A mangrove recognition method based on super-resolution images

Abstract

The present invention discloses a mangrove identification method based on super-resolution images, belonging to the field of mangrove identification technology. The method comprises the following steps: S1, performing bicubic interpolation processing on the forest image to be identified to obtain a super-resolution forest image; S2, extracting a detection data set corresponding to the super-resolution forest image, and generating updated pixel values ​​for each pixel in the detection data set based on the deviation coefficient of each pixel; S3, screening the mangrove area in the forest image to be identified based on the updated pixel values ​​corresponding to each pixel. The present invention gradually enhances the separability of mangrove features, improves the accuracy of mangrove ecological identification, and provides key technical support for mangrove protection and ecological restoration.

Description

Technical Field

[0001] The invention belongs to the technical field of mangrove identification, and in particular relates to a mangrove identification method based on super-resolution images. Background Art

[0002] Mangroves, as unique vegetation communities found in tropical and subtropical coastal zones, perform important ecological functions, including maintaining biodiversity, providing wind and wave protection, and carbon sequestration and storage. Strengthening mangrove protection and restoration, and accurately identifying their spatial distribution and health status, are key requirements in ecological monitoring. While traditional imaging techniques have been applied to mangrove identification, existing methods lack accuracy and detail due to limitations in sensor resolution, complex environmental interference, and the mixed species nature of mangroves.

[0003] Furthermore, mangroves grow in intertidal zones, and their color characteristics are easily affected by soil salinity, tidal levels, and surrounding vegetation. Furthermore, different mangrove species (such as Kandelia candel, Tung blossom trees, and Avicennia marina) often form mixed stands, making it difficult for traditional image-based classification methods to accurately extract species distribution and effectively identify mangrove areas from high-resolution imagery. Summary of the Invention

[0004] In order to solve the above problems, the present invention proposes a mangrove recognition method based on super-resolution images.

[0005] The technical solution of the present invention is: a mangrove identification method based on super-resolution images comprises the following steps:

[0006] S1, performing bicubic interpolation processing on the forest image to be identified to obtain a super-resolution forest image;

[0007] S2. extracting a detection data set corresponding to the super-resolution forest image, and generating an updated pixel value for each pixel point based on a deviation coefficient of each pixel point in the detection data set;

[0008] S3. Filter the mangrove area of ​​the forest image to be identified according to the updated pixel value corresponding to each pixel point.

[0009] Furthermore, S2 includes the following sub-steps:

[0010] S21, taking the pixel values ​​of all pixels in the super-resolution forest image as a detection data set;

[0011] S22. Determine the deviation coefficient of each pixel point based on the detection data set;

[0012] S23. Determine an updated pixel value based on the deviation coefficient of each pixel point.

[0013] The beneficial effect of the above further scheme is: in the present invention, the deviation coefficient of each pixel point is extracted, which effectively quantifies the degree of outlier of each pixel point relative to its set nearest neighbor pixel point, and helps to identify pixel points that are significantly different from the surrounding pixels. These pixel points often correspond to the edges of mangroves or special texture areas.

[0014] Furthermore, S22 includes the following sub-steps:

[0015] S221, extracting the outlier degree of each pixel in the detection data set using a local outlier factor algorithm based on the pixel value of the pixel;

[0016] S222. Determine the deviation coefficient of each pixel point according to the outlier degree of each pixel point in the detection data set.

[0017] The beneficial effect of the above further solution is that, in the present invention, when using the local outlier factor algorithm to extract the outlier degree of each pixel in a data set, the neighborhood must first be defined: the k nearest neighbors are selected for each point (the k value must be pre-set); the local reachable density is then calculated: the local density of the point is quantified based on the distance from the point to other points in the neighborhood; and finally, the outlier degree is obtained by comparing the density of the target point with that of its neighbors. The extracted outlier degree is converted into a specific deviation coefficient, achieving the conversion from outlier degree to adjustment parameter. This allows subsequent pixel value updates to be based on the actual outlier status of the pixel point, improving the targetedness and effectiveness of the adjustment.

[0018] Furthermore, in S222, the pixel deviation coefficient The calculation formula is:

[0019] ;

[0020] Where, Indicates the outlier degree of the pixel in the detection data set, Represents the standard deviation of the outliers of all pixels in the detection data set, Indicates the number of nearest neighbors set for a pixel by the local outlier factor algorithm, It represents the maximum number of nearest neighbors set by the local outlier factor algorithm for all pixels. Represents a very small number.

[0021] Furthermore, S23 includes the following sub-steps:

[0022] S231, matching an attracting pixel point for each pixel point according to the deviation coefficient and pixel value of each pixel point;

[0023] S232: Determine an updated pixel value for each pixel point according to each pixel point and its corresponding attracted pixel point.

[0024] The beneficial effect of the above further solution is that, in the present invention, by comparing the deviation coefficient and pixel value of the pixel point, it is possible to accurately match the attracting pixel points. The deviation coefficient reflects the degree of outlier of the pixel point. Selecting the pixel point with the larger deviation coefficient as the candidate ensures that the attracting pixel point has significant outlier characteristics. Among the candidate pixels, further extracting the pixel point with the largest pixel value as the attracting pixel point helps to highlight the high-brightness or high-contrast areas, which often correspond to the edges of mangroves or areas with rich texture, effectively enhancing the characteristics of the mangrove area.

[0025] Further, in S231, The method of matching the pixel points with the attracting pixel points is: From the pixel point set with the deviation coefficient of each pixel point, the pixel point with the maximum pixel value is extracted as the The attracting pixels of pixels.

[0026] The beneficial effect of the above further solution is that in the present invention, when the segregation coefficient of a pixel point is the largest, the pixel point corresponding to the maximum pixel value of the remaining pixels is used as the adjacent pixel point to avoid update failure due to lack of attraction points.

[0027] Further, in S232, Update pixel value of each pixel The calculation formula is:

[0028] ;

[0029] Where, Indicates the The pixel value of each pixel, Indicates the The normalized distance between a pixel and its matching attracting pixel, represents the exponential function, Indicates the The deviation coefficient of each pixel, represents the random perturbation parameter, Indicates generating a random number between 0 and 1.

[0030] Furthermore, in S3, The pixel points are regarded as the mangrove area, where Indicates the The pixel value of each pixel, Indicates the Update pixel value of pixel points, Indicates the The sum of the pixel values ​​of the four neighboring pixels around the pixel.

[0031] The beneficial effect of the above further scheme is: in the present invention, the threshold on the right side of the constraint condition is dynamically adjusted with the pixel point itself, and the four-neighborhood summation operation implicitly contains a spatial continuity constraint, requiring the identified pixel to form a spatially continuous structure with the surrounding pixels.

[0032] The beneficial effects of the present invention are as follows: It interpolates the initial image to more accurately reconstruct image details; by calculating characteristic parameters such as the deviation coefficient and outlier degree of pixels in the super-resolution forest image, it selects pixels with larger deviation coefficients and higher pixel values ​​as attraction points, thereby determining updated pixel values ​​for the pixels based on the attraction points; and finally, by combining the color of the pixel's neighborhood, it extracts the mangrove area. This invention gradually enhances the separability of mangrove features, improves the accuracy of mangrove ecological identification, and provides key technical support for mangrove protection and ecological restoration. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 Flowchart of the mangrove recognition method based on super-resolution images. DETAILED DESCRIPTION

[0034] The embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0035] like Figure 1 As shown, the present invention provides a mangrove identification method based on super-resolution images, comprising the following steps:

[0036] S1, performing bicubic interpolation processing on the forest image to be identified to obtain a super-resolution forest image;

[0037] S2. extracting a detection data set corresponding to the super-resolution forest image, and generating an updated pixel value for each pixel point based on a deviation coefficient of each pixel point in the detection data set;

[0038] S3. Filter the mangrove area of ​​the forest image to be identified according to the updated pixel value corresponding to each pixel point.

[0039] In this embodiment of the present invention, S2 includes the following sub-steps:

[0040] S21, taking the pixel values ​​of all pixels in the super-resolution forest image as a detection data set;

[0041] S22. Determine the deviation coefficient of each pixel point based on the detection data set;

[0042] S23. Determine an updated pixel value based on the deviation coefficient of each pixel point.

[0043] In the present invention, the deviation coefficient of each pixel point is extracted, which effectively quantifies the degree of outlier of each pixel point relative to its set nearest neighbor pixel point, and helps to identify pixels that are significantly different from the surrounding pixels. These pixels often correspond to the edges of mangroves or special texture areas.

[0044] In this embodiment of the present invention, S22 includes the following sub-steps:

[0045] S221, extracting the outlier degree of each pixel in the detection data set using a local outlier factor algorithm based on the pixel value of the pixel;

[0046] S222. Determine the deviation coefficient of each pixel point according to the outlier degree of each pixel point in the detection data set.

[0047] In this paper, the local outlier factor algorithm is used to extract the outlier degree of each pixel in a data set. The algorithm first defines a neighborhood: for each point, its k nearest neighbors are selected (the k value must be pre-set). The local reachable density is then calculated: the local density of the point is quantified based on its distance from other points in the neighborhood. Finally, the outlier degree is determined by comparing the density of the target point with that of its neighbors. The extracted outlier degree is converted into a specific deviation coefficient, achieving a conversion from outlier degree to adjustment parameter. This allows subsequent pixel value updates to be based on the actual outlier status of the pixel, improving the relevance and effectiveness of the adjustments.

[0048] In the embodiment of the present invention, in S222, the deviation coefficient of the pixel point The calculation formula is:

[0049] ;

[0050] Where, Indicates the outlier degree of the pixel in the detection data set, Represents the standard deviation of the outliers of all pixels in the detection data set, Indicates the number of nearest neighbors set for a pixel by the local outlier factor algorithm, It represents the maximum number of nearest neighbors set by the local outlier factor algorithm for all pixels. Represents a very small number.

[0051] In this embodiment of the present invention, S23 includes the following sub-steps:

[0052] S231, matching an attracting pixel point for each pixel point according to the deviation coefficient and pixel value of each pixel point;

[0053] S232: Determine an updated pixel value for each pixel point according to each pixel point and its corresponding attracted pixel point.

[0054] In this method, by comparing the deviation coefficients and pixel values ​​of pixels, we can accurately match attracting pixels. The deviation coefficient reflects the degree of outlier of a pixel. Selecting pixels with larger deviation coefficients as candidates ensures that the attracting pixels have significant outlier characteristics. Among the candidate pixels, further extracting the pixels with the largest pixel values ​​as attracting pixels helps to highlight areas of high brightness or high contrast, which often correspond to the edges of mangroves or areas with rich texture, effectively enhancing the characteristics of the mangrove areas.

[0055] In the embodiment of the present invention, in S231, The method of matching the pixel points with the attracting pixel points is: From the pixel point set with the deviation coefficient of each pixel point, the pixel point with the maximum pixel value is extracted as the The attracting pixels of pixels.

[0056] In the present invention, when the segregation coefficient of a pixel point is the largest, the pixel point corresponding to the maximum pixel value of the remaining pixels is used as the adjacent pixel point to avoid update failure due to lack of attraction points.

[0057] In the embodiment of the present invention, in S232, Update pixel value of each pixel The calculation formula is:

[0058] ;

[0059] Where, Indicates the The pixel value of each pixel, Indicates the The normalized distance between a pixel and its matching attracting pixel, represents the exponential function, Indicates the The deviation coefficient of each pixel, represents the random perturbation parameter, Indicates generating a random number between 0 and 1.

[0060] In the embodiment of the present invention, in S3, The pixel points are regarded as the mangrove area, where Indicates the The pixel value of each pixel, Indicates the Update pixel value of pixel points, Indicates the The sum of the pixel values ​​of the four neighboring pixels around the pixel.

[0061] In the present invention, the threshold on the right side of the constraint condition is dynamically adjusted with the pixel itself, and the four-neighborhood summation operation implicitly contains a spatial continuity constraint, requiring that the identified pixel must form a spatially continuous structure with the surrounding pixels.

[0062] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and it should be understood that the scope of protection of the present invention is not limited to such specific descriptions and embodiments. Those skilled in the art can make various other specific variations and combinations based on the technical teachings disclosed in the present invention without departing from the essence of the present invention, and such variations and combinations are still within the scope of protection of the present invention.

Claims

1. A mangrove identification method based on super-resolution images, characterized in that: The following steps are involved: S1, performing bicubic interpolation processing on the forest image to be identified to obtain a super-resolution forest image; S2. extracting a detection data set corresponding to the super-resolution forest image, and generating an updated pixel value for each pixel point based on a deviation coefficient of each pixel point in the detection data set; S3, screening the mangrove area of ​​the forest image to be identified according to the updated pixel value corresponding to each pixel point; The S2 includes the following sub-steps: S21, taking the pixel values ​​of all pixels in the super-resolution forest image as a detection data set; S22. Determine the deviation coefficient of each pixel point based on the detection data set; S23, determining an updated pixel value based on the deviation coefficient of each pixel point; The S22 includes the following sub-steps: S221, extracting the outlier degree of each pixel in the detection data set using a local outlier factor algorithm based on the pixel value of the pixel; S222, determining a deviation coefficient of each pixel point based on the outlier degree of each pixel point in the detection data set; In the above S222, the pixel deviation coefficient The calculation formula is: ； Where, Indicates the outlier degree of the pixel in the detection data set, Represents the standard deviation of the outliers of all pixels in the detection data set, Indicates the number of nearest neighbors set for a pixel by the local outlier factor algorithm, It represents the maximum number of nearest neighbors set by the local outlier factor algorithm for all pixels. Represents a very small number; The S23 includes the following sub-steps: S231, matching an attracting pixel point for each pixel point according to the deviation coefficient and pixel value of each pixel point; S232, determining an updated pixel value for each pixel point according to each pixel point and its corresponding attracted pixel point; In the above S232, Update pixel value of each pixel The calculation formula is: ； Where, Indicates the The pixel value of each pixel, Indicates the The normalized distance between a pixel and its matching attracting pixel, represents the exponential function, Indicates the The deviation coefficient of each pixel, represents the random perturbation parameter, Indicates generating a random number between 0 and 1; The S3 will satisfy The pixel points are regarded as the mangrove area, where Indicates the The pixel value of each pixel, Indicates the Update pixel value of pixel points, Indicates the The sum of the pixel values ​​of the four neighboring pixels around the pixel.

2. The method for identifying mangroves based on super-resolution images according to claim 1, characterized in that: In the S231, The method of matching the pixel points with the attracting pixel points is: From the pixel point set with the deviation coefficient of each pixel point, the pixel point with the maximum pixel value is extracted as the The attracting pixels of pixels.

Images