Green vegetation extraction method based on visible light wave band

Abstract

The invention discloses a green vegetation extraction method based on a visible light wave band, and relates to the technical field of satellite remote sensing. Comprising the following steps: acquiring an RGB image of a to-be-detected area, preprocessing the RGB image to obtain an orthoimage, segmenting and extracting the orthoimage to obtain a segmented image, selecting a green vegetation object and a green non-vegetation object from the segmented image, and determining pixel average values of the green vegetation object and the green non-vegetation object in different wavebands; and determining a green vegetation index and an optimal parameter based on the pixel average value of the different wavebands of the green vegetation object, and extracting a green vegetation image of the to-be-detected area based on the optimal parameter. According to the method, other green ground features can be prevented from being extracted mistakenly, and the purpose of accurately extracting green vegetation is achieved.

Description

Technical Field

[0001] This invention relates to the field of satellite remote sensing technology, and in particular to a method for extracting green vegetation based on the visible light band. Background Technology

[0002] Vegetation, atmosphere, water, and soil together constitute an ecosystem, with vegetation playing a crucial role as a producer in the entire ecological cycle and serving as an important indicator for assessing carbon storage, land cover, and ecosystem conditions. The size of vegetation cover has a significant impact on the temperature of ecosystems and biodiversity; accurately identifying green vegetation and understanding its spatial distribution is essential for investigating vegetation diversity and community structure. Remote sensing technology, with its rapid, real-time, and large-scale data acquisition capabilities, has become the primary means of vegetation monitoring. Unmanned aerial vehicle (UAV) remote sensing offers high flexibility, short image acquisition cycles, minimal weather interference, and the ability to carry visible light lenses, resulting in high spatial resolution images rich in information. However, the construction of the visible light band differential vegetation index has not adequately considered the differences between green non-vegetation features and green vegetation. When UAV RGB imagery contains green water bodies, green artificial grass, green basketball courts, etc., using the visible light band differential vegetation index to extract green vegetation can easily misclassify these non-vegetation green features as vegetation.

[0003] Therefore, providing a method for extracting green vegetation based on the visible light band to overcome the difficulties of existing technologies is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0004] In view of this, the present invention provides a method for extracting green vegetation based on the visible light band, which can avoid erroneously extracting other green features and achieve the goal of accurately extracting green vegetation.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] A method for extracting green vegetation based on the visible light band includes the following steps:

[0007] The RGB image of the region to be detected is acquired and preprocessed to obtain an orthophoto. The orthophoto is then segmented and extracted to obtain a segmented image.

[0008] In the segmented image, green vegetation objects and green non-vegetation objects are selected, and the average pixel values ​​of green vegetation objects and green non-vegetation objects in different bands are determined.

[0009] The green vegetation index and optimal parameters are determined based on the average pixel values ​​of different bands of the green vegetation object; the green vegetation image of the area to be detected is extracted based on the optimal parameters.

[0010] Optionally, obtaining orthophotos includes:

[0011] Using drones to obtain environmental images of the area to be inspected and the corresponding laser point clouds;

[0012] The three-dimensional coordinates, camera parameters, and acquisition time of the laser point cloud are obtained. The laser points are then projected onto the environmental space to determine the spatial coordinates of the environmental pixels.

[0013] The spatial point weights are determined based on spatial coordinates and camera parameters, and the spatial points are projected onto the environmental image. The target pixel value is then determined by combining the spatial coordinates and spatial point weights.

[0014] Generate orthophotos based on target pixel values.

[0015] Optionally, determining the spatial point weights includes:

[0016] The solar altitude angle and solar azimuth angle are determined based on spatial coordinates and acquisition time, and the solar reflection vector is determined based on the solar altitude angle and solar azimuth angle.

[0017] The camera's light-sensing vector is determined based on spatial coordinates and camera parameters, and the probability factor of overexposure is determined by combining the solar reflection vector.

[0018] Spatial point weights are determined based on probability factors.

[0019] Optionally, segmentation extraction includes using the orthophoto as three layers (R, G, and B) as segmentation layers, performing parameter filtering and multi-scale segmentation respectively, to obtain segmented images.

[0020] Optionally, determining the pixel average includes selecting green vegetation images and green non-vegetation images in the segmented image, and determining the mean of the pixel average of green non-vegetation objects in the red band R, the mean of the pixel average in the green band G, and the mean of the pixel average in the blue band B, as well as the mean of the pixel average of green vegetation objects in the red band R, the mean of the pixel average in the green band G, and the mean of the pixel average in the blue band B.

[0021] Optionally, the green vegetation index is determined based on the mean of the average pixel value of green vegetation ground objects in the red band R, the average pixel value in the green band G, and the average pixel value in the blue band B.

[0022] The optimal parameters are determined based on the mean values ​​of the pixel averages of green non-vegetated ground objects in the red band R, the green band G, and the blue band B.

[0023] Optionally, extracting green vegetation images includes determining optimal parameters, generating an image threshold based on the optimal parameters, and extracting green vegetation based on the image threshold.

[0024] As can be seen from the above technical solution, compared with the prior art, the present invention provides a method for extracting green vegetation based on the visible light band, which has the following beneficial effects: 1) The present invention combines laser point cloud and environmental image methods, so that the generated orthophoto map contains richer texture information, saves the cost of generating orthophoto map, and improves the accuracy of subsequent information extraction; 2) The present invention can avoid erroneously extracting other green features and achieve the purpose of accurately extracting green vegetation. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0026] Figure 1 This is a flowchart of a method for extracting green vegetation based on the visible light band, as disclosed in this invention. Detailed Implementation

[0027] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] Reference Figure 1 As shown, this invention discloses a method for extracting green vegetation based on the visible light band, comprising the following steps:

[0029] The RGB image of the region to be detected is acquired and preprocessed to obtain an orthophoto. The orthophoto is then segmented and extracted to obtain a segmented image.

[0030] In the segmented image, green vegetation objects and green non-vegetation objects are selected, and the average pixel values ​​of green vegetation objects and green non-vegetation objects in different bands are determined.

[0031] The green vegetation index and optimal parameters are determined based on the average pixel values ​​of different bands of the green vegetation object; the green vegetation image of the area to be detected is extracted based on the optimal parameters.

[0032] Furthermore, obtaining orthophotos includes:

[0033] Using drones to obtain environmental images of the area to be inspected and the corresponding laser point clouds;

[0034] The three-dimensional coordinates, camera parameters, and acquisition time of the laser point cloud are obtained. The laser points are then projected onto the environmental space to determine the spatial coordinates of the environmental pixels.

[0035] The spatial point weights are determined based on spatial coordinates and camera parameters, and the spatial points are projected onto the environmental image. The target pixel value is then determined by combining the spatial coordinates and spatial point weights.

[0036] Generate orthophotos based on target pixel values.

[0037] Furthermore, obtaining environmental images of the area to be detected involves using a drone equipped with a camera and LiDAR to pre-traverse the area to acquire frames of environmental images and corresponding LiDAR point clouds. During the pre-traversal, the camera on the drone acquires environmental images, including those of the road, around the vehicle at fixed time intervals, while the LiDAR acquires LiDAR point clouds around the vehicle at fixed time intervals. Environmental images and LiDAR point clouds acquired at similar time points are then correlated.

[0038] Furthermore, determining the spatial point weights includes:

[0039] The solar altitude angle and solar azimuth angle are determined based on spatial coordinates and acquisition time, and the solar reflection vector is determined based on the solar altitude angle and solar azimuth angle.

[0040] The camera's light-sensing vector is determined based on spatial coordinates and camera parameters, and the probability factor of overexposure is determined by combining the solar reflection vector.

[0041] The weights of spatial points are determined based on probability factors.

[0042] Specifically, for all spatial points of overlapping pixels, the target pixel value is calculated by weighting according to the weights, and finally an orthophoto image without overexposure and distortion is generated.

[0043] Furthermore, the segmentation extraction involves using the orthophoto as three layers (R, G, and B) as segmentation layers, performing parameter filtering and multi-scale segmentation on each layer to obtain a segmented image.

[0044] Furthermore, determining the pixel average value includes selecting green vegetation images and green non-vegetation images in the segmented image, and determining the mean of the pixel average value of green non-vegetation objects under the red band R, the green band G, and the blue band B, as well as the mean of the pixel average value of green vegetation objects under the red band R, the green band G, and the blue band B.

[0045] Furthermore, the green vegetation index is determined based on the mean of the pixel average value of green vegetation ground objects in the red band R, the green band G, and the blue band B.

[0046] The optimal parameters are determined based on the mean values ​​of the pixel averages of green non-vegetated ground objects in the red band R, the green band G, and the blue band B.

[0047] Specifically, the expression for the green vegetation index is: Where x1, x2, and x3 are parameters, G g R is the mean of the pixel averages of green vegetation objects in the green band G. g B is the mean of the pixel values ​​of green vegetation features in the red band R. g This is the mean of the pixel averages of green vegetation features in the blue band B.

[0048] Furthermore, determining the optimal parameters includes:

[0049] The mean values ​​of the pixel average values ​​of each green non-vegetation ground object in the red band R, the green band G, and the blue band B are used to calculate the index A according to the expression of the adaptive green vegetation index GVI.

[0050] Generate a preset set of GVI parameters based on index A, and assign initial values;

[0051] Based on the preset GVI parameter set, calculate the set of differences between A and GVI in the parameter set set, and establish a mapping relationship between the parameter set set and the set of differences.

[0052] Based on the set of differences and the mapping relationship, an iterative operation is performed starting with the given initial value to update the GVI parameter set until the optimal parameters of the GVI are generated.

[0053] Specifically, the expression for A is: Among them, G n R is the mean of the pixel values ​​of green non-vegetation ground features in the green band G. n B is the mean of the pixel values ​​of green non-vegetated ground features in the red band R. n This is the mean of the pixel averages of green non-vegetated ground features in the blue band B.

[0054] Furthermore, extracting green vegetation images includes determining optimal parameters, generating an image threshold based on the optimal parameters, and extracting green vegetation based on the image threshold.

[0055] Specifically, image thresholding can be achieved using a bimodal histogram-based thresholding method or Otsu's thresholding method.

[0056] The bimodal histogram method is a commonly used thresholding method. If the histogram of an image includes two obvious main peaks, it indicates that the target area to be extracted from the image has a high degree of distinction from the background area. That is, the target objects to be extracted are concentrated in one peak, while the background objects are concentrated in the other peak of the histogram. The trough between the two peaks can be considered as the optimal thresholding point. There are fewer pixels at this position. The threshold of each vegetation index can be determined based on the position of the trough, thereby achieving the distinction between vegetation and non-vegetation.

[0057] Otsu's thresholding method, also known as the Otsu method, divides an image into two groups by setting a threshold: one group of gray levels corresponds to the target, and the other group corresponds to the background. The greater the inter-class variance between the target and the background, the higher the distinction between them. The distinction between the target and the background reaches its maximum when the inter-class variance is at its maximum, which effectively avoids misclassification and omission between the target and the background.

[0058] Furthermore, green vegetation is extracted based on image thresholding. For each pixel in the orthophoto, its GVI value is calculated and compared with the optimal threshold. When the GVI meets the threshold requirement, it is determined to be green vegetation; otherwise, it is considered non-vegetation. Noise points are removed through morphological operations, and holes are filled using a region growing algorithm to finally generate a complete green vegetation image.

[0059] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for extracting green vegetation based on the visible light band, characterized in that, Includes the following steps: The RGB image of the region to be detected is acquired and preprocessed to obtain an orthophoto. The orthophoto is then segmented and extracted to obtain a segmented image. Select green vegetation objects and green non-vegetation objects in the segmented image, and determine the average pixel value of green vegetation objects and green non-vegetation objects in different bands. The green vegetation index and optimal parameters are determined based on the average pixel values ​​of different bands of the green vegetation object; the green vegetation image of the area to be detected is extracted based on the optimal parameters.

2. The method for extracting green vegetation based on the visible light band according to claim 1, characterized in that, Obtaining orthophotos includes: Using drones to obtain environmental images of the area to be inspected and the corresponding laser point clouds; The three-dimensional coordinates, camera parameters, and acquisition time of the laser point cloud are obtained. The laser points are then projected onto the environmental space to determine the spatial coordinates of the environmental pixels. The spatial point weights are determined based on spatial coordinates and camera parameters, and the spatial points are projected onto the environmental image. The target pixel value is then determined by combining the spatial coordinates and spatial point weights. Generate orthophotos based on target pixel values.

3. The method for extracting green vegetation based on the visible light band according to claim 2, characterized in that, Determining the spatial point weights includes: The solar altitude angle and solar azimuth angle are determined based on spatial coordinates and acquisition time, and the solar reflection vector is determined based on the solar altitude angle and solar azimuth angle. The camera's light-sensing vector is determined based on spatial coordinates and camera parameters, and the probability factor of overexposure is determined by combining the solar reflection vector. Spatial point weights are determined based on probability factors.

4. The method for extracting green vegetation based on the visible light band according to claim 1, characterized in that, Segmentation and extraction involves using the orthophoto as three layers (R, G, and B) for segmentation, performing parameter filtering and multi-scale segmentation on each layer to obtain segmented images.

5. The method for extracting green vegetation based on the visible light band according to claim 1, characterized in that, Determining the pixel average involves selecting green vegetation images and green non-vegetation images in the segmented image, and determining the mean of the pixel average of green non-vegetation objects in the red band R, the green band G, and the blue band B, as well as the mean of the pixel average of green vegetation objects in the red band R, the green band G, and the blue band B.

6. The method for extracting green vegetation based on the visible light band according to claim 5, characterized in that, The green vegetation index is determined based on the mean of the pixel average values ​​of green vegetation objects in the red band R, the green band G, and the blue band B. The optimal parameters are determined based on the mean of the pixel average values ​​of green non-vegetated objects in the red band R, the green band G, and the blue band B.

7. The method for extracting green vegetation based on the visible light band according to claim 1, characterized in that, Extracting green vegetation from images involves determining optimal parameters, generating an image threshold based on the optimal parameters, and extracting green vegetation based on the image threshold.

Images