A pseudo-color feature extraction method for enhancing the water body extraction effect of Sentinel-1 SAR images

Abstract

The application discloses a pseudo-color feature extraction method for enhancing water body extraction effect of a Sentinel-1 SAR image. The application considers that the current SAR image water body extraction method is insufficient in mining and utilization of color features of the SAR image, and therefore automatically extracts color features of the Sentinel-1 SAR image through pseudo-color synthesis and color feature construction, so as to effectively distinguish water bodies from non-water bodies in the SAR image, and then improve the precision and reliability of Sentinel-1 SAR water body extraction. The application has the advantages that the proposed method does not need manual intervention, and can realize automatic extraction of color information of the Sentinel-1 SAR image. The method proposed by the application has the advantages of fast calculation efficiency, high automation degree and strong operability, is easy to implement, and has strong scalability of the whole model. Therefore, the application has high practical value.

Description

Technical Field

[0001] This invention belongs to the field of remote sensing image processing, and in particular relates to a pseudo-color feature extraction method to enhance the water body extraction effect of Sentinel-1 SAR images. Background Technology

[0002] In recent years, with the increasing severity of climate change, large-scale flood events have occurred frequently. Floods, due to their immense destructive power, often cause enormous economic and property losses, and even casualties. Flood mapping based on remote sensing imagery can provide spatial distribution information of floods to aid in the formulation of disaster relief measures and post-disaster reconstruction. SAR imagery, with its advantages of being available 24 / 7 and unaffected by weather conditions, is one of the main data sources for remote sensing flood mapping. Therefore, developing effective SAR imagery water body extraction methods is crucial for achieving timely and accurate flood mapping.

[0003] Unlike common optical remote sensing images, SAR images have limited spectral information and are subject to significant noise. Common SAR water body extraction methods often utilize backscatter intensity information from SAR images for threshold segmentation or construct and extract spatial features based on backscatter intensity maps to improve accuracy. However, in addition to intensity information and spatial features, SAR images still offer considerable potential for color feature extraction. Current SAR image water body extraction work suffers from insufficient information mining, particularly a severe underutilization of color information, resulting in significant room for improvement in accuracy and reliability. Therefore, exploring effective methods to mine color information from SAR images to enhance SAR water body extraction is crucial. Summary of the Invention

[0004] The purpose of this invention is to address the aforementioned shortcomings of existing technologies by providing a pseudo-color feature extraction method to enhance water body extraction from Sentinel-1 SAR images. This method first acquires Sentinel-1 SAR VV and VH polarimetric images of the target area. After a series of data preprocessing steps, including radiometric calibration, filtering, topographic correction, and decibel conversion, the Sentinel-1 dual-polarization water index (SDWI) is calculated. Then, the VV, VH polarimetric images and the SDWI are synthesized into a pseudo-color image. The synthesized pseudo-color image is then converted to the CIE XYZ color space through color space transformation. Color information such as the dominant wavelength and color saturation of the image are extracted within the CIE XYZ color space as new features to improve the accuracy of water body information extraction from Sentinel-1 SAR images.

[0005] The technical solution adopted in this invention is: a pseudo-color feature extraction method to enhance the water body extraction effect of Sentinel-1 SAR images, comprising the following steps:

[0006] Step 1: Acquire Sentinel-1 SAR images of the target area, including VV and VH polarimetric images, and preprocess the SAR images.

[0007] Step 2: Calculate the Sentinel-1 dual polarization water index SDWI using the preprocessed Sentinel-1 SAR VV and VH polarization images;

[0008] Step 3: Treat the VV polarization image, VH polarization image and water index SDWI image as the R, G and B bands of the color image respectively, and perform pseudo-color synthesis to obtain pseudo-color image I;

[0009] Step 4: Perform color space transformation on the pseudo-color image I generated in Step 3, converting it from the CIE RGB color space to the CIE XYZ color space. The CIE XYZ color space is a horseshoe-shaped region.

[0010] Step 5: Extract the dominant wavelength information of the image; Assuming that point N is a pixel in the original pseudo-color image I obtained by color space transformation, connect the point E (equal white point) and point N, and extend the line along the direction from point E to point N, intersecting the horseshoe curve at point M. The horseshoe curve is the outer contour line of the horseshoe region; then, the wavelength corresponding to the intersection point M is the dominant wavelength corresponding to point N. Obtain the dominant wavelength information of all pixels in the image in the CIE XYZ color space, which together constitute a new feature of the image;

[0011] Step 6: Calculate the length of line segment EM as a new feature of the image to enhance the similarity between pixels of the same color.

[0012] Step 7: Calculate the ratio ρ of the lengths of EN and EM as a new feature of the image;

[0013] Step 8: When using Sentinel-1 SAR imagery to extract water bodies in a certain area, use the features from steps 5-7 as input features for the water body extraction model to improve the accuracy of water body extraction.

[0014] Furthermore, the preprocessing described in step 1 includes radiometric calibration, filtering, topographic correction, and decibel conversion of the SAR image.

[0015] Furthermore, the dual polarization water index SDWI = log(10 × VV × VH).

[0016] Furthermore, the conversion formula for pseudo-color image I from CIE RGB color space to CIE XYZ color space in step 4 is as follows:

[0017]

[0018]

[0019] Pixels (m) in pseudo-color image I in RGB color space r ,m g ,m b After conversion using formulas (1) and (2), it can be represented in the CIEXYZ color space as (m x ,m y ,m z ).

[0020] Furthermore, in step 6, the length of line segment EM is measured using Euclidean distance, calculated as follows:

[0021]

[0022] In the formula, (M x M y () represents the coordinates of point M in the horseshoe-shaped region; point E is an isoenergetic white point with known coordinates (1 / 3, 1 / 3).

[0023] Furthermore, in step 7, the lengths of line segments EN and EM are both measured using Euclidean distance, and the ratio ρ of the lengths of line segments EN and EM is calculated as follows:

[0024]

[0025] The advantages of this invention are:

[0026] (1) The method proposed in this invention does not require manual intervention and can realize the automatic extraction of color information of Sentinel-1 SAR images, which has strong practical application value.

[0027] (2) The method proposed in this invention fully exploits the color information of Sentinel-1 SAR images through pseudo-color synthesis and color feature construction. The constructed color features can effectively distinguish between water bodies and non-water bodies in Sentinel-1 SAR images. To a certain extent, this compensates for the problem of insufficient mining and utilization of SAR image color features in current SAR image water extraction methods.

[0028] (3) The method proposed in this invention has fast computational efficiency, strong operability, high degree of automation, and strong scalability of the entire model. Therefore, this invention has high practical value.

[0029] In summary, the method proposed in this invention can effectively construct and extract color feature information from Sentinel-1 SAR imagery, thereby improving the accuracy of water body extraction from Sentinel-1 SAR imagery. It boasts high accuracy and automation, is easy to implement and expand, and has significant practical value. Attached Figure Description

[0030] Figure 1 It uses the CIE XYZ color space.

[0031] Figure 2 This is a flowchart of an embodiment of the present invention. Detailed Implementation

[0032] To facilitate understanding and implementation of the present invention by those skilled in the art, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0033] This invention provides a pseudo-color feature extraction method to enhance the water body extraction effect of Sentinel-1 SAR images. The technical solution of this invention will be further described below with reference to the accompanying drawings and embodiments.

[0034] like Figure 2 As shown, the process of this embodiment of the invention includes the following steps:

[0035] Step 1: For the target area, acquire Sentinel-1 SAR images with a high signal-to-noise ratio, including VV and VH polarimetric images. Then, use the SNAP tool provided by the European Space Agency to perform a series of data preprocessing on the SAR images, including radiometric calibration, filtering, terrain correction, and decibel conversion.

[0036] Step 2: In order to highlight the water body information in the SAR image, the Sentinel-1 dual polarization water body index SDWI is calculated using the preprocessed Sentinel-1 SAR VV and VH polarization images, where SDWI = log(10×VV×VH).

[0037] Step 3: Treat the VV polarization image, VH polarization image, and water index SDWI image as the R, G, and B bands of the color image respectively, and perform pseudo-color synthesis to obtain pseudo-color image I.

[0038] Step 4: Perform a color space transformation on the pseudo-color image I generated in Step 3, converting it from the CIE RGB color space to the CIE XYZ color space. The color space conversion formula is as follows:

[0039]

[0040]

[0041] Pixels (m) in pseudo-color image I in RGB color space r ,m g ,m b After conversion using formulas (1) and (2), it can be represented in the CIEXYZ color space as (m x ,m y ,m z ).

[0042] The CIE XYZ color space can be used as follows: Figure 1 The horseshoe-shaped area shown is represented.

[0043] Step 5: Extract the dominant wavelength information of the image. Assume point N is a pixel in the original pseudo-color image I obtained through color space transformation. Connect point E (equal-energy white point) and point N, and extend the line along the direction from point E to point N, intersecting with the horseshoe curve (i.e.,... Figure 1 The outer contour lines of the horseshoe-shaped region shown intersect at point M (denoted as (M)). x M y Therefore, the wavelength corresponding to the intersection point M is the dominant wavelength corresponding to point N. Obtaining the dominant wavelength information of all pixels in the image in the CIE XYZ color space collectively constitutes a new feature of the image.

[0044] Step 6: Calculate the length of line segment EM. All points on line segment EM have the same dominant wavelength, but their colors differ in the proportion of the spectral color corresponding to the dominant wavelength. The length of line segment EM is calculated as a new feature of the image to enhance the similarity between pixels of the same color family.

[0045] The length of line segment EM is measured using Euclidean distance, calculated as follows:

[0046]

[0047] In the formula, (M x M y ) indicates that point M is in such a position. Figure 1 The coordinates of the horseshoe-shaped region shown are given; point E is an isoenergetic white point with known coordinates (1 / 3, 1 / 3).

[0048] Step 7: Calculate the length ratio ρ of EN and EM. From a color science perspective, the length ratio ρ of EN and EM reflects the saturation information of the corresponding color. Therefore, the ρ value corresponding to all pixels can be calculated and used as a new feature of the image.

[0049] The lengths of line segments EN and EM are both measured using Euclidean distance, and the ratio ρ of the lengths of line segments EN and EM is calculated as follows:

[0050]

[0051] Step 8: When using Sentinel-1 SAR imagery to extract water bodies in a certain area, use the features from steps 5-7 as input features for the water body extraction model to improve the accuracy of water body extraction.

[0052] This invention addresses the insufficient utilization of color features in current SAR image water extraction methods, which still require improvement in water extraction accuracy. Therefore, it proposes a pseudo-color feature extraction method to enhance water information extraction from Sentinel-1 SAR images. This method requires no manual intervention, enabling automated extraction of color information from Sentinel-1 SAR images. Furthermore, the constructed color features can effectively distinguish between water and non-water bodies in the image, thus improving the accuracy and reliability of water extraction from Sentinel-1 SAR images. This method is not only computationally efficient and highly operable but also highly automated and scalable, making it highly valuable for practical applications.

[0053] It should be understood that any parts not described in detail in this specification belong to the prior art.

[0054] It should be understood that the above description of the preferred embodiments is quite detailed, but it should not be considered as a limitation on the scope of protection of this invention. Those skilled in the art, under the guidance of this invention, can make substitutions or modifications without departing from the scope of protection of the claims of this invention, and all such substitutions or modifications fall within the scope of protection of this invention. The scope of protection of this invention should be determined by the appended claims.

Claims

1. A pseudo-color feature extraction method to enhance water body extraction in Sentinel-1 SAR images, characterized in that, Includes the following steps: Step 1: Acquire Sentinel-1 SAR images of the target area, including VV and VH polarimetric images, and preprocess the SAR images. Step 2: Calculate the Sentinel-1 dual polarization water index SDWI using the preprocessed Sentinel-1 SAR VV and VH polarization images; Step 3: Treat the VV polarization image, VH polarization image and water index SDWI image as the R, G and B bands of the color image respectively, and perform pseudo-color synthesis to obtain pseudo-color image I; Step 4: Perform color space transformation on the pseudo-color image I generated in Step 3, converting it from the CIE RGB color space to the CIE XYZ color space. The CIE XYZ color space is a horseshoe-shaped region. Step 5: Extract the dominant wavelength information of the image; Suppose that point N is a pixel in the original pseudo-color image I obtained by color space transformation. Connect point E, the point of equal whiteness, with point N, and extend the line from point E to point N. The line intersects the horseshoe curve at point M. The horseshoe curve is the outer contour line of the horseshoe region. Then, the wavelength corresponding to the intersection point M is the dominant wavelength corresponding to point N. Obtain the dominant wavelength information of all pixels in the image in the CIE XYZ color space, which together constitute a new feature of the image. Step 6: Calculate the length of line segment EM as a new feature of the image to enhance the similarity between pixels of the same color. Step 7: Calculate the ratio ρ of the lengths of EN and EM as a new feature of the image; Step 8: When using Sentinel-1SAR imagery to perform water body extraction in a certain area, use the features from steps 5-7 as input features for the water body extraction model to improve the accuracy of water body extraction.

2. The pseudo-color feature extraction method for enhancing water body extraction in Sentinel-1 SAR images as described in claim 1, characterized in that: The preprocessing described in step 1 includes radiometric calibration, filtering, topographic correction, and decibel conversion of the SAR image.

3. The pseudo-color feature extraction method for enhancing water body extraction in Sentinel-1 SAR images as described in claim 1, characterized in that: The dual polarization water index SDWI = log(10 × VV × VH).

4. The pseudo-color feature extraction method for enhancing water body extraction in Sentinel-1 SAR images as described in claim 1, characterized in that: The conversion formula for pseudo-color image I from CIE RGB color space to CIE XYZ color space in step 4 is as follows: Pixels (m) in pseudo-color image I in RGB color space r ,m g ,m b After conversion using formulas (1) and (2), it can be expressed in the CIE XYZ color space as (m x ,m y ,m z ).

5. The pseudo-color feature extraction method for enhancing water body extraction in Sentinel-1 SAR images as described in claim 1, characterized in that: In step 6, the length of line segment EM is measured using Euclidean distance, calculated as follows: In the formula, (M x M y () represents the coordinates of point M in the horseshoe-shaped region; point E is an isoenergetic white point with known coordinates (1 / 3, 1 / 3).

6. The pseudo-color feature extraction method for enhancing water body extraction in Sentinel-1 SAR images as described in claim 1, characterized in that: In step 7, the lengths of line segments EN and EM are both measured using Euclidean distance, and the ratio ρ of the lengths of line segments EN and EM is calculated as follows:

Images