A fast method for extracting bare land based on high-resolution satellite data

Abstract

The invention relates to the field of satellite remote sensing data technology, in particular to a method for quickly extracting bare land based on high-resolution satellite data, comprising the following steps: S1, establishing a database, collecting image data of bare land through a network, storing relevant data in an established training set, constructing an image data conversion model in the database, inputting the image data in the training set into the image data conversion model, processing the image data conversion model to obtain the radiation brightness value of each pixel point in the bare land image data, and forming a radiation brightness image of the bare land. The invention can effectively solve the problem that currently monitoring personnel generally extract bare land from satellite image data by naked eye recognition, and as the resolution of remote sensing images continues to increase, the information in satellite image data becomes more complex, and the method of extracting bare land by human eye recognition has low work efficiency and takes a long time.

Description

Technical Field

[0001] The invention relates to the technical field of satellite remote sensing data, and in particular to a method for quickly extracting bare land based on high-resolution satellite data. Background Art

[0002] Remote sensing technology is a general term for a comprehensive technical system for observing the earth and celestial bodies from the ground to space. It can obtain satellite data from remote sensing technology platforms, and receive, process and analyze information through remote sensing instruments. Remote sensing technology is a rapidly developing high-tech technology. The information network it has formed is constantly providing people with a large amount of scientific data and dynamic information.

[0003] With the development of the economy, global environmental damage is becoming increasingly serious. Problems such as droughts and floods, illegal occupation of agricultural land, and deforestation frequently occur, seriously affecting people's lives, socio-economic development, and social stability. The use of remote sensing information to quickly, scientifically, and accurately classify and evaluate land cover, timely grasp the regional land cover situation, and take corresponding measures in a timely manner plays an important role in the rational planning and protection of land resources. Remote sensing monitoring of land cover is mostly carried out through monitoring of land cover type, area, and changes. If you want to know the changes in land cover area, you must rely on land cover classification.

[0004] At present, monitoring personnel generally extract exposed land from satellite image data through naked eye identification. As the resolution of remote sensing images continues to increase, the information in satellite image data becomes more complex. The method of extracting exposed land with human eyes has reduced work efficiency and is time-consuming.

[0005] In summary, the present invention solves the existing problems by designing a method for rapid extraction of bare land based on high-resolution satellite data. Summary of the invention

[0006] The purpose of the present invention is to provide a method for quickly extracting bare land based on high-resolution satellite data, so as to solve the problem that currently monitoring personnel generally extract bare land from satellite image data by naked eye recognition. As the resolution of remote sensing images continues to increase, the information in satellite image data becomes more complex, and the method of extracting bare land by human eye recognition has low work efficiency and is time-consuming.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A method for rapidly extracting bare land based on high-resolution satellite data comprises the following steps:

[0009] S1, establish a database, collect image data of bare land through the network, and store relevant data in the established training set, build an image data conversion model in the database, input the image data in the training set into the image data conversion model, the image data conversion model processes to obtain the radiation brightness value of each pixel in the bare land image data, and composes the radiation brightness image of the bare land, and calculates the radiation brightness threshold of the bare land according to the threshold calculation formula;

[0010] S2, inputting the image data taken by the high-resolution satellite into the image data conversion model, the image data conversion model processes to obtain the radiance value of each pixel in the image data taken by the high-resolution satellite, and composes a first radiance image;

[0011] S3, comparing all the radiation brightness values ​​in the first radiation brightness image with the radiation brightness threshold in order from left to right and from top to bottom, removing the radiation brightness values ​​in the first radiation brightness image that are greater than the radiation brightness threshold, thereby obtaining a second radiation brightness image;

[0012] S4, overlapping the second radiance image with the image data taken by a high-resolution satellite, extracting the portion of the image data taken by the high-resolution satellite that overlaps with the second radiance image, and obtaining an image of bare land in the image data taken by the high-resolution satellite.

[0013] Preferably, the specific processing steps of the image data conversion model in S1 are:

[0014] S11, using P(r N ,g N ,b N ) represents any pixel in the image, using the grayscale function H N (P) scanning each pixel in the image data in order from left to right and from top to bottom to obtain the gray value of each pixel, and forming a gray value image in the order of scanning;

[0015] S12, using the radiation brightness conversion formula La N The gray value representing each pixel in the gray value image is scanned in sequence from left to right and from top to bottom, so as to obtain the radiation brightness value of each pixel, and an initial radiation brightness image is formed in the scanning order;

[0016] S13, using Gaussian filtering to perform noise reduction processing on the initial radiometric brightness image to obtain a first radiometric brightness image of the input image data.

[0017] Preferably, the grayscale function H in S11 N (P) is:

[0018] HN (P) = 0.3r N +0.59g N +0.11b N

[0019] where r N Represents the value of the R component of the pixel, g N represents the value of the G component of the pixel, b N Represents the value of the B component of the pixel, N∈(1,2,3,...,N).

[0020] Preferably, the radiation brightness conversion formula La in S12 is N for:

[0021] La N =Gain×H N (P)+Bias

[0022] Wherein Gain is the gain value of the image data, Bias is the offset of the image data, and N∈(1,2,3,...,N).

[0023] Preferably, the specific steps of the noise reduction process in S13 are:

[0024] S21, construct a Gaussian template U(x, y) of size (2k+1)×(2k+1), the calculation formula of U(x, y) is:

[0025] Wherein, k is a positive integer and k≥1, x, y represent the horizontal coordinate and vertical coordinate of the element in the Gaussian template respectively, σ>0, is the smoothness parameter of the first radiation brightness image;

[0026] S22, take k=1, the template values ​​of the 9 elements in the Gaussian template U(x, y) of size 3×3 are M1, M2, M3, M4, M5, M6, M7, M8, M9 respectively, the 8 neighboring pixels of any pixel except the four edges with a width of one pixel on the left, top, right and bottom in the initial radiation brightness image, that is, the pixels in the eight directions of the upper left, upper right, left, right, lower left, bottom and lower right adjacent to the pixel, the radiation brightness values ​​of the pixel and the 8 neighboring pixels are H1, H2, H3, H4, H5, H6, H7, H8, H9 in the order from left to right and from top to bottom in the first grayscale image, use the Gaussian template U(x, y) to scan each pixel except the four edges with a width of one pixel on the left, top, right and bottom in the first grayscale image in the order from left to right and from top to bottom, and use the Gaussian template U(x, y) to perform Gaussian filtering on each pixel;

[0027] S23, in order from left to right and from top to bottom, use the Gaussian template U(x, y) to perform Gaussian filtering on each pixel point, and replace the radiation brightness of the current pixel point P with the weighted average radiation brightness value of all pixels in the neighborhood determined by the Gaussian template U(x, y), and obtain the first radiation brightness image after the scanning is completed.

[0028] Preferably, the calculation formula for the radiation brightness threshold of the bare land in S1 is:

[0029]

[0030] n=a1+a2+a3+...+a n

[0031] where a n Indicates that the radiation brightness value is H n (P) is the number of occurrences, and n∈(1,2,3,...,n).

[0032] Preferably, the calculation formula for Gaussian filtering of each pixel point by Gaussian template U(x, y) in S22 is:

[0033]

[0034]

[0035] Where U0 represents the radiation brightness of the pixel after Gaussian filtering with Gaussian template U(x, y), and S is an intermediate variable, which represents the sum of the radiation brightness values ​​of pixel P and its 8 neighboring pixels.

[0036] Preferably, the radiance conversion formula La N The specific values ​​of Gain and Bias can be obtained from the header file of the image data using ENVI software, and the header file is located in the original data of the image data.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] 1. In the present invention, by designing an image data conversion model, the image data conversion model can convert high-resolution satellite image data into a radiation brightness image. According to the difference between the radiation brightness of bare land and that of other objects, the calculated radiation brightness threshold of the bare land is used to remove the non-conforming radiation brightness values ​​in the converted radiation brightness image, thereby extracting the image data related to the bare land in the satellite image data, with fast response speed and high work efficiency.

[0039] 2. In the present invention, a Gaussian template is designed to perform Gaussian filtering to reduce noise on the output radiation brightness image, thereby removing interference values ​​in the image and improving the accuracy of the extraction result.

[0040] 3. In the present invention, a weighted calculation method is designed to calculate the radiation threshold of bare land. The calculated radiation threshold has high accuracy, which increases the reliability of the extracted results. DETAILED DESCRIPTION

[0041] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0042] The present invention provides a technical solution:

[0043] A method for rapidly extracting bare land based on high-resolution satellite data comprises the following steps:

[0044] S1, establish a database, collect image data of bare land through the network, and store relevant data in the established training set, build an image data conversion model in the database, input the image data in the training set into the image data conversion model, the image data conversion model processes to obtain the radiation brightness value of each pixel in the bare land image data, and composes the radiation brightness image of the bare land, and calculates the radiation brightness threshold of the bare land according to the threshold calculation formula;

[0045] S2, inputting the image data taken by the high-resolution satellite into the image data conversion model, the image data conversion model processes to obtain the radiance value of each pixel in the image data taken by the high-resolution satellite, and composes a first radiance image;

[0046] S3, comparing all the radiation brightness values ​​in the first radiation brightness image with the radiation brightness threshold in order from left to right and from top to bottom, removing the radiation brightness values ​​in the first radiation brightness image that are greater than the radiation brightness threshold, thereby obtaining a second radiation brightness image;

[0047] S4, overlapping the second radiance image with the image data taken by a high-resolution satellite, extracting the portion of the image data taken by the high-resolution satellite that overlaps with the second radiance image, and obtaining an image of bare land in the image data taken by the high-resolution satellite.

[0048] Furthermore, the specific processing steps of the image data conversion model in S1 are:

[0049] S11, using P(r N ,g N ,b N ) represents any pixel in the image, using the grayscale function H N (P) scanning each pixel in the image data in order from left to right and from top to bottom to obtain the gray value of each pixel, and forming a gray value image in the order of scanning;

[0050] S12, using the radiation brightness conversion formula La N The gray value representing each pixel in the gray value image is scanned in sequence from left to right and from top to bottom, so as to obtain the radiation brightness value of each pixel, and an initial radiation brightness image is formed in the scanning order;

[0051] S13, using Gaussian filtering to perform noise reduction processing on the initial radiometric brightness image to obtain a first radiometric brightness image of the input image data.

[0052] Furthermore, the grayscale function H in S11 N (P) is:

[0053] H N (P) = 0.3r N +0.59g N +0.11b N

[0054] where r N Represents the value of the R component of the pixel, g N represents the value of the G component of the pixel, b N Represents the value of the B component of the pixel, N∈(1,2,3,...,N).

[0055] Furthermore, the radiation brightness conversion formula La in S12 is N for:

[0056] La N =Gain×H N (P)+Bias

[0057] Wherein Gain is the gain value of the image data, Bias is the offset of the image data, and N∈(1,2,3,...,N).

[0058] Furthermore, the specific steps of the noise reduction process in S13 are:

[0059] S21, construct a Gaussian template U(x, y) of size (2k+1)×(2k+1), the calculation formula of U(x, y) is:

[0060] Wherein, k is a positive integer and k≥1, x, y represent the horizontal coordinate and vertical coordinate of the element in the Gaussian template respectively, σ>0, is the smoothness parameter of the first radiation brightness image;

[0061] S22, take k=1, the template values ​​of the 9 elements in the Gaussian template U(x, y) of size 3×3 are M1, M2, M3, M4, M5, M6, M7, M8, M9 respectively, the 8 neighboring pixels of any pixel except the four edges with a width of one pixel on the left, top, right and bottom in the initial radiation brightness image, that is, the pixels in the eight directions of the upper left, upper right, left, right, lower left, bottom and lower right adjacent to the pixel, the radiation brightness values ​​of the pixel and the 8 neighboring pixels are H1, H2, H3, H4, H5, H6, H7, H8, H9 in the order from left to right and from top to bottom in the first grayscale image, use the Gaussian template U(x, y) to scan each pixel except the four edges with a width of one pixel on the left, top, right and bottom in the first grayscale image in the order from left to right and from top to bottom, and use the Gaussian template U(x, y) to perform Gaussian filtering on each pixel;

[0062] S23, in order from left to right and from top to bottom, use the Gaussian template U(x, y) to perform Gaussian filtering on each pixel point, and replace the radiation brightness of the current pixel point P with the weighted average radiation brightness value of all pixels in the neighborhood determined by the Gaussian template U(x, y), and obtain the first radiation brightness image after the scanning is completed.

[0063] Furthermore, the calculation formula for the radiation brightness threshold of the bare land in S1 is:

[0064]

[0065] n=a1+a2+a3+...+a n

[0066] where a n Indicates that the radiation brightness value is H n (P) is the number of occurrences, and n∈(1,2,3,...,n).

[0067] Furthermore, the calculation formula for Gaussian filtering of each pixel point by Gaussian template U(x, y) in S22 is:

[0068]

[0069]

[0070] Where U0 represents the radiation brightness of the pixel after Gaussian filtering with Gaussian template U(x, y), and S is an intermediate variable, which represents the sum of the radiation brightness values ​​of pixel P and its 8 neighboring pixels.

[0071] Furthermore, the radiation brightness conversion formula La N The specific values ​​of Gain and Bias can be obtained from the header file of the image data using ENVI software, and the header file is located in the original data of the image data.

[0072] Specific implementation cases

[0073] Establish a database, collect image data of bare land through the network, and store the relevant data in the established training set. Build an image data conversion model in the database, input the image data in the training set into the image data conversion model, and process the image data conversion model to obtain the radiation brightness value of each pixel in the bare land image data, and form a radiation brightness image of the bare land. According to the threshold calculation formula, calculate the radiation brightness threshold of the bare land. The calculation formula of the radiation brightness threshold of the bare land is:

[0074]

[0075] n=a1+a2+a3+...+a n

[0076] where a n Indicates that the radiation brightness value is H n (P) is the number of occurrences, and n∈(1,2,3,...,n);

[0077] The image data taken by high-resolution satellites are input into the image data conversion model, and P(r N ,g N ,b N ) represents any pixel in the image, using the grayscale function H N (P) Scan each pixel in the image data in order from left to right and from top to bottom to obtain the grayscale value of each pixel, and form a grayscale value image in the order of scanning. The grayscale function H N (P) is: H N (P) = 0.3r N +0.59g N +0.11b N , where r N Represents the value of the R component of the pixel, g N represents the value of the G component of the pixel, b N Represents the value of the B component of the pixel, N∈(1,2,3,...,N), using the radiation brightness conversion formula La NScan the grayscale value representing each pixel in the grayscale image in order from left to right and from top to bottom to obtain the radiation brightness value of each pixel. The initial radiation brightness image is formed in the order of scanning. The radiation brightness conversion formula La N For: La N =Gain×H N (P)+Bias, where Gain is the gain value of the image data, Bias is the offset of the image data, N∈(1,2,3,...,N), and the radiation brightness conversion formula La N The specific values ​​of Gain and Bias in can be obtained from the header file of the image data using ENVI software, and the header file is located in the original data of the image data. The initial radiation brightness image is denoised using Gaussian filtering to construct a Gaussian template U(x, y) of size (2k+1)×(2k+1). The calculation formula of U(x, y) is:

[0078]

[0079] Wherein, k is a positive integer and k≥1, x, y represent the abscissa and ordinate of the elements in the Gaussian template respectively, σ>0, is the smoothness parameter of the first radiation brightness image, and the first radiation brightness image of the input image data is obtained. Take k=1, and the template values ​​of the 9 elements in the Gaussian template U(x, y) with a size of 3×3 are M1, M2, M3, M4, M5, M6, M7, M8, M9 respectively. In the initial radiation brightness image, except for the four edges with a width of one pixel on the left, top, right, and bottom, the 8 neighboring pixels of any pixel point, namely, the upper left, upper, upper right, left, right, lower left, The radiation brightness values ​​of the pixels in the eight directions of the bottom, bottom and right are H1, H2, H3, H4, H5, H6, H7, H8 and H9 respectively in the order from left to right and from top to bottom in the first grayscale image. The Gaussian template U(x, y) is used to scan each pixel except the four edges with a width of one pixel on the left, top, right and bottom in the first grayscale image from left to right and from top to bottom. The Gaussian template U(x, y) is used to perform Gaussian filtering on each pixel. The calculation formula for Gaussian filtering of each pixel by the Gaussian template U(x, y) is as follows:

[0080]

[0081]

[0082] Where U0 represents the radiant brightness of the pixel after Gaussian filtering with Gaussian template U(x, y), S is an intermediate variable, which represents the sum of the radiant brightness values ​​of pixel P and its 8 neighboring pixels. From left to right and from top to bottom, Gaussian template U(x, y) is used to perform Gaussian filtering on each pixel. The weighted average radiant brightness value of all pixels in the neighborhood determined by Gaussian template U(x, y) replaces the radiant brightness of the current pixel P. After the scanning is completed, the first radiant brightness image is obtained;

[0083] All the radiation brightness values ​​in the first radiation brightness image are compared with the radiation brightness threshold in order from left to right and from top to bottom, and the radiation brightness values ​​in the first radiation brightness image that are greater than the radiation brightness threshold are removed to obtain a second radiation brightness image. The second radiation brightness image is overlapped with the image data taken by a high-resolution satellite, and the portion of the image data taken by the high-resolution satellite that overlaps with the second radiation brightness image is extracted to obtain the image of bare land in the image data taken by the high-resolution satellite.

[0084] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for rapid extraction of exposed land based on high-resolution satellite data, characterized in that: The following steps are involved: S1, establish a database, collect image data of bare land through the network, and store relevant data in the established training set, build an image data conversion model in the database, input the image data in the training set into the image data conversion model, the image data conversion model processes to obtain the radiation brightness value of each pixel in the bare land image data, and composes the radiation brightness image of the bare land, and calculates the radiation brightness threshold of the bare land according to the threshold calculation formula; S2, inputting the image data taken by the high-resolution satellite into the image data conversion model, the image data conversion model processes to obtain the radiance value of each pixel in the image data taken by the high-resolution satellite, and composes a first radiance image; S3, comparing all the radiation brightness values ​​in the first radiation brightness image with the radiation brightness threshold in order from left to right and from top to bottom, removing the radiation brightness values ​​in the first radiation brightness image that are greater than the radiation brightness threshold, thereby obtaining a second radiation brightness image; S4, overlapping the second radiance image with the image data taken by a high-resolution satellite, extracting the portion of the image data taken by the high-resolution satellite that overlaps with the second radiance image, and obtaining an image of bare land in the image data taken by the high-resolution satellite.

2. The method for rapid extraction of bare land based on high-resolution satellite data according to claim 1, characterized in that: The specific processing steps of the image data conversion model in S1 are: S11, using P(r N ,g N ,b N ) represents any pixel in the image, using the grayscale function H N (P) scanning each pixel in the image data in order from left to right and from top to bottom to obtain the gray value of each pixel, and forming a gray value image in the order of scanning; S12, using the radiation brightness conversion formula La N The gray value representing each pixel in the gray value image is scanned in sequence from left to right and from top to bottom, so as to obtain the radiation brightness value of each pixel, and an initial radiation brightness image is formed in the scanning order; S13, using Gaussian filtering to perform noise reduction processing on the initial radiometric brightness image to obtain a first radiometric brightness image of the input image data.

3. The method for rapid extraction of bare land based on high-resolution satellite data according to claim 2, characterized in that: The grayscale function H in S11 N (P) is: H N (P)=0.3r N +0.59g N +0.11b N where r N Represents the value of the R component of the pixel, g N represents the value of the G component of the pixel, b N Represents the value of the B component of the pixel, N∈(1,2,3,...,N).

4. The method for rapid extraction of bare land based on high-resolution satellite data according to claim 2, characterized in that: The radiation brightness conversion formula La in S12 N for: The N =Gain×H N (P)+Bias Wherein Gain is the gain value of the image data, Bias is the offset of the image data, and N∈(1,2,3,...,N).

5. The method for rapid extraction of bare land based on high-resolution satellite data according to claim 2, characterized in that: The specific steps of the noise reduction process in S13 are: S21, construct a Gaussian template U(x, y) of size (2k+1)×(2k+1), the calculation formula of U(x, y) is: Wherein, k is a positive integer and k≥1, x, y represent the horizontal coordinate and vertical coordinate of the element in the Gaussian template respectively, σ>0, is the smoothness parameter of the first radiation brightness image; S22, take k=1, the template values ​​of the 9 elements in the Gaussian template U(x, y) of size 3×3 are M1, M2, M3, M4, M5, M6, M7, M8, M9 respectively, the 8 neighboring pixels of any pixel except the four edges with a width of one pixel on the left, top, right and bottom in the initial radiation brightness image, that is, the pixels in the eight directions of the upper left, upper right, left, right, lower left, lower and lower right adjacent to the pixel, the radiation brightness values ​​of the pixel and the 8 neighboring pixels are H1, H2, H3, H4, H5, H6, H7, H8, H9 in the order from left to right and from top to bottom in the first grayscale image, use the Gaussian template U(x, y) to scan each pixel except the four edges with a width of one pixel on the left, top, right and bottom in the first grayscale image in the order from left to right and from top to bottom, and use the Gaussian template U(x, y) to perform Gaussian filtering on each pixel; S23, in order from left to right and from top to bottom, use the Gaussian template U(x, y) to perform Gaussian filtering on each pixel point, and replace the radiation brightness of the current pixel point P with the weighted average radiation brightness value of all pixels in the neighborhood determined by the Gaussian template U(x, y), and obtain the first radiation brightness image after the scanning is completed.

6. The method for rapid extraction of bare land based on high-resolution satellite data according to claim 1, characterized in that: The calculation formula for the radiation brightness threshold of the bare land in S1 is: <h2 style=";text-align:left;direction:ltr">n = a1 + a2 + a3 +... + a<h2 style=";text-align:left;direction:ltr"> n where a n Indicates that the radiation brightness value is H n (P) is the number of occurrences, and n∈(1,2,3,...,n).

7. The method for rapid extraction of bare land based on high-resolution satellite data according to claim 5, characterized in that: The calculation formula for Gaussian filtering of each pixel point by Gaussian template U(x, y) in S22 is: Where U0 represents the radiation brightness of the pixel after Gaussian filtering with Gaussian template U(x, y), and S is an intermediate variable, which represents the sum of the radiation brightness values ​​of pixel P and its 8 neighboring pixels.

8. The method for rapid extraction of bare land based on high-resolution satellite data according to claim 4, characterized in that: The radiation brightness conversion formula La N The specific values ​​of Gain and Bias in are obtained from the header file of the image data using ENVI software, and the header file is located in the original data of the image data.

Images