An eucalyptus extraction method based on SWIR difference enhancement and eucalyptus canopy characteristic factors

Abstract

The present application belongs to the field of remote sensing monitoring and forestry resource management, and discloses an eucalyptus extraction method based on SWIR (short wave infrared) difference enhancement and eucalyptus canopy characteristic factor. The method first constructs a SWIR difference enhancement factor, uses the difference between the two bands of the short wave infrared reflectance of the Sentinel-2 multispectral satellite to extract the short wave infrared characteristics of eucalyptus, and at the same time constructs an eucalyptus canopy characteristic factor, which strengthens the spectral difference between eucalyptus and other vegetation by weighted fusion of red edge slope and near infrared enhancement characteristics. Then, the two types of factors are subjected to spatial regularization and Fisher-entropy thresholding to generate a preliminary mask, and the core eucalyptus area is extracted by intersection operation; then, morphological opening and closing operations and small area filtering are combined to eliminate noise and non-target interference, and a high-reliability extraction mask is obtained. The present application realizes efficient extraction of eucalyptus by relying on the advantages of the Sentinel-2 multispectral data, and is suitable for forestry resource monitoring and eucalyptus extraction verification.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of remote sensing image processing and forestry resource monitoring, and particularly relates to an eucalyptus extraction method based on SWIR difference enhancement and eucalyptus crown layer characteristic factors. BACKGROUND

[0002] As a typical fast-growing plantation tree species, eucalyptus has important value in wood processing, paper industry and forestry economy. However, with the continuous expansion of planting scale, its large-scale felling activities have brought serious challenges to the sustainable use of forest resources and the protection of ecological environment. Therefore, realizing effective extraction of eucalyptus has become an important goal of researchers and management departments. Traditional supervision methods mainly rely on manual patrol and unmanned aerial vehicle survey. Such methods can obtain high-precision results in local areas, but when applied in large-scale forest land, they are difficult to meet the needs of dynamic monitoring due to the constraints of labor cost and operation efficiency. In contrast, eucalyptus extraction methods based on remote sensing images have become a research hotspot due to their wide coverage and high acquisition efficiency. However, the existing technology still has obvious deficiencies. On the one hand, methods based on a single spectral index have limited accuracy in distinguishing eucalyptus from other vegetation, and are easily disturbed by light conditions, seasonal differences and atmospheric effects, resulting in serious false detection and missed detection problems. On the other hand, although methods based on supervised classification or deep learning can improve recognition accuracy, they rely heavily on a large number of high-quality labeled samples and high-performance computing resources, have high processing complexity, and have insufficient generalization ability across regions and seasons, which restricts their practical application and promotion. In addition, the existing methods also have problems of insufficient feature selection and insufficient feature fusion in utilizing hyperspectral characteristics, which makes it difficult to stably guarantee the eucalyptus recognition effect.

[0003] Current research trends mainly focus on three types of methods. The detection method based on spectral index uses a single or a small number of spectral factors for vegetation extraction, which is simple to calculate, but it is difficult to accurately distinguish eucalyptus from other vegetation. The method based on statistics and classification can improve the recognition effect to a certain extent through machine learning or clustering algorithm, but it has high dependence on sample quantity and feature selection. The method based on deep learning uses convolutional neural network or semantic segmentation model to extract eucalyptus information from remote sensing data, which has high accuracy, but it needs a large number of labeled samples and computing power support, and it has difficulties in large-scale and low-cost applications. The patent (publication number CN118823579A, application number CN202410855855.6, invention name "Eucalyptus extraction method based on hyperspectral data and device, equipment and medium") disclosed by Zhuhai Orbiter Satellite Big Data Co., Ltd. proposes a eucalyptus extraction method based on hyperspectral data, which expands the eucalyptus recognition technology path. The patent technology uses "multi-period image + exclusive ESI index + farmland mask" design, which has a significant improvement over traditional methods in eucalyptus recognition scenes in farmland, but it still has application limitations: first, it has strong dependence on the waveband of hyperspectral data, and requires specific Band18 and Band26 wavebands of satellites. When using other satellites, the index formula needs to be re-adjusted due to differences in waveband settings, and the generalization ability across satellite platforms is insufficient. Second, the multi-period image processing and multi-threshold filtering increase the computational complexity, and the computing power requirement for data preprocessing is high. In large-scale regional applications, the data processing period is long. Third, although the multi-period image reduces some seasonal interference, the adaptability to extreme light conditions is still limited, and the recognition accuracy in such scenes is easily reduced.

[0004] In summary, the existing eucalyptus extraction technology has not broken through the "precision-efficiency-universality" triangle balance. Traditional single spectral methods are efficient but have low accuracy, supervised classification and deep learning methods have high accuracy but high cost and weak generalization, and the above hyperspectral patent technology focuses on farmland scenes, but still has room for improvement in cross-platform adaptation, extreme environment adaptation, and large-scale computing efficiency. Therefore, there is an urgent need for a eucalyptus extraction method that can fully utilize the unique spectral response of eucalyptus in the short-wave infrared, red edge, and near-infrared wavebands, and balance high efficiency and strong adaptability, to break through the existing technical bottlenecks and provide more comprehensive technical support for forestry resource supervision and farmland protection. SUMMARY

[0005] The purpose of the present application is to provide a eucalyptus extraction method based on SWIR difference enhancement and eucalyptus canopy characteristic factors, which provides a new method to solve the problems of weak adaptability to complex surfaces and insufficient discrimination of ground objects in existing eucalyptus recognition methods.

[0006] The present application provides a eucalyptus extraction method based on SWIR difference enhancement and eucalyptus canopy characteristic factors, which comprises the following steps:

[0007] S1: Obtain Sentinel-2 multispectral image data, normalize the reflectivity of each waveband, convert the digital value of the original image, referred to as DN, into normalized data in the range of 0-1; then extract the feature wavebands required for eucalyptus identification from the image, including the red waveband, red edge 1 waveband, red edge 2 waveband, red edge 3 waveband, near-infrared waveband, short-wave infrared 1 waveband, and short-wave infrared 2 waveband;

[0008] S2: Calculate the SWIR difference enhancement factor for eucalyptus feature identification and the eucalyptus canopy feature factor; the SWIR difference enhancement factor is constructed based on the reflectivity difference of the short-wave infrared double waveband, used to represent the structural difference characteristics of eucalyptus in the short-wave infrared region, helping to strengthen the structural difference identification of eucalyptus and other ground objects in the short-wave infrared waveband, and reduce the interference of non-eucalyptus ground objects on subsequent feature extraction; the eucalyptus canopy feature factor is constructed by weighted fusion of red edge slope, near-infrared enhancement, and short-wave infrared structural characteristics, used to extract the spectral properties of eucalyptus that are different from other vegetation, helping to amplify the spectral difference between eucalyptus and confused vegetation, and improve the spectral discrimination accuracy of eucalyptus in complex vegetation coverage scenarios;

[0009] S3: Perform Gaussian smoothing and normalization processing on the SWIR enhancement factor and the eucalyptus canopy feature factor respectively to suppress noise and unify the gray scale range;

[0010] S4: In the two-dimensional feature space constructed by the preprocessed SWIR difference enhancement factor and the eucalyptus canopy feature factor, use the spatial regularization joint Fisher-entropy threshold method, referred to as SR-JFET, to determine the optimal segmentation threshold to , the method considers the inter-class difference, information entropy, and pixel spatial neighborhood consistency, and obtains the globally optimal segmentation threshold in the spectral and spatial domains through joint optimization, respectively generating the corresponding SWIR mask and canopy mask;

[0011] S5: Perform intersection operation on the SWIR mask and the canopy mask to obtain the initial eucalyptus extraction mask;

[0012] S6: Perform morphological opening and closing operation on the initial eucalyptus extraction mask to eliminate small noise points;

[0013] S7: Perform connected region analysis on the mask after morphological opening operation, remove connected regions with an area less than a preset threshold, and obtain the optimized eucalyptus extraction mask;

[0014] S8: Output the optimized eucalyptus extraction mask as the eucalyptus extraction result.

[0015] Further, in step S2, the SWIR difference enhancement factor is calculated based on the reflectivity difference of the short-wave infrared double waveband, and the eucalyptus canopy feature factor is calculated by weighted fusion of the red edge slope, near-infrared enhancement, and short-wave infrared structural characteristics. ​This vegetation spectral characteristic factor, constructed based on the reflectance difference between two shortwave infrared bands, helps to address the limitations of traditional single shortwave infrared (SWIR) band characteristic factors, which cannot simultaneously incorporate "vegetation moisture-sensitive" features. "and "sensitive to vegetation cell structure" "Two types of key spectral information lead to confusion and difficulty in distinguishing eucalyptus trees from similar vegetation in terms of spectral characteristics;

[0016] SWIR differential enhancement factor The calculation formula is as follows:

[0017]

[0018] in: It is the baseline offset used to adjust the baseline level of the factor and avoid insufficient discrimination due to excessively small difference values; The reflectance of shortwave infrared band 1 in multispectral images is sensitive to vegetation moisture content. The reflectance of shortwave infrared band 2 in multispectral images is sensitive to vegetation cell structure; This is the difference amplification factor, used to amplify the SWIR differences between ground features and enhance the ability to distinguish between eucalyptus and non-eucalyptus trees; To prevent division by zero of small constants;

[0019] The eucalyptus canopy characteristic factors As a weighted fusion factor, it helps to integrate red-edge feature factors. Near-infrared characteristic factors Shortwave infrared structural characteristic factor Multi-dimensional canopy information can solve the problem that a single feature factor cannot fully characterize the multi-dimensional differences of eucalyptus canopy and is easily confused with the canopy features of other vegetation, thereby enhancing the ability to distinguish eucalyptus from non-eucalyptus in terms of canopy feature dimensions.

[0020] Eucalyptus canopy characteristic factors The calculation formula is as follows:

[0021]

[0022] in: , , For the weighting coefficients, satisfying ;

[0023] The red-edge slope factor is calculated using the following formula:

[0024]

[0025] In the formula, , , These are the reflectivities of red-edge band 1, red-edge band 2, and red-edge band 3, respectively. The red edge slope coefficient;

[0026] The near-infrared enhancement factor is calculated using the following formula:

[0027]

[0028] In the formula, For near-infrared band reflectivity, The reflectivity is in the red band. To prevent small constants from being divided by zero, Near-infrared enhancement coefficient;

[0029] The SWIR structure factor is calculated using the following formula:

[0030]

[0031] In the formula, The reflectance of shortwave infrared band 1 in the multispectral image; The reflectance of shortwave infrared band 2 in the multispectral image; This is the SWIR structure adjustment coefficient, used to adapt to the short-wave infrared spectral characteristics of eucalyptus.

[0032] In the above formula, It is Limited to a specified range between; To prevent division by zero of small constants;

[0033] Furthermore, S4 specifically includes the following steps:

[0034] S4-1: Estimate the initial thresholds for the normalized SWIR difference enhancement factor and the eucalyptus canopy characteristic factor using the Otsu method or Gaussian mixture model, respectively. As a starting point for subsequent optimizations;

[0035] S4-2: Construct a joint optimization objective function in a two-dimensional feature space:

[0036]

[0037] in: Fisher's discriminant is based on the difference between inter-class means and the covariance within classes. It is an intra-class negative entropy index used to measure the purity of information within a class; It is a spatial consistency index based on the proportion of similar pixels in a neighborhood; , , are three weight coefficients for balancing spectral discriminability, information concentration and spatial coherence;

[0038] S4-3: A two-stage strategy based on grid search and derivative-free local optimization is adopted to automatically search the optimal solution of the maximum objective function in the two-dimensional threshold space ;

[0039] S4-4: According to the optimal threshold, the SWIR difference enhancement factor and the eucalyptus canopy feature factor are respectively subjected to binaryzation segmentation to generate an initial mask of the eucalyptus area, and the determination rule is:

[0040]

[0041] wherein, if the pixel satisfies both and , the pixel is marked as 1 in the mask; otherwise, it is marked as 0.

[0042] The beneficial effects produced by the above technical scheme are that the present application constructs an eucalyptus extraction method based on SWIR difference enhancement and eucalyptus canopy feature factor, which can realize full-dimensional collaborative analysis from spectral characteristics to spatial structure, accurately locate the eucalyptus distribution area and generate a high-reliability initial extraction mask, effectively improve the spectral difference recognition degree of eucalyptus and related vegetation, the collaborative representation ability of the unique canopy structure and species characteristics of eucalyptus, and the anti-interference ability to local noise and the spatial continuity of the segmentation result in the threshold segmentation process.

[0043] The core mechanism of "double feature factor customized design + SR-JFET threshold joint optimization" of the present application can more accurately break through the tree species confusion problem caused by "same spectrum different things", avoid misjudgment caused by insufficient information dimension of a single factor, eliminate local noise interference and isolated point missegmentation phenomenon through spatial regularization constraint, and improve the overall accuracy and spatial consistency of eucalyptus extraction, thereby significantly strengthening the stability and practicality of eucalyptus extraction in complex ground object scenes. BRIEF DESCRIPTION OF DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0045] ​​Figure 1 is a flowchart of an eucalyptus extraction method based on SWIR difference enhancement and eucalyptus canopy characteristic factors provided by the present application;

[0046] Figure 2 is a mask map based on SWIR difference enhancement and eucalyptus canopy characteristic factors provided by the present application;

[0047] Figure 3 is a preliminary eucalyptus mask map obtained after intersection operation of double masks provided by the present application;

[0048] Figure 4 is an effect picture of eucalyptus extraction result provided by the present application. DETAILED DESCRIPTION

[0049] To make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0050] As shown in Figure 1 , the present application discloses a eucalyptus extraction method based on SWIR difference enhancement and eucalyptus canopy characteristic factors, comprising the following steps:

[0051] S1: acquiring a multi-spectral remote sensing image of Sentinel-2, covering red, near-infrared, red edge three bands and short wave infrared two bands;

[0052] In the present embodiment, since the core feature of using Sentinel-2 L2A product is that “radiometric calibration + atmospheric correction” has been completed, the output is directly the ground reflectance data, and the image does not need to be radiometrically corrected, atmospherically corrected and spatially resampled, only the reflectance of each band needs to be normalized to provide reliable data basis for subsequent feature extraction.

[0053] The read red band, red edge 1 band, red edge 2 band, red edge 3 band, near-infrared band, short wave infrared 1 band and short wave infrared 2 band are normalized band by band, and the DN value of the original image is converted into normalized data in the range of 0-1.

[0054] The normalized band data in the range of 0-1, the calculation formula is as follows:

[0055]

[0056] ​​wherein, is the single-band original data; , is the minimum value and the maximum value of the band original data; is a small constant to prevent division by zero;

[0057] S2: Calculate the SWIR difference enhancement factor for eucalyptus feature recognition and the eucalyptus canopy feature factor;

[0058] SWIR difference enhancement factor is a vegetation spectral feature factor based on the difference in double short-wave infrared band reflectance, and the calculation formula is as follows:

[0059]

[0060] wherein, is the basic offset, which is used to adjust the baseline level of the factor to avoid insufficient discrimination caused by too small difference value; The reflectance of short-wave infrared band 1 in the multispectral image is sensitive to the water content of vegetation, The reflectance of short-wave infrared band 2 in the multispectral image is sensitive to the cell structure of vegetation; is the difference amplification coefficient, which is used to amplify the SWIR difference between ground objects and enhance the discrimination ability of eucalyptus and non-eucalyptus; is a small constant to prevent division by zero;

[0061] Eucalyptus canopy feature factor The calculation formula is as follows:

[0062]

[0063] wherein, , , is the weighted coefficient, which satisfies ;

[0064] is the red edge slope factor, and the calculation formula is as follows:

[0065]

[0066] In the formula, , , are the reflectances of red edge band 1, red edge band 2, and red edge band 3 respectively, is the red edge slope coefficient;

[0067] is the near-infrared enhancement factor, and the calculation formula is as follows:

[0068]

[0069] wherein, is the reflectance of the near-infrared band, is the reflectance of the red band, is a small constant to prevent division by zero, is the near-infrared enhancement coefficient;

[0070] is the SWIR structure factor, and the calculation formula is as follows:

[0071]

[0072] wherein, is the reflectance of the short-wave infrared band 1 in the multispectral image; is the reflectance of the short-wave infrared band 2 in the multispectral image; is the SWIR structure adjustment coefficient, which is used to adapt to the short-wave infrared spectral characteristics of eucalyptus;

[0073] In the above formula, is the reflectance of the near-infrared band, is limited to a specified range ; is a small constant to prevent division by zero;

[0074] S3: Gaussian smoothing and normalization are respectively performed on the SWIR enhancement difference factor and the eucalyptus canopy characteristic factor to suppress noise and unify the gray scale range;

[0075] S4: In the two-dimensional feature space constructed by the preprocessed SWIR difference enhancement factor and the eucalyptus canopy characteristic factor, the spatial regularization joint Fisher-entropy threshold method is used to determine the optimal segmentation threshold , which comprehensively considers the inter-class difference degree, information entropy and pixel spatial neighborhood consistency, and obtains the global optimal segmentation threshold in the spectral and spatial dual domains through joint optimization, and respectively generates the corresponding SWIR mask and canopy mask.

[0076] The S4 specifically includes the following steps:

[0077] S4-1: The Otsu method or Gaussian mixture model is respectively used to estimate the initial threshold of the normalized SWIR difference enhancement factor and the eucalyptus canopy characteristic factor as the starting point for subsequent optimization;

[0078] S4-2: A joint optimization objective function is constructed in the two-dimensional feature space:

[0079]

[0080] Wherein: Fisher discriminant based on the difference between the mean value of the class and the covariance within the class; Entropy within the class, which is used to measure the purity of information within the class; Spatial consistency index based on the proportion of the same class of neighborhood pixels; Three weight coefficients for balancing spectral discriminability, information concentration, and spatial continuity;

[0081] S4-3: A two-stage strategy based on grid search and derivative-free local optimization is adopted to automatically search for the optimal solution of the maximum objective function in the two-dimensional threshold space

[0082] S4-4: According to the optimal threshold, the SWIR difference enhancement factor and the eucalyptus canopy feature factor are respectively subjected to binary segmentation to generate the preliminary eucalyptus region mask, and the determination rule is:

[0083]

[0084] If the pixel satisfies both and , the pixel is marked as 1 in the mask; otherwise, it is marked as 0.

[0085] The preliminary eucalyptus region masks (SWIR mask and canopy mask) of the SWIR difference enhancement factor and the eucalyptus canopy feature factor are generated, as shown in Figure 2

[0086] S5: Dual-threshold intersection screening is performed, and the intersection of the two preliminary binary masks is taken as the original eucalyptus mask, i.e., only the region marked as 1 in both masks is retained—this region needs to satisfy the threshold conditions of the SWIR enhancement factor and the eucalyptus canopy feature factor, as shown in Figure 3 The white region is the eucalyptus region screened by the dual-threshold intersection.

[0087] S6: Morphological and small-area filtering post-processing is performed: morphological opening and closing operations are sequentially performed on the original eucalyptus mask—opening operation is used to remove small-area noise points, and closing operation is used to fill small holes in the mask.

[0088] S7: Connected region analysis is performed to filter regions with an area smaller than a preset threshold, and finally the eucalyptus extraction mask is generated.

[0089] S8: The superimposed display is realized by loading and displaying the remote sensing image as the base map first, then extracting the edge contour from the binary mask of the eucalyptus extraction, and finally superimposing and drawing the red contour line on the image, and the processed eucalyptus extraction mask result is output, as shown in​​​​​​Figure 4 as shown.

Claims

1. A method for extracting eucalyptus trees based on SWIR differential enhancement and eucalyptus canopy characteristic factors, characterized in that... Includes the following steps: S1: Acquire Sentinel-2 multispectral image data, normalize the reflectance of each band, and convert the digital values ​​(DN) of the original image into... Standardized data for the range; then extract the feature bands required for eucalyptus tree identification from the images, including: red band, red edge 1 band, red edge 2 band, red edge 3 band, near-infrared band, shortwave infrared 1 band and shortwave infrared 2 band; S2: Calculate the SWIR difference enhancement factor and eucalyptus canopy feature factor used for eucalyptus feature identification; the SWIR difference enhancement factor The vegetation spectral feature factor is constructed based on the difference in reflectance in the shortwave infrared band, which is used to characterize the structural difference features of eucalyptus in the shortwave infrared region; the eucalyptus canopy feature factor is constructed by weighted fusion of red edge slope, near-infrared enhancement and shortwave infrared structural characteristics, and is used to extract the spectral attributes that distinguish eucalyptus from other vegetation. The SWIR difference enhancement factor The calculation formula is as follows: in: It is the baseline offset used to adjust the baseline level of the factor and avoid insufficient discrimination due to excessively small difference values; The reflectance of the shortwave infrared band 1 in multispectral imagery is sensitive to vegetation moisture content. The reflectance of the shortwave infrared band 2 in multispectral images is sensitive to the cellular structure of vegetation. This is the difference amplification factor, used to amplify the SWIR differences between ground features and enhance the ability to distinguish between eucalyptus and non-eucalyptus trees; To prevent division by zero of small constants; S3: Gaussian smoothing and normalization are performed on the SWIR difference enhancement factor and the eucalyptus canopy characteristic factor, respectively. S4: Within the two-dimensional feature space constructed by the preprocessed SWIR difference enhancement factor and eucalyptus canopy characteristic factors, the optimal segmentation threshold pair is determined using spatial regularization combined with the Fisher-Entropy Thresholding method (SR-JFET). The method comprehensively considers inter-class differences, information entropy and pixel spatial neighborhood consistency, and obtains the globally optimal segmentation threshold in both spectral and spatial domains through joint optimization, thereby generating corresponding SWIR masks and canopy masks respectively. S5: Perform an intersection operation on the SWIR mask and the canopy mask to obtain the initial eucalyptus extraction mask; S6: Perform a morphological opening operation on the initial eucalyptus extraction mask to eliminate small noise points; S7: Perform connected component analysis and remove connected components with an area smaller than a preset threshold; S8: Output the optimized eucalyptus extraction mask as the eucalyptus extraction result.

2. The eucalyptus extraction method based on SWIR difference enhancement and eucalyptus canopy characteristic factors as described in claim 1, characterized in that, The eucalyptus canopy characteristic factors As a weighted fusion factor, it helps to integrate the red-edge slope factor. Near-infrared enhancement factor SWIR structural factors Multi-dimensional canopy information; eucalyptus canopy characteristic factors The calculation formula is as follows: in: , , For the weighting coefficients, satisfying ; Red-edge slope factor The calculation formula is as follows: In the formula, , , These are the reflectances of red-edge band 1, red-edge band 2, and red-edge band 3, respectively. The red edge slope coefficient; Near-infrared enhancement factor The calculation formula is as follows: In the formula, For near-infrared band reflectivity, The reflectivity is in the red band. Near-infrared enhancement coefficient; SWIR structural factor The calculation formula is as follows: In the formula, This is the SWIR structure adjustment coefficient, used to adapt to the short-wave infrared spectral characteristics of eucalyptus. In the above formula, It is Limited to a specified range between.

3. The eucalyptus extraction method based on SWIR difference enhancement and eucalyptus canopy characteristic factors as described in claim 1, characterized in that, S4 specifically includes the following steps: S4-1: Estimate the initial thresholds for the normalized SWIR difference enhancement factor and the eucalyptus canopy characteristic factor using the Otsu method or Gaussian mixture model, respectively. As a starting point for subsequent optimizations; S4-2: Construct a joint optimization objective function in a two-dimensional feature space: in: Fisher's discriminant is based on the difference between inter-class means and the covariance within classes. It is an intra-class negative entropy index used to measure the purity of information within a class; It is a spatial consistency index based on the proportion of similar pixels in a neighborhood; , , These are three weighting coefficients used to balance spectral discriminability, information concentration, and spatial coherence; S4-3: A two-stage strategy based on grid search and derivative-free local optimization is adopted to automatically search for the joint optimization objective function in the two-dimensional feature space. The optimal segmentation threshold ; S4-4: Based on the optimal segmentation threshold... Binarization segmentation was performed on the SWIR difference enhancement factor and the eucalyptus canopy feature factor respectively to generate a preliminary mask for the eucalyptus area. The determination rule is as follows: Among them, if the pixel simultaneously satisfies and If the value is 1, then the pixel is marked as 1 in the mask; otherwise, it is marked as 0.

Images