An Automated Vegetation Cover Extraction Method for Field Camera Surveys

Abstract

This invention discloses an automated method for extracting vegetation cover (FVC) in field camera surveys, comprising the following steps: inputting an RGB electronic photograph to be processed; calculating the greenness index and color saturation of the RGB electronic photograph; determining the threshold for segmenting vegetation based on the greenness index pixel by pixel using color saturation; identifying vegetation pixels based on the greenness index and the pixel-by-pixel threshold; and calculating the proportion of vegetation pixels to the total number of pixels in the RGB electronic photograph to obtain the vegetation cover. This method can determine the optimal threshold for segmenting vegetation using ExG on a pixel-by-pixel basis for RGB electronic photographs imaged under different lighting conditions and terrain scenes, thereby achieving high-precision automated extraction of vegetation cover (FVC). It facilitates the rapid and accurate acquisition of quadrat-scale vegetation cover data during field surveys. Its principle is simple, easy to implement, highly adaptable, highly automated, and widely applicable, making it valuable in fields such as ecology, geography, and remote sensing.

Description

Technical Field

[0001] This invention relates to the fields of remote sensing measurement and image processing technology and computer vision technology, and in particular to an automated method for extracting vegetation cover for field camera surveys using threshold segmentation of remote sensing images. Background Technology

[0002] Fractional Vegetation Cover (FVC) is an important ecological indicator describing the state of land cover and is widely used in fields such as carbon cycling, ecological assessment, and soil nutrient cycling. Ground-based measurements are crucial for obtaining accurate FVC data. Among these methods, photographic methods, due to their simplicity and efficiency, have gradually become an important FVC measurement method in field surveys. Photographic methods acquire true-color electronic photographs in visible light and utilize computer vision technology to identify and segment green vegetation areas, thereby obtaining quantitative information on vegetation cover. In vegetation identification methods based on RGB electronic photographs, the Excess Green Index (ExG) is a commonly used green vegetation identification indicator, widely applied because it can efficiently distinguish between vegetated and non-vegetated areas. However, the optimal segmentation threshold for ExG is highly dependent on imaging conditions and image background. Especially under complex lighting conditions and insufficient background contrast, its recognition accuracy significantly decreases, limiting its automated application in real-world scenarios.

[0003] Currently, common image thresholding methods include global thresholding and local thresholding. Global thresholding is suitable for images with uniform illumination and clear contrast between foreground and background. Commonly used global thresholding methods include maximum entropy thresholding and Otsu's method, which are automatic threshold determination methods. Maximum entropy thresholding performs well in low-density vegetation areas, while Otsu's method has advantages in segmentation that maximizes inter-class variance. However, regardless of the method, the recognition accuracy is limited in imaging scenes with strong illumination variations. Illumination variations cause the optimal segmentation threshold for ExG thresholding to differ between strong and weak light areas, making it difficult for a single global thresholding method to meet the segmentation needs of different illumination regions. In contrast, local thresholding, with its ability to refine local features, exhibits stronger robustness to images with uneven illumination. Summary of the Invention

[0004] This invention addresses the shortcomings of existing methods for extracting vegetation cover using the ExG segmentation threshold, which is easily affected by imaging lighting conditions. It proposes an automated vegetation cover extraction method for field camera surveys. This method utilizes color saturation (S) to determine the optimal ExG segmentation threshold for vegetation on a pixel-by-pixel basis, thereby achieving automated vegetation cover extraction from RGB electronic photographs. This method can determine the optimal ExG segmentation threshold for vegetation on a pixel-by-pixel basis for RGB electronic photographs imaged under different lighting conditions and with various terrain features, thus achieving high-precision automatic extraction of FVC (Frequency Value Capacity). This facilitates the rapid and accurate acquisition of quadrat-scale vegetation cover data during field surveys. Its principle is simple, easy to implement, highly adaptable, highly automated, and widely applicable.

[0005] To achieve this objective, the present invention provides an automated vegetation cover extraction method for field camera surveys, comprising the following steps:

[0006] Input the RGB digital image to be processed;

[0007] Calculate the greenness index and color saturation of RGB digital photographs;

[0008] The threshold for segmenting vegetation is determined pixel by pixel using the overgreen index;

[0009] Vegetation pixels are identified based on the greenness index and pixel-by-pixel threshold.

[0010] The vegetation coverage is obtained by statistically analyzing the proportion of vegetation pixels to the total number of pixels in an RGB electronic photograph.

[0011] Optionally, in the step of determining the threshold for segmenting vegetation using color saturation pixel-by-pixel overgreen index:

[0012] The threshold for segmenting vegetation using the overgreen index is determined pixel-by-pixel by color saturation.

[0013] The threshold for segmenting vegetation using the overgreen index is a regression fitting function of color saturation.

[0014] Optionally, the values ​​of the red, green, and blue bands of the RGB electronic photograph range from 0 to 255, and are in byte format.

[0015] Optionally, in the step of calculating the greenness index and color saturation of the RGB digital photograph: calculate the greenness index and color saturation of each pixel.

[0016] Optionally, in the step of determining the threshold for vegetation segmentation by using color saturation pixel by pixel to determine the overgreen index, the overgreen index threshold adaptive model used is obtained by training RGB electronic photos collected under different lighting conditions, vegetation cover and complex land cover combination scenarios.

[0017] Optionally, the training step may include:

[0018] Based on this data, vegetation and non-vegetation attributes are interpreted pixel by pixel;

[0019] The relationship between the threshold T of the greenness index and color saturation was fitted using regression methods.

[0020] The threshold adaptive model for the greenness index is as follows:

[0021]

[0022] Optionally, the process may also include the step of: using the color saturation of each pixel in the RGB electronic photograph to be processed, and combining it with an adaptive threshold model for the green index to calculate the segmentation threshold of the green index for each pixel.

[0023] Optionally, in the step of identifying vegetation pixels based on the greenness index and pixel-by-pixel threshold, for each pixel of the RGB electronic photo to be processed, the calculated greenness index is compared with the threshold to determine whether the pixel is vegetation.

[0024] Optionally, the formula for calculating color saturation S is:

[0025]

[0026] In the formula, , and These are the normalized values ​​of the red, green, and blue bands in the RGB digital photograph, respectively. It is a minimum value used to avoid the case where the denominator is zero.

[0027] Optionally, for each pixel in the RGB digital photograph, its greenness index ExG, denoted as ExG(x, y), is compared with an adaptive threshold T to determine whether the pixel attribute is vegetation or non-vegetation. The determination criteria are as follows:

[0028]

[0029] in, The results are for pixel classification, where 1 represents vegetation and 0 represents non-vegetation.

[0030] This invention has the following characteristics:

[0031] (1) It has strong adaptability and is suitable for imaging RGB electronic photos under different lighting conditions and ground scenes;

[0032] (2) The principle is simple and easy to implement. The optimal segmentation threshold per pixel of the green index ExG is determined only by the color saturation S.

[0033] (3) High degree of automation. Once the ExG adaptive threshold model is established based on the training samples, the vegetation coverage can be extracted with high precision automatically in the study area.

[0034] (4) It has a wide range of applications and can be widely used in fields such as ecology, geography, and remote sensing. Attached Figure Description

[0035] Figure 1 This is a flowchart of the automated vegetation cover extraction method for field camera surveys according to the present invention.

[0036] Figure 2 The present invention provides an automated vegetation cover extraction method for field camera surveys, using RGB electronic photographs taken from a drone as an example.

[0037] Figure 3 The present invention provides an automated vegetation cover extraction method for field camera surveys, using the ExG feature image of the overgreen index.

[0038] Figure 4 The present invention provides an automated method for extracting vegetation cover from field camera surveys, using color saturation S-feature images.

[0039] Figure 5 This invention relates to an automated vegetation cover extraction method for field camera surveys, using ExG-S scatter plots and an ExG adaptive threshold model.

[0040] Figure 6 The image shown is a vegetation identification result image of the automated vegetation cover extraction method for field camera surveys according to the present invention, wherein the black area represents vegetation and the white area represents non-vegetation. Detailed Implementation

[0041] The inventors discovered in their research on threshold segmentation of remote sensing images that color saturation (S), as an indicator describing the vibrancy of colors, can reflect the visual characteristics of vegetation under different lighting conditions. Therefore, S can be used to indicate different lighting conditions for vegetation in RGB electronic photographs. Both S and ExG values ​​show a synergistic change with lighting conditions: under strong light conditions, vegetation colors are vibrant, S and ExG values ​​are high, and the vegetation segmentation threshold needs to be increased accordingly; under weak light conditions, vegetation colors are dull, S and ExG values ​​are low, and the vegetation segmentation threshold should be appropriately decreased. Based on this, the S value can serve as a key factor in determining the optimal ExG segmentation threshold. It can be used to overcome the influence of lighting variations in images on vegetation recognition, thereby achieving high-precision and automated FVC extraction based on RGB electronic photographs.

[0042] Therefore, to address the shortcomings of existing methods for extracting vegetation cover using the ExG segmentation threshold, which is susceptible to the influence of imaging lighting conditions, an automated method for extracting vegetation cover from field camera surveys is proposed, including the following steps:

[0043] Input the RGB digital image to be processed;

[0044] Calculate the greenness index and color saturation of RGB digital photographs;

[0045] The threshold for segmenting vegetation is determined pixel by pixel using the overgreen index;

[0046] Vegetation pixels are identified based on the greenness index and pixel-by-pixel threshold.

[0047] The vegetation coverage is obtained by statistically analyzing the proportion of vegetation pixels to the total number of pixels in an RGB electronic photograph.

[0048] In one embodiment, in the step of determining the threshold for segmenting vegetation using the overgreen index on a pixel-by-pixel basis using color saturation:

[0049] The threshold for segmenting vegetation using the overgreen index is determined pixel-by-pixel by color saturation.

[0050] The threshold for segmenting vegetation using the overgreen index is a regression fitting function of color saturation.

[0051] In one embodiment, the values ​​of the red, green, and blue bands of the RGB electronic photograph range from 0 to 255, and are in byte format.

[0052] In one embodiment, in the step of calculating the greenness index and color saturation of an RGB electronic photograph: the greenness index and color saturation of each pixel are calculated.

[0053] In one embodiment, in the step of determining the threshold for vegetation segmentation by using color saturation pixel by pixel to determine the overgreen index, the overgreen index threshold adaptive model used is obtained by training RGB electronic photos collected under different lighting conditions, vegetation cover conditions, and complex land cover combination scenarios.

[0054] In one embodiment, the training step includes:

[0055] Based on this data, vegetation and non-vegetation attributes are interpreted pixel by pixel;

[0056] The relationship between the threshold T of the greenness index and color saturation was fitted using regression methods.

[0057] The threshold adaptive model for the greenness index is as follows:

[0058]

[0059] In one embodiment, the method further includes the step of: using the color saturation of each pixel in the RGB electronic photograph to be processed, and combining it with an adaptive threshold model for the green index to calculate the segmentation threshold of the green index for each pixel.

[0060] In one embodiment, in the step of identifying vegetation pixels using the greenness index and pixel-by-pixel threshold, for each pixel of the RGB electronic photograph to be processed, the calculated greenness index is compared with the threshold to determine whether the pixel is vegetation.

[0061] In one embodiment, the formula for calculating color saturation S is:

[0062]

[0063] In the formula, , and These are the normalized values ​​of the red, green, and blue bands in the RGB digital photograph, respectively. It is a minimum value used to avoid the case where the denominator is zero.

[0064] In one embodiment, for each pixel of an RGB electronic photograph, its greenness index ExG, denoted as ExG(x, y), is compared with an adaptive threshold T to determine whether the pixel attribute is vegetation or non-vegetation. The determination condition is as follows:

[0065]

[0066] in, The results are for pixel classification, where 1 represents vegetation and 0 represents non-vegetation.

[0067] To further illustrate the technical implementation of this invention, the following, in conjunction with the accompanying drawings and using a UAV RGB electronic photograph as an example, details the technical implementation process of an automated vegetation cover extraction method for field camera surveys. The overall technical flow is as follows: Figure 1 As shown. The specific implementation steps are as follows:

[0068] Beforehand, use a drone's field camera to acquire RGB electronic photos of the area to be investigated.

[0069] Input the RGB digital image to be processed.

[0070] In this step, input an RGB electronic photo of the drone to be processed, such as... Figure 2 As shown. In this embodiment, the image spatial resolution is 5cm, with a total of 5472 columns × 3648 rows of pixels. The numerical range of each band (red, green, and blue) is 0-255, and the data is in byte format.

[0071] Next, the greenness index ExG and color saturation S of the RGB digital photograph are calculated.

[0072] Specifically, based on the RGB electronic photograph to be processed, the greenness index ExG and color saturation S of each pixel are calculated.

[0073] For each pixel of an RGB digital photograph, the green index ExG and its corresponding color saturation S are calculated.

[0074] The formula for calculating ExG is:

[0075] , , (1)

[0076] , , (2)

[0077] (3)

[0078] In the formula, , and These represent the pixel values ​​for the red, green, and blue bands in an RGB digital photograph. , and These represent the maximum pixel values ​​for the red, green, and blue bands, respectively. , and These are the normalized pixel values ​​for the red, green, and blue bands, respectively.

[0079] The formula for calculating S is:

[0080] (4)

[0081] In the formula, , and These are the normalized values ​​of the red, green, and blue bands in the RGB electronic photograph (Equation 1). It is a minimum value used to avoid the case where the denominator is zero.

[0082] The ExG image and S feature image obtained in this step are as follows: Figure 3 and Figure 4 As shown.

[0083] Next, the threshold for segmenting vegetation is determined pixel by pixel using the color saturation S of each pixel, and the threshold is the optimal threshold.

[0084] Using the color saturation S of each pixel in the RGB digital photograph, we substitute it into the ExG adaptive thresholding model to determine the optimal ExG threshold for segmenting vegetation pixel by pixel. The ExG adaptive thresholding model is as follows:

[0085] (5)

[0086] In the formula, For pixel-by-pixel adaptive thresholding of ExG, This represents the saturation value of the current pixel. This is a parameterized model driven by training data. The only input parameter of this model is S, and the model can be represented as a linear or nonlinear function. Typical forms include, but are not limited to, linear models, quadratic polynomial models, and exponential models. The model parameters are determined by regression fitting through the training set data of the study area.

[0087] In this embodiment, the threshold adaptive model for the overgreen index is trained using 50 RGB electronic photographs (resolution 1-5cm) collected under different lighting conditions, vegetation cover, and complex land cover combinations. First, the overgreen index ExG and color saturation S are calculated for each pixel. Based on this data, vegetation and non-vegetation attributes are interpreted pixel-by-pixel, and the optimal threshold T for the overgreen index and the relationship between color saturation S are fitted using a regression method (e.g., Support Vector Machine Regression SVM) and the overgreen index ExG and color saturation S. The relationship between ExG and S is determined through scatter plots. Figure 5 Visualize the data and obtain the threshold adaptive model for the overgreen index in this embodiment:

[0088]

[0089] Using the color saturation S value of each pixel in the RGB electronic photograph to be processed, and combining it with the above-mentioned adaptive threshold model for the green index, the optimal segmentation threshold T for the green index ExG of each pixel is calculated.

[0090] Next, vegetation pixels were identified based on the greenness index ExG and pixel-by-pixel threshold.

[0091] For each pixel in the RGB digital photograph, its ExG value (denoted as ExG(x, y)) is compared with an adaptive threshold T to determine whether the pixel attribute is vegetation or non-vegetation. The determination criteria are as follows:

[0092] (6)

[0093] in, The results are for pixel classification, where 1 represents vegetation and 0 represents non-vegetation.

[0094] For each pixel in the RGB digital image to be processed, the calculated ExG value is compared with a threshold T to determine whether the pixel is vegetation. The final vegetation identification result is as follows: Figure 6 As shown, the black areas represent vegetation, and the white areas represent non-vegetation.

[0095] Finally, the proportion of vegetation pixels to the total number of pixels in the RGB electronic photograph is calculated to obtain the vegetation coverage (FVC).

[0096] By statistically analyzing the ratio of vegetation pixels to total pixels, the vegetation cover (FVC) is calculated. The FVC calculation formula is as follows:

[0097] (7)

[0098] In the formula, This represents the number of pixels in an RGB digital photograph that are identified as vegetation. This represents the total number of pixels.

[0099] The vegetation cover (FVC) is calculated by statistically analyzing the ratio of vegetation pixels to total pixels in an RGB digital photograph. In this case, the calculated vegetation cover of the RGB digital photograph is 0.68 (68%), with a recognition accuracy of 91%.

[0100] This invention enables automated extraction of vegetation cover from RGB electronic photographs. The method can determine the optimal ExG threshold for vegetation segmentation pixel-by-pixel in RGB electronic photographs imaged under different lighting conditions and terrain features, thereby achieving high-precision automatic extraction of vegetation cover (FVC). This facilitates the rapid and accurate acquisition of quadrat-scale vegetation cover data during field surveys. Its principle is simple, easy to implement, highly adaptable, highly automated, and widely applicable. It has the following characteristics:

[0101] (1) It has strong adaptability and is suitable for imaging RGB electronic photos under different lighting conditions and ground scenes;

[0102] (2) The principle is simple and easy to implement. The optimal segmentation threshold per pixel of the green index ExG is determined only by the color saturation S.

[0103] (3) High degree of automation. Once the ExG adaptive threshold model is established based on the training samples, the vegetation coverage can be extracted with high precision automatically in the study area.

[0104] (4) It has a wide range of applications and can be widely used in fields such as ecology, geography, and remote sensing.

[0105] The above description is merely one specific embodiment of the present invention. The ExG threshold adaptive model is not unique or fixed; its form includes, but is not limited to, a univariate linear model, and the model parameters may vary with changes in the training dataset. Therefore, the scope of protection of the present invention should not be limited to specific regression methods or parameter ranges. Any reasonable changes or substitutions made by those skilled in the art based on the technical content disclosed in this invention should be included within the scope of protection of this invention. Therefore, the scope of protection of this invention should be determined by the scope of the claims.

Claims

1. An automated method for extracting vegetation cover in field camera surveys, characterized in that, Includes the following steps: Input the RGB digital image to be processed; Calculate the greenness index and color saturation of RGB digital photographs; The threshold for segmenting vegetation is determined pixel by pixel using the overgreen index; Vegetation pixels are identified based on the greenness index and pixel-by-pixel threshold. The vegetation coverage is obtained by statistically analyzing the proportion of vegetation pixels to the total number of pixels in an RGB electronic photograph. In the step of determining the threshold for vegetation segmentation by using color saturation pixel by pixel to determine the green index, the adaptive model for the green index threshold is obtained by training RGB electronic photos collected under different lighting conditions, vegetation cover and complex land cover combination scenarios. The training steps include: Based on this data, vegetation and non-vegetation attributes are interpreted pixel by pixel; Calculate the greenness index ExG and color saturation S for each pixel; The relationship between the overgreen index threshold T and color saturation S was fitted using regression methods, and the adaptive threshold model for the overgreen index was obtained as follows: T = 0.21 × S + 0.06; In the step of determining the threshold for vegetation segmentation using color saturation pixel-by-pixel overgreen index: The threshold for segmenting vegetation using the overgreen index is determined pixel-by-pixel by color saturation. The threshold for segmenting vegetation using the overgreen index is a regression fitting function of color saturation; It also includes the step of: using the color saturation of each pixel in the RGB electronic photo to be processed, and combining it with the threshold adaptive model of the green index to calculate the segmentation threshold of the green index of each pixel.

2. The method for automated extraction of vegetation cover in field camera surveys according to claim 1, characterized in that, The values ​​of the red, green, and blue bands of the RGB electronic photograph range from 0 to 255, and are in byte format.

3. The method for automatically extracting vegetation cover for field camera surveys according to claim 1, characterized in that, In the steps of calculating the greenness index and color saturation of an RGB digital photograph: calculate the greenness index and color saturation of each pixel.

4. The method for automated extraction of vegetation cover in field camera surveys according to claim 1, characterized in that, In the step of identifying vegetation pixels based on the greenness index and pixel-by-pixel threshold, for each pixel of the RGB electronic photo to be processed, the calculated greenness index is compared with the threshold to determine whether the pixel is vegetation.

5. The method for automatically extracting vegetation cover for field camera surveys according to claim 1, characterized in that, The formula for calculating color saturation S is: In the formula, R ' G ' and B ' These are the normalized values ​​of the red, green, and blue bands in the RGB electronic photograph, respectively. ∈ is a minimum value used to avoid the case where the denominator is zero.

6. The method for automatically extracting vegetation cover for field camera surveys according to claim 1, characterized in that, For each pixel in an RGB digital photograph, its greenness index ExG, denoted as ExG(x, y), is compared with an adaptive threshold T to determine whether the pixel attribute is vegetation or non-vegetation. The determination criteria are as follows: Where f(x, y) is the pixel classification result, 1 represents vegetation and 0 represents non-vegetation.

Images