric distortion of the image caused by terrain undulation during shooting; S1024) Finally, perform fusion processing on the multi-spectral remote sensing image to obtain a high-resolution image.

[0012] Further, in S103, water body pixels, building pixels, vegetation pixels, and bare land pixels are selected in the image to form a feature matrix of different ground object types, and the mean values of each type of ground object in all bands are obtained in turn to obtain the spectral curve of each type of ground object.

[0013] Further, the building pixels are divided into dark buildings and light buildings, and their respective spectral curves are formed.

[0014] Further, the feature matrix is:

[0015]

[0016] In the above formula, A iThe feature vectors composed of the characteristic pixels representing a certain type of ground object, where i represents one type of ground object among water bodies, buildings, vegetation, and bare land; n represents the nth pixel, m represents the mth band, and a mn represents the pixel reflectance value of the nth pixel in the mth band;

[0017] Obtain the mean value of all characteristic pixels of the ith type of ground object in the mth band. The calculation method is shown in the following formula:

[0018]

[0019] In the above formula, represents the mean value of all characteristic pixels of the ith type of ground object in the mth band. Sequentially obtain the mean values of the 4 types of ground objects in all bands;

[0020] After the calculation is completed, the following vector is formed:

[0021]

[0022] In the above formula, i represents one type of ground object among water bodies, buildings, vegetation, and bare land; B i represents the spectral curve of the ith type of ground object obtained after calculation.

[0023] Furthermore, the calculation method of the normalized difference vegetation index NDVI in the step S104 is as follows:

[0024]

[0025] The calculation method of the infrared index IR in the step S104 is as follows:

[0026] IR = NIR + SWIR;

[0027] NIR represents the near-infrared band with a wavelength range between 750nm - 910nm, R represents the red band with a wavelength range between 590nm - 700nm, and SWIR represents the short-wave infrared band with a wavelength range between 1550nm - 1800nm.

[0028] Furthermore, the calculation method of the visible-infrared fusion index VIFWI in the step S104 is as follows:

[0029]

[0030] Furthermore, the calculation method of the visible-infrared fusion index VIFWI in the step S104 is as follows:

[0031]

[0032] Further, it also includes selecting the Normalized Difference Water Index for comparison and verification. For the remotely sensed image after the input is preprocessed, the water pixels are extracted through the Normalized Difference Water Index algorithm model. The calculation method of the Normalized Difference Water Index NDWI is as follows:

[0033]

[0034] G represents the remotely sensed image of the green band, and the band range is between 520nm - 580nm.

[0035] The present invention has the following beneficial effects compared with the prior art: The multi-spectral remotely sensed image water body extraction method for multi-service system integration provided by the present invention constructs a visible light and infrared fusion water index based on the spectral characteristics of water bodies, realizes the efficient extraction of water bodies in urban areas, and can well suppress the interference of shadows, thereby being able to solve the problem that water bodies and building shadows in remotely sensed images are extremely easy to be confused, quickly master the changes in surface water, meet the needs of underpass rainstorm waterlogging simulation and analysis, urban waterlogging deduction, water pollution tracing, etc., and realize the precise management and efficient governance of urban water resources. Description of the Drawings

[0036] Figure 1 It is a schematic diagram of the water body extraction process of the multi-spectral remotely sensed image for multi-service system integration of the present invention;

[0037] Figure 2 It is a flowchart of the preprocessing of the multi-spectral remotely sensed image of the present invention;

[0038] Figure 3 It is the spectral curves of different ground objects of the present invention;

[0039] Figure 4 It is to verify the extraction result by using the Landsat 9 image in the present invention. In the figure, a is the original satellite image, b is the water body extraction result of the visible light and infrared fusion water index VIFWI algorithm model, and c is the water body extraction result of the Normalized Difference Water Index NDWI algorithm model;

[0040] Figure 5 It is to verify the extraction results of VIFWI and NDWI by selecting urban areas with more dark buildings in the present invention. In the figure, a is the original satellite image, b is the water body extraction result of the visible light and infrared fusion water index VIFWI algorithm model, and c is the water body extraction result of the Normalized Difference Water Index NDWI algorithm model. Detailed Embodiments

[0041] The present invention will be further described below with reference to the drawings and embodiments.

[0042] Figure 1 It is the construction process of the visible light and infrared fusion water index VIFWI.

[0043] Please refer to Figure 1 , the multi-spectral remote sensing image water body extraction method for multi-service system integration provided by the present invention has the following specific operation steps:

[0044] Step 101:

[0045] Select the multi-spectral remote sensing image data of the target area according to the monitoring time. It is required that the selected multi-spectral remote sensing image data must have a red band, a near-infrared band, and a short-wave infrared band. Among them, the band range of the near-red band supports 750nm - 910nm, the band range of the red band is 590nm - 700nm, and the band range of the short-wave infrared band is 1550nm - 1800nm. The present invention supports any data within this interval of band ranges. At present, the tests of three types of image data, namely Landsat7, Lsandat8, and Landsat9, have been completed.

[0046] Step 102:

[0047] During the process of a remote sensing satellite sensor receiving electromagnetic waves for imaging, it often receives the influence of factors such as the atmosphere and terrain undulation. Therefore, it is necessary to preprocess the original satellite image data. Radiometric calibration is based on the absolute radiometric calibration coefficients released by the satellite sensor, and the calibration coefficients are used to eliminate the errors of the sensor itself. Atmospheric correction is to process the apparent reflectance formed after radiometric correction. The atmospheric correction model is used to perform atmospheric correction on the apparent reflectance image to eliminate the influence of various molecules in the atmosphere on the reflection of ground objects. Orthorectification is a process carried out to avoid geometric distortion of the image caused by terrain undulation during shooting. Finally, if high-resolution images are required, the images can be fused, as Figure 2 shown.

[0048] Step 103:

[0049] Select typical ground objects of water bodies and non-water bodies for analysis. Among them, non-water body ground objects mainly include buildings, bare land, and vegetation. In urban areas, building shadow pixels are often easily confused with water body pixels. Select 100 water body pixels, building pixels, vegetation pixels, and bare land pixels each in the image to form a feature matrix of different ground object types, specifically as follows:

[0050]

[0051] In the above formula, A i represents the feature vector jointly composed of the feature pixels of a certain type of ground object, and i represents one of the ground object types of water body, building, vegetation, and bare land. n represents the nth pixel, m represents the mth band, and a mn represents the pixel reflectance value of the nth pixel in the mth band.

[0052] Calculate the mean value of all feature pixels of the i-th type of ground object in the m-th band. The specific calculation method is shown in the following formula:

[0053]

[0054] In the above formula, represents the mean value of all feature pixels of the i-th type of ground object in the m-th band. It is necessary to calculate the mean values of 4 types of ground objects in all bands in sequence.

[0055] After the calculation is completed, the following vector is formed:

[0056]

[0057] In the above formula, i represents one type of ground object among water body, building, vegetation, and bare land. B i represents the spectral curve of the i-th type of ground object obtained after calculation.

[0058] The curve of the reflectance of an object changing with wavelength is called the reflection spectrum, and its shape reflects the spectral characteristics of the ground object. Therefore, taking the landsat 9 data as an example for illustration, spectral curves of different ground objects are constructed based on the above algorithm. Usually, only dark buildings are prone to be confused with water bodies. To better distinguish water bodies and non-water bodies, the building pixels that are prone to be confused with water bodies are divided into dark buildings and light buildings, and the formed spectral curves are as Figure 3 shown.

[0059] From Figure 3 it can be found that the band range of the near-infrared band of the water body supports 750nm - 910nm, and the band range of the short-wave infrared band is 1550nm - 1800nm, which can be well distinguished from other ground objects. After introducing the red band (the band range of the red band is 590nm - 700nm), the waveform of the spectral curve shows an obvious downward trend, specifically manifested as "sharp decline - reaching the lowest value - slight rise", while the waveforms of vegetation, bare land, dark buildings, and light buildings obviously show "sharp rise - reaching the highest value - slight decline". Therefore, the red band, near-infrared band, and short-wave infrared band are selected to construct the model.

[0060] Step 104:

[0061] The normalized difference vegetation index NDVI can well distinguish vegetation from other ground objects, and at the same time NDVI can significantly increase the difference between vegetation and water bodies. Therefore, the present invention takes the normalized difference vegetation index NDVI as a variable.

[0062]

[0063] In public expressions, NIR represents the near infrared band with a wavelength range of 750nm-910nm, R represents the red band with a wavelength range of 590nm-700nm, and SWIR represents the short-wave infrared band with a wavelength range of 1550nm-1800nm.

[0064] At the same time, it can be found from the spectral curve that water bodies are clearly distinguished from other ground objects in the near-infrared band and the short-wave infrared band. Therefore, the two bands are combined to further highlight the difference between water bodies and vegetation, buildings, and bare land, and the process infrared index IR is constructed. The calculation method is as follows:

[0065] IR=NIR+SWIR

[0066] Finally, the NDVI and IR indices are combined for calculation, because the non-water body value in NDVI is much higher than the water body value, and the water body value in IR index is much higher than the non-water body value. Therefore, a ratio index is constructed - Visible Infrared Fusion Water Index (VIFWI).

[0067]

[0068] In order to further realize the convenience of VIFWI calculation, the above three calculation formulas are combined and simplified to form the visible light infrared fusion water index VIFWI. The final visible light infrared fusion index is as follows:

[0069]

[0070] In the formula, NIR represents the near infrared band with a wavelength range of 750nm-910nm, R represents the red band with a wavelength range of 590nm-700nm, and SWIR represents the short-wave infrared band with a wavelength range of 1550nm-1800nm.

[0071] Step 105:

[0072] The preprocessed remote sensing image is input and the water body pixels are extracted through the visible infrared fusion water index VIFWI algorithm model.

[0073] In order to further verify that VIFWI has a better extraction effect on water areas and can suppress the interference of shadow areas, the normalized water index NDWI is selected as a comparison for verification. The calculation method is as follows:

[0074]

[0075] Where G represents the remote sensing image of the green band, and the band range is between 520nm-580nm.

[0076] The pre - processed remote sensing image is used to extract water pixels through the Normalized Difference Water Index (NDWI) algorithm model.

[0077] Step 106:

[0078] Use Landsat 9 images to verify the extraction results of small water bodies. Figure 4 In [figure], a is the original satellite image, b is the water body extraction result of the Visible Infrared Fusion Water Index (VIFWI) algorithm model, and c is the water body extraction result of the Normalized Difference Water Index (NDWI) algorithm model. Figure 4 In [figure], the white areas in b and c are water body areas, and the black areas are non - water body areas. By comparing b and c, it can be found that the extraction effect of the Visible Infrared Fusion Water Index (VIFWI) algorithm on water bodies is better than that of the Normalized Difference Water Index (NDWI) algorithm. The Visible Infrared Fusion Water Index (VIFWI) algorithm can better distinguish non - water bodies and avoid the mis - extraction of water bodies.

[0079] Select urban areas with more dark - colored buildings to verify the extraction results of VIFWI and NDWI. Figure 5 In [figure], a is the original satellite image, b is the water body extraction result of the Visible Infrared Fusion Water Index (VIFWI) algorithm model, and c is the water body extraction result of the Normalized Difference Water Index (NDWI) algorithm model. Figure 5 In [figure], the white areas in b and c are water body areas, and the black areas are non - water body areas. In the original image, there is only a very small artificial lake in the upper left corner, and the rest are dark - colored buildings. By comparing the extraction results of VIFWI and NDWI, it can be found that VIFWI can well suppress the interference of shadows on water bodies. Thus, by checking and evaluating the Visible Infrared Fusion Water Index (VIFWI) algorithm, the present invention has a high water body extraction accuracy. The urban surface water body boundary has both good regional integrity and maintains local details, and at the same time has a good inhibitory effect on urban building shadows, improving the water body extraction accuracy.

[0080] Based on the spectral characteristics of water bodies, the visible-infrared fusion water body index VIFWI is constructed in the present invention. And based on the waveform characteristics of water bodies and non-water bodies: the waveform of water bodies is "sharp decline - reaching the lowest value - slight rise", while the waveforms of vegetation, bare land, dark buildings, and light buildings significantly show "sharp rise - reaching the highest value - slight decline". The red band, near-infrared band, and short-wave infrared band are selected to construct a model. At the same time, the data information is deeply mined from the spectral curve. Based on the obvious distinction between water bodies and other ground objects in the near-infrared band and the short-wave infrared band, the combination operation of the two bands is carried out to further highlight the differences between water bodies and vegetation, buildings, and bare land, and the process infrared index IR is constructed. And the NDVI and the IR index are combined for calculation. Since the non-water body value in NDVI is much higher than the water body value, and the water body value in the IR index is much higher than the non-water body value, the ratio index - visible-infrared fusion water body index VIFWI is constructed, realizing the efficient extraction of water bodies in urban areas and can well suppress the interference of shadows.

[0081] Urban ground objects are composed of different types of ground objects such as buildings, vegetation, bare land, and water bodies. The performance of each type of ground object in remote sensing images is affected by its own structure and the surrounding environment. There are often "same object, different spectra" and "same spectra, different objects" phenomena in spectral performance, and the texture and structure often also show different performances. It is precisely because of this problem that the extraction of water bodies in urban areas becomes more complicated. Water body pixels are often easily mixed with building shadows. Therefore, the present invention develops a multi-spectral remote sensing image water body extraction algorithm for urban intelligent management spatio-temporal integration - visible-infrared fusion water body index VIFWI. The visible-infrared fusion water body index VIFWI can effectively distinguish urban shadows and water bodies, reduce the misjudgment of water bodies caused by urban shadows, resulting in over-extraction of water body areas. Especially in areas with dense dark buildings, it can significantly reduce the misjudged pixels. At the same time, the visible-infrared fusion water body index VIFWI can also better distinguish the boundary between water bodies and non-water bodies and has great advantages in the extraction of small water bodies. For example, it can well distinguish the ridges between paddy fields.

[0082] In summary, the visible light and infrared fusion water body index VIFWI can meet the demand for real-time extraction of urban water body data in the operations of multiple departments such as the water affairs bureau, emergency bureau, ecological bureau, and agriculture and rural affairs bureau. It provides real-time water body distribution data support for business service plugins such as underpass rainstorm waterlogging simulation and analysis plugins, water and drought disaster warning plugins, pumping station drainage analysis plugins, urban waterlogging deduction plugins, water pollution source tracing business plugins, potential water pollution source identification plugins, water environment assessment plugins, water environment quality monitoring and warning plugins, and water environment sensitive point pre-research and pre-judgment plugins in the urban intelligent management spatio-temporal integration system. Through the visible light and infrared fusion water body index VIFWI algorithm, the distribution of water resources within the city can be quickly obtained. Especially in the case of urban waterlogging, the water body distribution area within the city can be quickly obtained, providing an effective decision-making basis for urban waterlogging treatment and emergency response. This algorithm can also assist in water environment monitoring and assessment. Urban water bodies are usually closely related to the urban ecosystem. Changes in the water body area will directly affect the urban ecological environment, leading to changes in the urban microclimate. The visible light and infrared fusion water body index VIFWI algorithm can provide real-time water body data for water environment monitoring, which is of great significance for water ecological protection. Based on the importance of the urban water resources environment, with the water environment and water resources security guarantee as the core, the present invention realizes the real-time and efficient extraction of water bodies, providing high-quality services such as urban waterlogging treatment, water and drought disaster warning, and water ecological environment guarantee for urban governance.

[0083] Although the present invention has been disclosed above with preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications and improvements without departing from the spirit and scope of the present invention. Therefore, the protection scope of the present invention shall be defined by the claims.

Claims

1. A method for extracting water bodies from multi-spectral remote sensing images for the integration of multi-service systems, characterized in that, It includes the following steps: S101) Obtain the multi-spectral remote sensing image data of the target area; S102) Preprocess the multi-spectral remote sensing image data; S103) Select typical water and non-water ground objects for analysis, and construct spectral curves of different ground objects; S104) Fuse the normalized difference vegetation index and the process infrared index to construct a visible-infrared fusion index model; S105) Input the preprocessed remote sensing image, and extract water pixels through the visible-infrared fusion water index algorithm model; In S103, select water pixels, building pixels, vegetation pixels, and bare land pixels in the image to form a feature matrix of different ground object types, and successively calculate the mean value of each type of ground object in all bands to obtain the spectral curve of each type of ground object; The feature matrix is: In the above formula, represents the feature vector jointly constituted by the feature pixels of a certain type of ground object, i represents one type of ground object among water bodies, buildings, vegetation, and bare land; n represents the n th pixel, m represents the m th band, represents the pixel reflectance value of the m th band at the n th pixel; Obtain the i average value of all characteristic pixels of the m type of ground object on the band, and the calculation method is shown in the following formula: In the above formula, represents the mean value of all characteristic pixels of the i type of ground object in the m th band, and the mean values of the 4 types of ground objects in all bands are obtained in sequence; After the calculation is completed, the following vector is formed: In the above formula, i represents a type of ground object among water bodies, buildings, vegetation, and bare land; represents the spectral curve of the i th type of ground object obtained after calculation; The calculation method of the normalized difference vegetation index NDVI in step S104 is as follows: ； The calculation method of the process infrared index IR in step S104 is as follows: ； NIR represents the near-infrared band with a band range between 750nm and 910nm, R represents the red band with a band range between 590nm and 700nm, SWIR represents the short-wave infrared band with a band range between 1550nm and 1800nm; The calculation method of the visible-infrared fusion index VIFWI in step S104 is as follows: 。 2. The method for extracting water bodies from multi-spectral remote sensing images for the integration of multi-service systems according to claim 1, characterized in that, The multi-spectral remote sensing image data has a red band, a near-infrared band, and a short-wave infrared band. The band range of the near-infrared band is 750nm - 910nm, the band range of the red band is 590nm - 700nm, and the band range of the short-wave infrared band is 1550nm - 1800nm.

3. The method for extracting water bodies from multi-spectral remote sensing images for the integration of multi-service systems according to claim 1, characterized in that, Step S102 includes: S1021) Use the absolute radiometric calibration coefficient released by the satellite sensor to eliminate the error of the sensor itself; S1022) Perform atmospheric correction on the apparent reflectance formed after radiometric correction to eliminate the influence of various molecules in the atmosphere on the ground object reflection; S1023) Perform orthorectification on the geometric distortion of the image caused by terrain undulation during shooting; S1024) Finally, perform fusion processing on the multi-spectral remote sensing image to obtain a high-resolution image.

4. The method for extracting water bodies from multi-spectral remote sensing images for the integration of multi-service systems according to claim 1, characterized in that, The building pixels are divided into dark buildings and light buildings, and their respective spectral curves are formed.

5. The method for extracting water bodies from multi-spectral remote sensing images for the integration of multi-service systems according to claim 1, characterized in that, The simplified calculation method of the visible-infrared fusion index VIFWI in step S104 is as follows: 。 6. The multi-spectral remote sensing image water body extraction method for multi-service system integration according to claim 1, characterized in that, It also includes selecting the normalized difference water index for comparison and verification. For the input preprocessed remote sensing image, extract water pixels through the normalized difference water index algorithm model; the calculation method of the normalized difference water index NDWI is as follows: ； G represents the remote sensing image of the green band, and the band range is between 520nm and 580nm.

Images