A Smart Assessment Method for Coral Reef Health

Abstract

This invention provides an intelligent method for assessing the health status of coral reefs. It involves acquiring color images of coral reef areas using an underwater camera, preprocessing the images to improve image quality, constructing an image enhancement and classification model to enhance the images, improve classification accuracy, and identify the coral species in the acquired images. Coral coverage, coral health index, and the average color score of all corals in the area are used as evaluation indicators for coral reef health status, resulting in a comprehensive score. This invention solves the problem that weak underwater light greatly limits the quality of coral images captured by cameras, posing a significant obstacle to the health assessment of coral reef areas.

Description

Technical Field

[0001] This invention relates to the field of intelligent computing technology, and in particular to an intelligent assessment method for the health status of coral reefs. Background Technology

[0002] Coral reefs are among the most biodiverse and diverse ecosystems in the ocean. However, due to global climate change and human activities, corals are under immense pressure, and are currently in a state of continuous decline. With the rapid development of underwater robots, video recordings of warm-water corals are being used to study their distribution and health. However, the strong scattering and absorption properties of light in water result in weak light, significantly limiting the quality of coral images captured by cameras and posing a major obstacle to the health assessment of coral reef areas. Summary of the Invention

[0003] The technical problem this invention aims to solve is that the weak light in water greatly limits the quality of coral images captured by cameras, posing a significant obstacle to the health assessment of coral reef areas. Therefore, this invention provides an intelligent method for assessing the health status of coral reefs.

[0004] The intelligent assessment method for coral reef health described in this invention includes the following steps:

[0005] S1. Acquire color images of the coral reef area using an underwater camera, and preprocess the images to improve image quality;

[0006] S2. Construct an image enhancement classification model to improve classification accuracy and identify coral species in the acquired images by enhancing the images;

[0007] S3. Coral coverage, coral health index, and average color score of all corals in the area are used as evaluation indicators for coral reef health status to obtain a comprehensive score for coral reef health status.

[0008] Furthermore, step S1 includes:

[0009] S11. Use an underwater camera to acquire images of the coral reef area, and preprocess the images to improve image quality. The coral images acquired by the underwater camera are color images. Analyze the coral image s(i,j), treating the color image as a two-dimensional pixel array with three independent channels (R, G, B), where (i,j) represents the row and column of the two-dimensional pixel array. The grayscale image of the three channels is s. R (i, j), s G (i, j), s B(i, j), the color image is processed into a grayscale image in different color channels; in each color channel, each pixel corresponds to the response value of the corresponding photosensitive element to red, green and blue light, thus converting the color image into three grayscale images;

[0010] The original image is processed using a multi-scale homomorphic filtering algorithm with a color restoration factor:

[0011]

[0012] Among them, S l (i, j) represents the two-dimensional pixel matrix of the l-th channel of the processed image, l∈{R, G, B}, s l (i, j) represents the two-dimensional pixel array of the l-th channel of the original image, N represents the scale number, and W n It is the weight corresponding to the nth scale, n∈N, F n (i, j) represents the Gaussian kernel function corresponding to the nth scale;

[0013] S12. Then, the image is denoised using a diffusion filter method. First, a time operator t is introduced to indicate that the denoising is related to the diffusion duration:

[0014] S l (i, j) = I0 = I(i, j, 0)

[0015] Where I0 represents the image signal at the initial time; calculate the partial derivative of the image signal at time t:

[0016]

[0017] Where div represents the divergence operator. Representing image gradient The diffusion equation, For gradient operators, Let I represent the gradient magnitude, and k represent the gradient threshold. When the image gradient is much greater than the gradient threshold k, the diffusion equation tends to 0; when the image gradient is much less than the gradient threshold k, the diffusion equation tends to 1. The original image I0 acts as a medium, spreading across the image at a non-constant speed. The regions it spreads through gradually form a series of smooth images I(i,j,t).

[0018] Furthermore, step S2 includes:

[0019] The acquired image is divided into M small squares of size U×V×3 to construct an image enhancement classification model. By enhancing the image, the classification accuracy is improved, and the coral species in the acquired image are identified. The image enhancement classification model includes an encoder, a decoder, and a discriminator. The encoder consists of multiple convolutional layers, the number of which is determined according to actual needs. The decoder consists of multiple deconvolutional layers that correspond to the mirror images of the convolutional layers. There is a "solid-state pointer fusion" operation of feature maps between the convolutional layers and the deconvolutional layers.

[0020] Assuming the encoder and decoder consist of m convolutional and deconvolutional layers, the "solid-state pointer fusion" operation performs convolutional fusion on the channel between the feature map output by the q-th convolutional layer and the feature map output by the mq-th deconvolutional layer.

[0021] The convolution kernel H of the convolutional layer 1 The deconvolutional layer has a 3×3 structure, 3 channels, a stride of 2, and a kernel H. 2 It is 3×3, with a step size of

[0022] Initialize encoder parameters μ e The decoder parameter μ d and the discriminant's parameters μ a The network input is represented as I = I 1 I 2 , ..., I M First, the encoder obtains image features through convolution operations:

[0023]

[0024] Where conv is the output convolutional feature map. Let γ be the input image, γ∈M, c1 represents the number of input channels, c1=3, c2 represents the number of output channels, and i0, j0 represent the channel coordinates; the convolution operation of the encoder enhances the ability of the image augmentation classification model to extract image features;

[0025] After each sampling process by the encoder through a convolutional layer, the resulting image feature size is reduced to its original value. The obtained feature map is input into the decoder and subjected to a deconvolution operation, which increases the image size by a factor of four. The deconvolution operation is as follows:

[0026]

[0027] in, c3 is the output deconvolution feature map, where c3 is the number of output channels of the deconvolution.

[0028] The feature map output from the q-th convolutional layer and the feature map output from the mq-th deconvolutional layer are convolved and fused along the channels using the "solid pointer fusion" operation.

[0029]

[0030] After fusion In the image feature input discriminator, the "solid-state pointer fusion" operation is used to alleviate the artifacts and mosaic caused by deconvolution upsampling, thereby improving the quality of image generation.

[0031] The discriminator consists of two parts: the first part is a module composed of convolution and rectified linear units, and the second part is a module composed of convolution and batch normalization.

[0032] The convolution kernel in the first part is The output is:

[0033] The convolution kernel in the second part is The output is: in σ is the mean of the input samples, and σ is the variance of the input samples. The final output is a two-dimensional matrix, where each element maps to a sub-region in the input image, used to determine the coral species corresponding to each feature point in the input image.

[0034] Furthermore, step S3 includes:

[0035] The identified coral color is compared with a coral health color chart to obtain the coral color score Si. The specific method is as follows:

[0036] Convert the coral color values ​​from RGB to HSV, and calculate the color distance between the coral color values ​​in the HSV domain and each color square on the coral health color chart:

[0037]

[0038] Where (x0, y0, z0) are the coordinates of the coral color value within the HSV domain; (x i y i , z i ) represents the coordinates of the color value of each color square on the coral health color chart within the HSV domain; the coral color score Si = min d is obtained based on the color square information corresponding to the minimum color distance.

[0039] Coral coverage, coral health index, and color score Si of all corals in the area are used as evaluation indicators for coral reef health. Based on the weight values ​​of different evaluation indices, a comprehensive score W for coral reef health is obtained, and the coral reef health assessment result is obtained based on the comprehensive score W.

[0040] Furthermore, step S3 includes:

[0041] The diversity index of corals is:

[0042]

[0043] Where HI is the diversity index, p i Let be the number of the i-th type of coral, P be the total number of corals, and s be the number of coral species; the coral health index is:

[0044]

[0045] Among them, H i H is the diversity index of the i-th biological group. i0 is the health value of the diversity index of the i-th type of biological group, and n is the number of biological groups;

[0046] Calculate coral coverage: A represents the number of small squares occupied by corals in an image that is divided into U×V squares.

[0047] Calculate the average color score (MI) for all corals:

[0048]

[0049] The corresponding evaluation scores Y1, Y2, and Y3 are obtained based on the calculated coral coverage (TC), coral health index (CI), and average color score (MI) of all corals:

[0050] Y1 = TC × 100

[0051]

[0052]

[0053] Where α1, α2, α3, α4, α5, b1, b2, b3, and b4 are all parameters, and δ1, δ2, δ3, and δ4 are all exponential thresholds, which are obtained from experiments;

[0054] Coral coverage, coral health index, and average color score of all corals in the area were used as evaluation indicators for coral reef health. Based on the weights of the different evaluation indices, a comprehensive score W for coral reef health was obtained.

[0055] W = ω1Y1 + ω2Y2 + ω3Y3

[0056] Wherein, ω1 is the weight value of coral coverage; ω2 is the weight value of coral health index; and ω3 is the weight value of the average color score of all corals.

[0057] The beneficial effects of this invention are:

[0058] 1. Achieve noise reduction for images while effectively preserving image features such as boundaries and lines, thereby increasing image recognition accuracy.

[0059] 2. Constructing the "solid pointer fusion" operation can realize low-order information sharing between input and output, alleviate problems such as artifacts and mosaic caused by deconvolution upsampling, and help improve the quality of image generation.

[0060] 3. Use coral coverage, coral health index, and the color score of all corals in the area as evaluation indicators for the health status of coral reefs to increase the accuracy of the assessment. Attached Figure Description

[0061] Figure 1 Flowchart of an intelligent assessment method for coral reef health status as described in this invention;

[0062] Figure 2 The structural diagram of the image enhancement classification model described in this invention. Detailed Implementation

[0063] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings, thereby enabling a full understanding of how the present invention uses technical means to solve technical problems and achieve technical effects, and allowing for implementation accordingly. It should be noted that, as long as there is no conflict, the various features in the various embodiments of the present invention can be combined with each other, and the resulting technical solutions are all within the protection scope of the present invention.

[0064] Reference Figure 1 The intelligent assessment method for coral reef health status described in this invention includes:

[0065] S1. Acquire color images of the coral reef area using an underwater camera, and preprocess the images to improve image quality;

[0066] S2. Construct an image enhancement classification model to improve classification accuracy and identify coral species in the acquired images by enhancing the images;

[0067] S3. Coral coverage, coral health index, and average color score of all corals in the area are used as evaluation indicators for coral reef health status to obtain a comprehensive score for coral reef health status.

[0068] Furthermore, the image preprocessing method in step S1 is as follows:

[0069] S11. Use an underwater camera to acquire images of the coral reef area, and preprocess the images to improve image quality. The coral images acquired by the underwater camera are color images. Analyze the coral image s(i,j), treating the color image as a two-dimensional pixel array with three independent channels (R, G, B), where (i,j) represents the row and column of the two-dimensional pixel array. The grayscale image of the three channels is s. R (i, j), s G (i, j), s B (i, j), the color image is processed into a grayscale image in different color channels. In each color channel, each pixel corresponds to the response value of the corresponding photosensitive element to red, green and blue light, thereby converting the color image into three grayscale images. The above technology is prior art, and will not be elaborated on further here.

[0070] The original image is processed using a multi-scale homomorphic filtering algorithm with a color restoration factor:

[0071]

[0072] Among them, S l (i, j) represents the two-dimensional pixel matrix of the l-th channel of the processed image, l∈{R, G, B}, s l (i, j) represents the two-dimensional pixel array of the l-th channel of the original image, N represents the scale number, and W n It is the weight corresponding to the nth scale, n∈N, F n (i, j) represents the Gaussian kernel function corresponding to the nth scale.

[0073] S12. Then, the image is denoised using a diffusion filter method. First, a time operator t is introduced to indicate that the denoising is related to the diffusion duration:

[0074] S l (i, j) = I0 = I(i, j, 0)

[0075] Where I0 represents the image signal at the initial time. Calculate the partial derivative of the image signal at time t:

[0076]

[0077] Where div represents the divergence operator. Representing image gradient The diffusion equation, For gradient operators, Let I represent the gradient magnitude, and k represent the gradient threshold. When the image gradient is much greater than the gradient threshold k, the diffusion equation tends to 0; when the image gradient is much less than the gradient threshold k, the diffusion equation tends to 1. The original image I0 acts as a medium, spreading across the image at a non-constant speed. The regions it spreads through gradually form a series of smooth images I(i,j,t).

[0078] The beneficial effect of the diffusion filtering method is that it can reduce the noise of the image while effectively preserving the image's boundary and line features, thereby increasing the accuracy of image recognition.

[0079] Furthermore, the coral species identification method in step S2 is as follows:

[0080] The acquired image is divided into M small squares of size U×V×3, and an image enhancement classification model is constructed, such as... Figure 2 As shown, image enhancement improves classification accuracy and identifies coral species in the acquired images. The image enhancement classification model includes an encoder, a decoder, and a discriminator. The encoder consists of multiple convolutional layers, the number of which is determined according to actual needs. The decoder consists of multiple deconvolutional layers corresponding to the mirror images of the convolutional layers, and there is a "solid-state pointer fusion" operation of feature maps between the convolutional and deconvolutional layers.

[0081] The "solid-state pointer fusion" operation enables low-order information sharing between input and output, which is beneficial for improving the quality of image generation. Assuming that the encoder and decoder consist of m convolutional and deconvolutional layers, the "solid-state pointer fusion" operation performs convolutional fusion on the channel between the feature map output by the q-th convolutional layer and the feature map output by the mq-th deconvolutional layer.

[0082] The convolution kernel H of the convolutional layer 1 The deconvolutional layer has a 3×3 structure, 3 channels, a stride of 2, and a kernel H. 2 It is 3×3, with a step size of

[0083] Initialize encoder parameters μ e The decoder parameter μ d and the discriminant's parameters μ a The network input is represented as I = I 1 I 2 , ..., I M First, the encoder obtains image features through convolution operations:

[0084]

[0085] Where conv is the output convolutional feature map. Let be the γ-th input image, γ∈M, c1 represents the number of input channels (c1=3), c2 represents the number of output channels, and i0 and j0 represent the channel coordinates. The convolution operation of the encoder enhances the ability of the image augmentation classification model to extract image features.

[0086] After each sampling process by the encoder through a convolutional layer, the resulting image feature size is reduced to its original value. The obtained feature map is input into the decoder and deconvolution is performed to increase the image size by four times. The deconvolution operation is as follows:

[0087]

[0088] in, c3 is the output deconvolution feature map, and c3 is the number of output channels of the deconvolution.

[0089] The feature map output from the q-th convolutional layer and the feature map output from the mq-th deconvolutional layer are convolved and fused along the channels using the "solid pointer fusion" operation.

[0090]

[0091] After fusion In the image feature input discriminator, the "solid-state pointer fusion" operation alleviates problems such as artifacts and mosaic caused by deconvolution upsampling, thereby improving the quality of image generation.

[0092] The discriminator consists of two parts: the first part is a module composed of convolution and rectified linear units, and the second part is a module composed of convolution and batch normalization.

[0093] The convolution kernel in the first part is The output is:

[0094] The convolution kernel in the second part is The output is: in Let σ be the mean of the input samples and σ be the variance of the input samples. The final output is a two-dimensional matrix, where each element maps to a sub-region in the input image, used to determine the coral species corresponding to each feature point in the input image.

[0095] Furthermore, the method for assessing the health status of coral reefs in step S3 is as follows:

[0096] The identified coral color is compared with a coral health color chart to obtain the coral color score Si. The specific method is as follows:

[0097] Convert the coral color values ​​from RGB to HSV, and calculate the color distance between the coral color values ​​in the HSV domain and each color square on the coral health color chart:

[0098]

[0099] Where (x0, y0, z0) are the coordinates of the coral color value within the HSV domain; (x i y i , z i ) represents the coordinates of the color value of each color square on the coral health color chart within the HSV domain; the coral color score Si = min d is obtained based on the color square information corresponding to the minimum color distance.

[0100] Coral coverage, coral health index, and color score Si of all corals in the area are used as evaluation indicators of coral reef health status. Based on the weight values ​​of different evaluation indices, a comprehensive score W of coral reef health status is obtained, and the coral reef health assessment result is obtained based on the comprehensive score W of coral reef health status.

[0101] The specific method for calculating the comprehensive score W of the coral reef health status is as follows:

[0102] The diversity index of corals is:

[0103]

[0104] Where HI is the diversity index, p i Let be the number of the i-th type of coral, P be the total number of corals, and s be the number of coral species. The coral health index is:

[0105]

[0106] Among them, H i H is the diversity index of the i-th biological group. i0 is the health value of the diversity index of the i-th biological group, and n is the number of biological groups.

[0107] Calculate coral coverage: A represents the number of small squares occupied by corals in an image that is divided into U×V squares.

[0108] Calculate the average color score (MI) for all corals:

[0109]

[0110] The corresponding evaluation scores Y1, Y2, and Y3 are obtained based on the calculated coral coverage (TC), coral health index (CI), and average color score (MI) of all corals:

[0111] Y1 = TC × 100

[0112]

[0113]

[0114] Among them, α1, α2, α3, α4, α5, b1, b2, b3, and b4 are all parameters, and δ1, δ2, δ3, and δ4 are all exponential thresholds, which are derived from experiments.

[0115] Coral coverage, coral health index, and average color score of all corals in the area were used as evaluation indicators for coral reef health. Based on the weights of the different evaluation indices, a comprehensive score W for coral reef health was obtained.

[0116] W = ω1Y1 + ω2Y2 + ω3Y3

[0117] Wherein, ω1 is the weight value of coral coverage; ω2 is the weight value of coral health index; and ω3 is the weight value of the average color score of all corals.

[0118] The beneficial effect of the proposed coral reef health assessment method is that it uses coral coverage, coral health index, and the color score Si of all corals in the area as evaluation indicators of coral reef health status, thereby increasing the accuracy of the assessment.

[0119] In summary, this invention provides a method for intelligently assessing the health status of coral reefs.

[0120] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0121] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0122] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0123] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A smart assessment method for coral reef health status, characterized in that, Includes the following steps: S1. Acquire color images of the coral reef area using an underwater camera, and preprocess the images to improve image quality; S2. Construct an image enhancement and classification model to enhance images and identify coral species in the acquired images; Image enhancement classification models include encoders, decoders, and discriminators; the encoder consists of convolutional layers. The decoder consists of deconvolutional layers, and there is a "solid pointer fusion" operation of feature maps between the convolutional and deconvolutional layers; Divide the acquired image into equal parts A small square, assuming the encoder and decoder together consist of... It consists of a convolutional layer and a deconvolutional layer, and the "solid-state pointer fusion" operation will... The feature map output by the first convolutional layer and the first convolutional layer The feature maps output by each deconvolutional layer are convolved and fused across channels; By using the "solid-state pointer fusion" operation, the first... The feature map output by the first convolutional layer and the first convolutional layer The feature maps output by each deconvolutional layer are then convolved and fused across channels: ， in, The output is the convolutional feature map. The output is the deconvolution feature map. The fused feature map is then used to... In the image feature input discriminator, the discriminator is composed of a module consisting of convolution and modified linear units, and a module consisting of convolution and batch normalization. The convolution kernel in the first part is The output is: ; The convolution kernel in the second part is The output is: ,in, For the parameters of the discriminator, The mean of the input samples. The variance of the input sample; The final output is a two-dimensional matrix, where each element of the matrix maps to a sub-region of the input image. S3. Coral coverage, coral health index, and average color score of all corals in the area are used as evaluation indicators for coral reef health status to obtain a comprehensive score for coral reef health status.

2. The intelligent assessment method for coral reef health status according to claim 1, characterized in that, Step S1 includes: S11. Use an underwater camera to acquire images of the coral reef area, and preprocess the images to improve image quality; the coral images acquired by the underwater camera are color images, and the coral images... The analysis treats the color image as a two-dimensional pixel matrix with three independent channels: R, G, and B. Representing the rows and columns of a two-dimensional pixel array, a three-channel grayscale image is... , , The color image is processed into a grayscale image in different color channels; in each color channel, each pixel corresponds to the response value of the corresponding photosensitive element to red, green and blue light, thus converting the color image into three grayscale images; The original image is processed using a multi-scale homomorphic filtering algorithm with a color restoration factor: ， in, Indicates the processed image number 1 The two-dimensional pixel array of the channel, , Represents the original image. The two-dimensional pixel array of the channel, Indicates the scale quantity. It is the first The weights corresponding to each scale , Indicates the first Gaussian kernel functions corresponding to each scale; S12. Then, the image is denoised using a diffusion filter method, first introducing a time operator. This indicates that noise reduction is related to the duration of diffusion. ， in, Represents the image signal at the initial moment; calculate Partial derivative of the image signal at time: ， in, Denotes the divergence operator, Representing image gradient The diffusion equation, For gradient operators, Indicates the gradient magnitude. This represents the gradient threshold, which is used when the image gradient is much larger than the gradient threshold. When the diffusion equation approaches 0, the image gradient is much smaller than the gradient threshold. When the diffusion equation approaches 1, the original image... As a medium, it diffuses across the image at a non-constant speed, and the areas it diffuses through gradually form a series of smooth images. .

3. The intelligent assessment method for coral reef health status according to claim 2, characterized in that, Step S2 includes: The convolution kernel of the convolutional layer for The number of channels is 3, the stride is 2, and the convolution kernel of the deconvolution layer is... for Step size is ; Initialize encoder parameters Decoder parameters and the parameters of the discriminator Network input is represented as First, the encoder obtains image features through convolution operations: ， in, The output is the convolutional feature map. For the input of the first One image, , Indicates the number of input channels. , Indicates the number of output channels. The channel coordinates are represented; the convolution operation of the encoder enhances the ability of the image augmentation classification model to extract image features. After each sampling process by the encoder through a convolutional layer, the resulting image feature size is reduced to its original value. The obtained feature map is input into the decoder and deconvolution is performed to increase the image size by 4 times; the deconvolution operation is as follows: ， in, The output is the deconvolution feature map. It is the number of output channels of the deconvolution.

4. The intelligent assessment method for coral reef health status according to claim 3, characterized in that, Step S3 includes: The identified coral color is compared with a coral health color chart to obtain the coral color score. The specific method is as follows: Convert the coral color values ​​from RGB to HSV, and calculate the color distance between the coral color values ​​in the HSV domain and each color square on the coral health color chart: ， in, These are the coordinates of the coral color values ​​within the HSV domain; These are the coordinates of the color values ​​of each color square on the HSV coral health color chart; the coral color score is obtained based on the color square information corresponding to the minimum color distance. ; Coral coverage, coral health index, and color scores of all corals in the area. As an indicator of coral reef health, a comprehensive score of coral reef health is obtained based on the weight values ​​of different evaluation indices. And based on a comprehensive score of coral reef health status. Obtain the results of the coral reef health assessment.

5. The intelligent assessment method for coral reef health status according to claim 4, characterized in that, Step S3 includes: The diversity index of corals is: ， in, It is a diversity index. It is the first The number of coral species It refers to the total number of corals. It refers to the number of coral species; the coral health index is: ， in, It is the first Diversity index of biological groups It is the first The health value of the biodiversity index of biological groups The number of biological groups; Calculate coral coverage: , The acquired image is divided into equal parts of size. The number of small squares occupied by coral; Calculate the average color score (MI) for all corals: ， Based on the calculated coral coverage Coral Health Index and the average color score of all corals Obtain its corresponding evaluation score , and : ， ， ， in, , , , , , , , , All are parameters. , , , All are exponential thresholds, derived from experiments; Coral coverage, coral health index, and average color score of all corals in the area were used as evaluation indicators for coral reef health. A comprehensive score for coral reef health was obtained based on the weights of the different evaluation indices. : ， in, This is a weighted value for coral coverage. This is the weighting value for the coral health index; This represents the weighted value for the average color score of all corals.

Images