CN120472323B

CN120472323B - A stratified water extraction method driven by enhanced water index

Info

Publication number: CN120472323B
Application number: CN202510954509.8A
Authority: CN
Inventors: 吴立新; 矫志军
Original assignee: Central South University
Current assignee: Central South University
Priority date: 2025-07-11
Filing date: 2025-07-11
Publication date: 2025-10-10
Anticipated expiration: 2045-07-11
Also published as: CN120472323A

Abstract

The present invention relates to the field of remote sensing science, and discloses a layered water body extraction method driven by an enhanced water body index, including remote sensing image preprocessing; remote sensing index calculation; ground feature segmentation; layered feature fusion and decision voting; water body spatial distribution and mapping. The present invention utilizes the spatiotemporal invariance of ground feature spectral information to construct an enhanced water body index EWI, and combines remote sensing indices such as vegetation and clouds to jointly analyze the ground feature characteristics in the scene, thereby achieving accurate quantitative identification of surface water distribution. In response to the problem that water bodies and other interfering ground features in remote sensing images show similar trends in spectral characteristics, this method combines EWI with other remote sensing indices to propose a layered water body extraction strategy, which can not only significantly enhance the spectral signal of water bodies under thin cloud cover conditions, but also effectively suppress the interference caused by clouds and shadows, thereby improving the accuracy of water body extraction; the present invention has broad application prospects in the fields of flood disaster emergency response, water resources management, etc.

Description

Layered water body extraction method for enhancing water body index drive

Technical Field

The invention relates to the field of remote sensing science, in particular to a layered water body extraction method for enhancing water body index drive.

Background

The accurate extraction of the water body information has great significance for disaster prevention and reduction, water resource management and ecological environment protection. Dynamic change of water distribution is closely related to natural disasters such as flood, drought and the like, and directly affects balance of an ecological system. Therefore, a rapid and reliable water body extraction method is important for disaster prevention and early warning, water resource management and ecological protection.

However, in practical applications, complex environmental conditions (such as clouds, turbid water bodies, cloud shadows, etc.) form serious interference to water body extraction, so that accurate and real-time acquisition of water body distribution becomes very difficult. The traditional water body extraction method has obvious limitations under the complex conditions, and is difficult to accurately identify the water body region, and mainly comprises the following steps of 1. Cloud and water body confusion problem, namely, in a remote sensing image, a cloud layer is usually represented as a highlight region, reflection characteristics of the cloud layer are overlapped with water body trend in partial wave bands, so that the cloud is easy to be misjudged as the water body by the traditional spectrum index method, and in addition, the cloud accumulation or uneven distribution of low-layer cloud and complex spectrum information of the edge region of the cloud are further increased, so that the difficulty of distinguishing the cloud from the water body is further increased. 2. The problem of spectrum characteristic variation of the thin cloud and turbid water body is that light path scattering and mixing effects can be introduced by thin cloud coverage or sediment mixing, so that spectrum signals of the blocked water body are suddenly raised or distorted. For example, the spectral characteristics of the water under thin cloud coverage are significantly elevated, making it difficult for traditional water index methods to extract the complete water boundary. Meanwhile, due to the influence of sediment and suspended particles, the spectral reflection characteristic of the turbid water body in the near infrared band may be similar to that of bare soil or vegetation, so that missed detection or false detection frequently occurs. 3. The cloud shadow can weaken ground object reflection signals of the shielded area, so that the spectral characteristics of dark target ground objects such as vegetation, bare soil and the like are similar to those of a water body. For example, in a region with larger mountain or topography fluctuation, the cloud shadow distribution is irregular and often overlaps with the actual water body region, so that the difficulty of accurately extracting the water body is increased, and in addition, the weak spectrum signal of the shadow region is easily misjudged as the water body by the traditional method, so that the extraction precision is reduced.

The complex remote sensing phenomenon shows that the traditional water body extraction method has obvious defects when dealing with complex scenes, and a more robust technical means is needed to effectively overcome the interferences of clouds, thin clouds, turbid water bodies, cloud shadows and the like, and realize accurate extraction of water body information. To cope with these problems, it is particularly necessary to develop a stratified water extraction method based on enhanced water index driving.

Disclosure of Invention

The invention aims to provide a layered water body extraction method for enhancing water body index drive, which can effectively distinguish surface water bodies from interference ground objects by fully utilizing space-time invariance of ground object spectrum information and combining a layered water body extraction strategy. The research and application of the invention can provide reliable technical support for remote sensing drawing of flood disaster range, rapid disaster condition assessment and emergency disaster relief response decision, and has important scientific significance and wide practical value.

In order to achieve the above object, the present invention provides a layered water extraction method for enhancing the index drive of a water body, comprising the following steps:

step S1, acquiring a multi-scene earth surface multi-spectrum image, and preprocessing the acquired multi-spectrum image, wherein the preprocessing step comprises the following steps:

s1.1, performing fine correction on the multispectral image, wherein the fine correction comprises geometric correction, homonymous image point pickup and registration;

s1.2, performing atmospheric correction and radiation calibration according to the multispectral image after the fine correction so as to obtain a multispectral image data block of a research area;

Step S2, respectively calculating an enhanced water index feature, an improved multispectral vegetation index feature and a cloud shadow index feature based on the multispectral image data block obtained by preprocessing in the step S1;

Step S3, threshold segmentation is carried out on the enhanced water body index features, the improved multispectral vegetation index features and the cloud shadow index features obtained by the calculation in the step S2, and corresponding ground object distribution is extracted in a layered mode;

S4, fusing the multi-layer ground object distribution obtained in the step S3 in a voting decision mode, and extracting water body distribution according to a voting result;

and S5, extracting and drawing an earth surface water body distribution map of the satellite observation area.

Further, in the step S2, a calculation formula of the enhanced water index feature EWI is:

Wherein Green is the ground object reflectivity of the Green light wave band in the multispectral image, NIR is the ground object reflectivity of the near infrared wave band in the multispectral image, and SWIR2 is the ground object reflectivity of the second short wave infrared wave band in the multispectral image.

Further, in the step S2, the calculation formula of the multispectral vegetation index feature MMSVI is:

Wherein, RE1 is the ground object reflectivity of the first Red band in the multispectral image, RE2 is the ground object reflectivity of the second Red band in the multispectral image, red is the ground object reflectivity of the Red band in the multispectral image, and Blue is the ground object reflectivity of the Blue band in the multispectral image.

Further, in the step S2, a calculation formula of the cloud shadow index feature is:

Wherein NIR is the ground object reflectivity of the near infrared band in the multispectral image, SWIR1 is the ground object reflectivity of the first short wave infrared band in the multispectral image.

Compared with the prior art, the invention has the following beneficial effects:

(1) The method for enhancing the water index integrates the advantages of multiband optical remote sensing images, the constructed spectrum enhancement model can obviously improve the spectrum characteristic contrast of thin cloud coverage and turbid water, and effectively weaken the influence of interference factors such as cloud, cloud shadow, dark targets and the like, thereby realizing efficient and accurate extraction of the water and providing reliable technical support for the fields of flood disaster monitoring, water resource management and the like.

(2) The water body layering extraction method provides a new technical scheme for the collaborative processing of multi-time-phase and multi-source optical images. The method has high automatic operation capability, does not need manual intervention, and can realize automatic identification and layered extraction of water distribution in satellite images, thereby remarkably improving the efficiency and precision of water information extraction.

(3) The hierarchical extraction strategy for enhancing the water index drive effectively improves the reliability of the water extraction result by constructing the multi-scale spectrum enhancement and hierarchical research model. The method fully considers the diversity of water distribution and the spectral characteristic change in a complex scene, provides a brand new technical thought, and realizes the accurate extraction of water information. The strategy not only perfects the water body extraction model in theory, but also provides solid technical support for the fields of flood disaster monitoring, water resource management and the like in practical application.

In addition to the objects, features and advantages described above, the present invention has other objects, features and advantages. The invention will be described in further detail with reference to the accompanying drawings.

Drawings

The accompanying drawings are included to provide a further understanding of embodiments of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain, without limitation, the embodiments of the invention. In the drawings:

FIG. 1 is a flow chart of a method for enhanced water index driven stratified water extraction provided by an embodiment of the present invention;

FIG. 2 (a) is a spectral curve characterization of water under the thin cloud and vegetation under the shadow in a Sentinel-2 satellite;

FIG. 2 (b) is a spectral curve characterization of water, vegetation and clouds in a Sentinel-2 satellite;

FIG. 2 (c) is a spectral plot of water under a thin cloud and vegetation under a shadow in a Landsat-8 satellite;

FIG. 2 (d) is a spectral curve characterization of water, vegetation and clouds in Landsat-8 satellites;

FIG. 2 (e) is a spectral curve characterization of water under the thin clouds and vegetation under the shadows in the Landsat-9 satellite

FIG. 2 (f) is a spectral curve characterization of water, vegetation and clouds in Landsat-9 satellites;

FIG. 3 (a) is a diagram of a Sentinel-2 false color composite image (blue channel: green, green channel: red, red channel: NIR);

FIG. 3 (b) shows the EWI index results obtained by Sentinel-2 calculation, with values ranging from (0-0.2);

FIG. 3 (c) shows the result of the MNDWI index calculated by Sentinel-2, with values ranging from (0.2-0.5);

FIG. 3 (d) shows the result of the MNDWI index calculated by Sentinel-2, with values ranging from (0.3-0.5);

FIG. 3 (e) shows the result of the MNDWI index calculated by Sentinel-2, with values ranging from (0.4-0.5);

FIG. 3 (f) shows a Landsat-9 pseudo-color composite image (blue channel: green, green channel: red, red channel: NIR);

FIG. 3 (g) shows the EWI index results calculated for Landsat-9, with values ranging from (0-0.2);

FIG. 3 (h) shows the result of the MNDWI index calculated by Landsat-9, with values ranging from (0.2-0.5);

FIG. 3 (i) shows the result of the MNDWI index calculated for Landsat-9, with values ranging from (0.3-0.5);

FIG. 3 (j) shows the result of the MNDWI index calculated by Landsat-9, with values ranging from (0.4-0.5);

FIG. 3 (k) shows a Landsat-8 pseudo-color composite image (blue channel: green, green channel: red, red channel: NIR);

FIG. 3 (l) shows the EWI index result calculated by Landsat-8, with values ranging from (0-0.2);

FIG. 3 (m) shows the result of the MNDWI index calculated by Landsat-8, with values ranging from (0.2-0.5);

FIG. 3 (n) shows the result of the MNDWI index calculated by Landsat-8, with values ranging from (0.3-0.5);

FIG. 3 (o) shows the result of the MNDWI index calculated by Landsat-8, showing the range of values (0.4-0.5).

Detailed Description

The present invention will be described in detail below with reference to the embodiments shown in the drawings, but it should be understood that the embodiments are not limited to the invention, and functional, method, or structural equivalents and alternatives according to the embodiments are within the scope of protection of the present invention by those skilled in the art.

By adopting the enhanced water index, the invention constructs a technical scheme specially aiming at layered extraction of water in a complex scene, and the main advantages of the technical scheme are that 1. The problem of cloud water confusion is effectively solved, and the traditional method is easy to generate misjudgment because cloud layers and water show similar spectral characteristics on partial wave bands. According to the invention, by integrating multi-band information, a high-efficiency spectrum difference model is constructed, and the distinguishing capability of cloud and water body can be remarkably improved, so that the accurate extraction of the water body area is realized. 2. Aiming at the problem of weakening of a water spectrum signal under the condition of thin cloud coverage, the invention adopts the enhanced water index to remarkably strengthen the water spectrum characteristics in the thin cloud area and highlight the spectrum difference between the water and other ground objects. Meanwhile, for the water body which is influenced by sediment and suspended particles and is in a turbid state, the spectral characteristics of the water body can be enhanced by enhancing the index of the water body, so that more water body distribution areas can be effectively identified. 3. The cloud shadow tends to weaken the spectrum signal of the ground object in the shielded area, so that the spectrum characteristics of dark target ground objects such as vegetation and the like and the water body tend to be similar. According to the method, the multispectral vegetation index and the cloud shadow index are improved in a combined mode to conduct targeted compensation on the cloud shadow effect, so that interference on a water body extraction result is reduced, and the extraction precision is further improved.

As shown in fig. 1, the method for extracting a layered water body (including flood, river, lake, etc.) for enhancing the index drive of the water body provided by the embodiment of the invention comprises the following steps:

S1, preprocessing an image after acquiring a multi-view surface multi-spectrum image in a research area. The method comprises the following steps:

And S1.1, performing image fine correction, namely performing fine registration on the sequential multispectral image, wherein the fine registration comprises homonymous image point pickup and image correction. And adopting ENVI5.3 software to effectively select the same-name image points, and finishing the fine registration of the heterogeneous multispectral image by a polynomial image correction method.

And S1.2, performing atmospheric correction and radiation calibration by adopting ENVI5.3 software according to the multispectral image after fine registration, thereby obtaining a multispectral image data block of the research area.

And S2, respectively calculating the enhanced water index characteristic, the improved multispectral vegetation index characteristic and the cloud shadow index characteristic based on the multispectral image data block obtained by preprocessing in the step S1.

And S2.1, carrying out enhanced water index feature calculation on the multispectral image data block. As shown in fig. 2 (a) -2 (f), a schematic diagram of an enhanced water index structure provided in this embodiment is specifically:

As shown in fig. 2 (a), in non-water features, green bands of vegetation, clouds and the like under shadows are significantly higher than other bands, which are very similar to the water spectrum characteristics, so that such feature interference is difficult to reject when calculating the index characteristics of the traditional water index. After the ground object reflectivity of the SWIR2 wave band and the ground object reflectivity of the NIR wave band are overlapped, it can be obviously seen that the water body reflectivity under the thin cloud is obviously higher than the vegetation reflectivity under the shadow, and the reflectivity of the non-water body ground object after being overlapped is obviously lower than the Green wave band of the non-water body ground object, so that the spectral information of the non-water body ground object is restrained while the water body information is effectively enhanced, and the spectral information is shown in a dotted line in fig. 2 (a). Then, by utilizing the difference between the reflectivity of the Green wave band and the reflectivity of the wave band after superposition, the water body and the non-water body can be effectively distinguished, as shown by the arrows ① and ② in the figure. As shown by arrows ③ and ④ in fig. 2 (a), the water body differenced to appear positive (arrow ④) and the non-water body differenced to appear negative (arrow ③). In order to further enhance the index robustness, a normalized difference index structure is introduced, green and NIR bands sensitive to water bodies are selected as denominators, and an index is constructed. The enhanced water index calculation formula constructed by the principle is as follows:

The EWI construction process is applied to other ground objects such as water bodies, vegetation and clouds, and is characterized in that the construction process is effective for distinguishing water from non-water, and the detailed process is shown in fig. 2 (b). Similarly, the principle is changed from Sentinel-2 satellite to Landsat-8 satellite (FIGS. 2 (c) and (d)), and Landsat-9 satellite (FIGS. 2 (e) and (f)), and the same effect is obtained.

S2.2, carrying out improved multispectral vegetation index characteristic calculation on the multispectral image data block, wherein a calculation formula is as follows:

wherein RE2 is the ground object reflectivity of the second Red band in the multispectral image, RE1 is the ground object reflectivity of the first Red band in the multispectral image, red is the ground object reflectivity of the Red band in the multispectral image, and Blue is the ground object reflectivity of the Blue band in the multispectral image. If the sensor has no red-side band, both RE2 and RE1 parameters can be replaced by near-infrared bands.

Step S2.3, calculating cloud shadow index features (Cloud Shadow Index, short for CSI) of the multispectral image data block, wherein the calculation formula is as follows:

And S3, carrying out threshold segmentation on the enhanced water body index features, the improved multispectral vegetation index features and the cloud shadow index features obtained by the calculation in the step S2, and extracting corresponding ground object distribution in a layering manner. The method comprises the following specific steps:

s3.1, acquiring water distribution calculation by threshold segmentation on the enhanced water index features, wherein the calculation formula is as follows:

wherein Water is the spatial distribution of the Water body layer characteristics.

S3.2, obtaining vegetation distribution calculation by threshold segmentation on the improved multispectral vegetation index features, wherein the calculation formula is as follows:

Wherein Vegetation is the vegetation layer characteristic space distribution.

Step S3.3, cloud shadow index features are subjected to threshold segmentation to obtain cloud and cloud shadow distribution calculation, wherein a calculation formula is as follows:

wherein Cloud is the spatial distribution of Cloud and Cloud shadow layer features.

And S4, fusing the multi-layer ground object distribution obtained in the step S3 in a voting decision mode, and extracting water body distribution according to voting result threshold segmentation. And the multi-layer ground object distribution is fused in a voting decision mode to obtain final water body distribution calculation, wherein the calculation formula is as follows:

FINALWATER is the final spatial distribution of the water body in the algorithm flow.

And S5, obtaining water distribution and drawing of the flood inundation area.

As shown in fig. 3 (a) -3 (o), a comparison chart of flood detection results is provided in this example. From the comparison of the EWI and MNDWI indices shown in FIGS. 3 (a) -3 (o), it can be concluded that EWI shows significant advantages over existing index methods in water extraction. The specific expression is as follows:

As shown in fig. 3 (a) -3 (e), in the Sentinel-2 image and index extraction result, the EWI can not only effectively inhibit the interference of cloud layers and cloud shadows, but also accurately extract the water body in the thin cloud coverage area. Fig. 3 (a) clearly shows the distribution of areas such as flood, thick cloud, thin cloud and cloud shadow, and comparison with the EWI index intensity distribution shown in fig. 3 (b) shows that the EWI significantly reduces the influence of cloud layers and cloud shadows on water extraction. Fig. 3 (c) shows that, in the conventional MNDWI, vegetation and thick clouds under cloud shadows are easily misjudged as water bodies. Further, fig. 3 (c) -3 (e) show that although cloud interference is reduced with the iteration of MNDWI threshold, the water extraction effect is still gradually reduced, and especially in the thin cloud and cloud shadow areas, the water information is obviously reduced. As can be seen from comparing fig. 3 (b) with fig. 3 (c), the EWI better suppresses the influence of clouds and cloud shadows while maintaining the water extraction accuracy.

Similarly, as shown in fig. 3 (f) -3 (j), in the Landsat-9 image and index extraction result, the EWI still can extract the surface water distribution to the greatest extent and inhibit the cloud interference, while as shown in fig. 3 (h) -3 (j), MNDWI gradually worsens the water extraction effect in the process of continuously optimizing the threshold value, although the cloud interference is eliminated. Further comparison in Landsat-8 images (FIGS. 3 (k) -3 (o)) has been verified that the EWI results in FIG. 3 (l) are consistent with the images described above, successfully eliminating cloud interference, while FIGS. 3 (m) -3 (o) show that even if the threshold is continuously adjusted, MNDWI still cannot completely eliminate cloud interference, resulting in poor water extraction.

In addition, in the consistency verification of the multi-source image platform, the comprehensive analysis results of the multi-source images such as Sentinel-2, landsat-9 and Landsat-8 show that the EWI shows excellent water extraction performance under different satellite platforms, particularly under the conditions of complex cloud layer, thin cloud and cloud shadow coverage, the influence of the cloud can be obviously reduced, and the accuracy and the robustness of the extraction result are ensured.

In conclusion, the hierarchical water body extraction strategy for enhancing the water body index drive not only effectively solves the precision problem of the traditional method under the interference of thin clouds, turbid water bodies and cloud shadows, but also realizes the accurate extraction of different water body types by constructing a multi-order spectral feature enhancement and hierarchical discrimination model. The method provides reliable water information support for flood disaster early warning, water resource monitoring and ecological environment protection, and has wide application value.

The above description is only of the preferred embodiments of the present invention and is not intended to limit the present invention, but various modifications and variations can be made to the present invention by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for extracting stratified water bodies driven by enhanced water index, characterized by comprising the following steps:

S1. Acquire multispectral images of multiple ground surfaces within the study area and preprocess the acquired multispectral images. The preprocessing is specifically as follows:

S1.1. Performing fine correction on the multispectral image, wherein the fine correction includes geometric correction, picking and registering homonymous image points;

S1.2. Perform atmospheric correction and radiometric calibration based on the precisely calibrated multispectral image to obtain multispectral image data blocks for the study area;

S2. Based on the multispectral image data block obtained by preprocessing in step S1, the enhanced water index feature, the improved multispectral vegetation index feature and the cloud shadow index feature are calculated respectively; the calculation formula of the enhanced water index feature EWI is:

Among them, Green is the reflectance of the ground objects in the green band of the multispectral image, NIR is the reflectance of the ground objects in the near infrared band of the multispectral image, and SWIR2 is the reflectance of the ground objects in the second short-wave infrared band of the multispectral image;

S3, performing threshold segmentation on the enhanced water index feature, the improved multispectral vegetation index feature, and the cloud shadow index feature calculated in step S2, and extracting the corresponding ground feature distribution in layers; the specific steps are as follows:

Step S3.1: Use threshold segmentation to obtain water body distribution calculation for the enhanced water body index feature. The calculation formula is:

Among them, Water is the spatial distribution of water layer characteristics;

Step S3.2: Use threshold segmentation to obtain vegetation distribution calculation for the improved multispectral vegetation index feature. The calculation formula is:

Among them, Vegetation is the spatial distribution of vegetation layer characteristics; MMSVI is the multispectral vegetation index characteristics;

Step S3.3: Use threshold segmentation to obtain cloud and cloud shadow distribution calculation based on cloud shadow index features. The calculation formula is:

Among them, Cloud is the spatial distribution of cloud and cloud shadow layer characteristics; CSI is the cloud shadow index characteristics;

S4. The multi-layer feature distribution obtained in step S3 is fused by voting, and the water body distribution is extracted according to the voting results. The final water body distribution is calculated by fusing the multi-layer feature distribution by voting. The calculation formula is:

Among them, FinalWater is the final spatial distribution of water bodies in the algorithm process;

S5. Draw a surface water distribution map of the study area based on the water distribution information extracted in step S4.

2. The method for extracting stratified water bodies according to claim 1, wherein in step S2, the calculation formula of the multispectral vegetation index feature MMSVI is:

Among them, RE1 is the reflectance of the ground objects in the first red edge band of the multispectral image, RE2 is the reflectance of the ground objects in the second red edge band of the multispectral image, Red is the reflectance of the ground objects in the red light band of the multispectral image, and Blue is the reflectance of the ground objects in the blue light band of the multispectral image.

3. The method for extracting stratified water bodies according to claim 1, wherein in step S2, the cloud shadow index feature is calculated as follows:

Among them, NIR is the reflectance of the ground objects in the near-infrared band of the multispectral image, and SWIR1 is the reflectance of the ground objects in the first short-wave infrared band of the multispectral image.