Method for generating representative blade based on blade contour analysis

Abstract

The invention belongs to the field of plant morphology, and discloses a method for generating a representative leaf based on leaf contour analysis, which comprises the following steps: S1, selecting a target tree species and collecting leaves; s3, preprocessing the grid image data of the blade contour; s4, superposing the blade contours to generate a density map; and S5, reconstructing a representative blade contour. According to the invention, through high-precision digital image processing, form registration and density threshold optimization algorithms, precise reconstruction of the blade contour and generation of the representative blade are systematically realized. According to the method, optimization is carried out in the aspects of standardized leaf collection, automatic morphological analysis and optimal threshold calculation, so that quantification of leaf morphological characteristics is more accurate and higher in stability, and the method has remarkable universality and is suitable for various tree species and different ecological environments.

Description

Technical Field

[0001] The present invention relates to the field of plant morphology, and specifically to a method for generating representative leaves based on leaf contour analysis. Background Art

[0002] Leaf morphology is an important phenotypic trait for plants to adapt to the environment and affects ecological processes such as photosynthesis, transpiration, and gas exchange. However, due to differences in tree diameter at breast height (DBH), canopy levels (upper, middle, and lower layers), and directions (east, south, west, and north) of tree species, leaf morphology has high spatial variability. Existing research mainly relies on direct measurement or manual image processing, making it difficult to maintain high precision and consistency in large-scale data processing. In addition, due to the complexity of leaf margin features (serrations, lobes, etc.) of different leaf species, the existing methods for constructing representative leaves lack an optimized selection of pixel density thresholds, resulting in deficiencies in the accuracy and adaptability of the reconstructed leaves.

[0003] Based on digital image processing, pixel density mapping, and optimal threshold calculation, the present invention proposes an automated method that can efficiently and standardly extract leaf contours and reconstruct representative leaves through the superposition of morphological features, so as to improve the reconstruction accuracy and ensure the applicability to different tree species. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for generating representative leaves based on leaf contour analysis to solve the problems raised in the above background art.

[0005] To achieve the above purpose, the present invention provides the following technical solutions:

[0006] A method for generating representative leaves based on leaf contour analysis, comprising the following steps:

[0007] Step S1, select a target tree species and collect leaves:

[0008] Step S2, obtain a raster image of a single leaf contour;

[0009] Step S3, preprocess the leaf contour raster image data;

[0010] Step S4, superpose the leaf contours to generate a density map;

[0011] Step S5, reconstruct the representative leaf contour.

[0012] Further, the step S1 includes:

[0013] Step S11, select healthy and representative sample trees from the target arbor tree species, and the number of sample trees is not less than 10;

[0014] Step S12: Divide the canopy of each sample tree into upper, middle, and lower layers. The middle layer is further divided into four azimuths: east, west, south, and north. A total of 6 sampling positions are set for each sample tree.

[0015] Step S13: Collect the same number of healthy and undamaged leaves at each sampling position.

[0016] Furthermore, in step S13, the number of collected leaves at each sampling position is not less than 10.

[0017] Furthermore, in step S1, the period for leaf collection is selected at the stable stage after the completion of the leaf phenology of leaf expansion.

[0018] Furthermore, step S2 includes:

[0019] Step S21: Place the collected leaves flat on the scanner without overlapping and scan them into digital raster images.

[0020] Step S22: Extract the images of individual leaf contours from the digital raster images and store them.

[0021] Furthermore, in step S21, the resolution of the digital raster image is not less than 300 PPI.

[0022] Furthermore, step S3 includes:

[0023] Step S31: Binarize each leaf contour image so that the pixel value of the leaf area is 1 and the pixel value of the background is 0. The expression is:

[0024]

[0025] In the formula, I(x, y) is the value after binarization of the original image, f(x, y) is the gray value of the original image, (x, y) is the coordinate of the original image, and T is the threshold.

[0026] Step S32: Normalize the image to unify the width and height of the leaf contour images. The expression is:

[0027] W fix = max(W)

[0028] H fix = max(H)

[0029] In the formula, W fix is the unified width of the image, H fix is the unified height of the image, W is the width of the original image, H is the height of the original image, and max is the maximum value function.

[0030] Step S33, morphological registration, align all leaf contours so that they have a consistent central position. The expression for the centroid coordinates of the leaf contour is:

[0031]

[0032] In the formula, C x is the abscissa of the centroid of the leaf contour image, x i is the abscissa of the leaf contour image, C y is the ordinate of the centroid of the leaf contour image, y i is the ordinate of the leaf contour image, and N is the total number of pixels in the leaf contour image.

[0033] Further, the said S4 includes:

[0034] Overlay the preprocessed leaf contour images to generate a pixel density map. The value of each pixel in the density map represents the frequency of the pixel being marked as the leaf area, ranging from 0 to 1. The expression is:

[0035]

[0036] In the formula, D(x,y) is the value of the pixel point (x,y) in the density map, I j (x,y) is the value of the pixel point (x,y) in the j-th leaf contour image, and M is the number of leaf contour images.

[0037] Further, the said S5 includes:

[0038] Step S51, set the initial density threshold, and the expression is:

[0039]

[0040] In the formula, T0 is the initial density threshold, and M is the number of leaf contour images;

[0041] Step S52, gradually adjust the density threshold and calculate the area of the leaf area corresponding to the current threshold. The expression is:

[0042]

[0043] In the formula, ΔT is the change in the density threshold;

[0044] T = T0 + i×ΔT

[0045] In the formula, T is the density threshold, T0 is the initial density threshold, i is a non-negative integer, and its value range is 0, 1, 2, …, M - 1, and ΔT is the change in the density threshold;

[0046] According to the density map produced in step S4, pixels with density values greater than the threshold T are marked as the leaf area, and the expression is:

[0047]

[0048] In the formula, A T is the leaf contour area corresponding to the current threshold, is the indicator function, which takes 1 when D(x, y) ≥ T, and 0 otherwise;

[0049] Step S53: Calculate the leaf area under different density values and determine the optimal density threshold T opt , and the expression:

[0050]

[0051] In the formula, A T is the leaf contour area corresponding to the current threshold, and A j is the area of the contour of the j-th leaf image, the value of T when the formula reaches the minimum;

[0052] Step S54: Reconstruct the representative leaf contour with the leaf area corresponding to the optimal density threshold T opt .

[0053] Furthermore, the adjustment method of the density value in step S52 is that the change amount each time is 1 / the number of all leaf contour images.

[0054] Compared with the prior art, the present invention systematically realizes the precise reconstruction of the leaf contour and the generation of representative leaves through high-precision digital image processing, morphological registration, and density threshold optimization algorithms. The present invention has been optimized in terms of standardized leaf collection, automated morphological analysis, and optimal threshold calculation, making the quantification of leaf morphological characteristics more accurate, more stable, and having significant universality, applicable to various tree species and different ecological environments.

[0055] The core innovation points of the present invention include: Morphological standardization processing: Using a morphological registration algorithm to align the centroid and correct the rotation of the leaf contour, ensuring data consistency and improving calculation accuracy; Density mapping and optimization: Automatically extracting representative leaf contours through pixel density calculation and optimal threshold selection, avoiding the manual intervention error of traditional methods; Adapting to different leaf morphological characteristics: Applicable to tree species with different leaf margin morphologies (serrated, lobed, smooth, etc.), improving the wide applicability of the method; Stable calculation results: Through algorithm optimization, efficient and batch processing of leaf data is achieved, reducing morphological analysis errors and improving data reproducibility. This method not only has important value in basic research such as plant classification, population dynamics analysis, and ecological environment monitoring, but also can be widely applied to forest resource management, carbon sink assessment, and ecosystem service function analysis, providing innovative technical support for the scientific management of ecosystems, the optimization of tree species genetic resources, and sustainable forest management. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 FIG. is a schematic flowchart of a method for generating a representative leaf based on leaf contour analysis provided by an embodiment of the present invention;

[0057] Figure 2 FIG. is an example contour diagram of the leaf of the target tree species Tilia amurensis provided by an embodiment of the present invention;

[0058] Figure 3 FIG. is a reconstructed representative leaf contour diagram of the target tree species Tilia amurensis provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0059] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0060] Embodiment

[0061] The present invention selects Tilia amurensis in the broad-leaved Korean pine forest in Changbai Mountain as the target tree species, and the specific steps are as follows:

[0062] Step S1, Selecting the target tree species and collecting leaves:

[0063] Step S11, In the broad-leaved Korean pine forest in Changbai Mountain, select healthy and representative Tilia amurensis sample trees, and the number of sample trees is not less than 10.

[0064] Preferably, according to the diameter at breast height of the Tilia amurensis tree species in the area, divide the diameter at breast height grade according to a scale of 10 cm, and select the Tilia amurensis sample trees.

[0065] Step S12: Divide the canopy of each sample tree into upper, middle, and lower layers, where the middle layer is further divided into east, west, south, and north directions, and a total of 6 sampling positions are set for each sample tree.

[0066] Step S13: Collect the same number of healthy and undamaged leaves at each sampling position.

[0067] Preferably, the number of leaves collected at each of the said sampling positions is not less than 10.

[0068] Preferably, the leaf collection period is selected at the stable stage after the leaf phenology of leaf unfolding is completed.

[0069] Optionally, for the collection of leaves of tall trees, it can be carried out through fresh fallen logs.

[0070] Step S2: Obtaining a single leaf contour raster image:

[0071] Step S21: Place the collected leaves flat on the scanner without overlapping and scan them into digital raster images.

[0072] Preferably, the resolution of the digital raster image is not less than 300 PPI.

[0073] Step S22: Extract the images of single leaf contours from the digital raster images and store them. An example contour of the leaves of the target tree species Tilia amurensis is as Figure 2 shown.

[0074] Step S3: Preprocessing of leaf contour image data:

[0075] Step S31: Binarize each leaf contour image so that the pixel value of the leaf area is 1 and the pixel value of the background is 0. The expression is:

[0076]

[0077] In the formula, I(x,y) is the value after binarization of the original image, f(x,y) is the gray value of the original image, (x,y) is the coordinate of the original image, and T is the threshold.

[0078] Step S32: Unify all leaf contour images to a fixed width and height for subsequent calculation and alignment;

[0079] W fix =max(W)

[0080] H fix =max(H)

[0081] In the formula, W fix is the width after image unification, and H fixLet \(U\) be the unified height of the images, \(W\) be the width of the original image, \(H\) be the height of the original image, and \(max\) be the maximum value function.

[0082] Preferably, the fixed size can accommodate all the blades;

[0083] Step S33: Align all the blade contour images using a conventional shape registration algorithm so that they have a consistent center position and rotation angle.

[0084] Preferably, extract the centroid of the blade as the alignment reference point. The expression for the centroid coordinates of the blade contour is:

[0085]

[0086] where \(C_x\) x is the abscissa of the centroid of the blade contour image, \(x\) i is the abscissa of the blade contour image, \(C_y\) y is the ordinate of the centroid of the blade contour image, \(y\) i is the ordinate of the blade contour image, and \(N\) is the total number of pixels in the blade contour image;

[0087] Preferably, correct the rotation angle using the directions of the leaf length and width.

[0088] Step S4: Generate a density map by superimposing the blade contours:

[0089] Superimpose the preprocessed blade contour images to generate a pixel density map. The value of each pixel in the density map represents the frequency of the pixel being marked as the blade area, ranging from 0 to 1;

[0090]

[0091] where \(D(x,y)\) is the value of the pixel point \((x,y)\) in the density map, \(I_j(x,y)\) j is the value of the pixel point \((x,y)\) in the \(j\)-th blade contour image, and \(M\) is the number of blade contour images.

[0092] Step S5: Reconstruct the representative blade contour:

[0093] Step S51: Set the initial density threshold:

[0094]

[0095] where \(T_0\) is the initial density threshold and \(M\) is the number of blade contour images.

[0096] Step S52: Gradually adjust the density threshold, with each change being \(1 / \) the number of all blade contour images. Calculate the area of the blade region corresponding to the current threshold. The expression is:

[0097]

[0098] In the formula, ΔT is the change in the density threshold, and M is the number of blade contour images.

[0099] T = T0 + i×ΔT

[0100] In the formula, T is the density threshold, T0 is the initial density threshold, i is a non - negative integer, and its value range is 0, 1, 2, …, M - 1, and ΔT is the change in the density threshold;

[0101] For the density map produced according to step S4, pixels with density values greater than the threshold T are marked as the blade area:

[0102]

[0103] In the formula, A T is the blade contour area corresponding to the current threshold, is the indicator function, which takes 1 when D(x, y)≥T, and 0 otherwise.

[0104] Step S53: Calculate the blade area at different density values. When the blade area is closest to the average area of all blade contour images, determine the optimal density value T opt , the optimal density value T of the leaves of Tilia amurensis Rupr. opt is 0.39, and the expression is:

[0105]

[0106] In the formula, A T is the blade contour area corresponding to the current threshold, A j is the area of the contour of the j - th blade image, M is the number of blade contour images, the value of T when the formula reaches the minimum;

[0107] Step S54: Reconstruct the representative blade contour with the blade area corresponding to the optimal density threshold T opt The representative blade contour is as Figure 3 shown.

[0108] The above embodiments have elaborated in detail the specific implementation process of the leaf contour analysis and representative leaf generation method of the present invention on different plant tree species, demonstrating the key technical steps such as leaf collection, image processing, contour superposition and reconstruction. Through a standardized process and high-precision data processing, the present invention has successfully achieved the accurate quantification of leaf morphological characteristics and the generation of representative contours, greatly improving the accuracy and reliability of leaf morphological analysis. This method has strong universality, can adapt to different plant tree species and habitat conditions, and is easy to operate with high data processing efficiency. The present invention not only provides innovative technical support for plant classification, population dynamics research and ecological environment monitoring, but also provides a reliable tool for ecosystem function assessment and plant morphology research, with broad application prospects.

[0109] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for generating a representative blade based on blade profile analysis, characterized in that, It includes the following steps: Step S1: Select the target tree species and collect leaves: Step S2: Obtain the raster image of the single leaf contour; Step S3: Preprocess the data of the raster image of the leaf contour; Step S4: Overlay the leaf contours to generate a density map; Step S5: Reconstruct the representative leaf contour.

2. The method for generating a representative blade based on blade profile analysis according to claim 1, wherein The said Step S1 includes: Step S11: Select healthy and representative sample trees from the target arbor species, and the number of sample trees is not less than 10; Step S12: Divide the canopy of each sample tree into upper, middle, and lower layers, and the middle layer is further divided into east, west, south, and north directions. A total of 6 sampling positions are set for each sample tree; Step S13: Collect the same number of healthy and undamaged leaves at each sampling position.

3. A method for generating a representative blade based on blade profile analysis according to claim 2, characterized in that In the said Step S13, the number of collected leaves at each sampling position is not less than 10.

4. A method for generating a representative blade based on blade profile analysis according to claim 1, wherein, In the said Step S1, the period of leaf collection is selected at the stable stage after the completion of the leaf phenology of leaf expansion.

5. A method for generating a representative blade based on blade profile analysis according to claim 1, characterized in that The said Step S2 includes: Step S21: Lay the collected leaves flat on the scanner without overlapping and scan them into digital raster images; Step S22: Extract the image of the single leaf contour from the digital raster image and store it.

6. A method for generating a representative blade based on blade profile analysis according to claim 5, characterized in that, In the said Step S21, the resolution of the digital raster image is not less than 300 PPI.

7. A method for generating a representative blade based on blade profile analysis according to claim 1, characterized in that, The said Step S3 includes: Step S31: Binarize each leaf contour image so that the pixel value of the leaf area is 1 and the pixel value of the background is 0. The expression is: In the formula, I(x,y) is the value after binarization of the original image, f(x,y) is the gray value of the original image, (x,y) is the coordinate of the original image, and T is the threshold; Step S32: Image normalization, unify the width and height of the leaf contour image. The expression is: W fix = max(W) H fix = max(H) Where, W fix is the width after image unification, H fix is the height after image unification, W is the width of the original image, H is the height of the original image, and max is the maximum value function; Step S33: Morphological registration, align all leaf contours to make them have a consistent center position. The expression of the centroid coordinates of the leaf contour is: Where C x is the abscissa of the centroid of the blade profile image, x i is the abscissa of the blade profile image, C y is the ordinate of the centroid of the blade profile image, y i is the ordinate of the blade profile image, and N is the total number of pixels of the blade profile image.

8. A method for generating a representative blade based on blade profile analysis according to claim 1, wherein The said S4 includes: Overlay the preprocessed leaf contour images to generate a pixel density map. The value of each pixel in the density map represents the frequency of the pixel being marked as the leaf area, and the range is from 0 to 1. The expression is: where D(x, y) is the value of the pixel point (x, y) in the density map, and I j (x, y) is the value of the pixel point (x, y) in the j-th blade contour image, and M is the number of blade contour images.

9. A method for generating a representative blade based on blade profile analysis according to claim 8, characterized in that The said S5 includes: Step S51: Set the initial density threshold. The expression is: In the formula, T0 is the initial density threshold and M is the number of leaf contour images; Step S52: Gradually adjust the density threshold and calculate the area of the leaf area corresponding to the current threshold. The expression is: In the formula, ΔT is the change amount of the density threshold; T = T0 + i×ΔT In the formula, T is the density threshold, T0 is the initial density threshold, i is a non-negative integer, and the value range is 0, 1, 2, …, M - 1, and ΔT is the change amount of the density threshold; According to the density map produced in Step S4, mark the pixels with density values greater than the threshold T as the leaf area. The expression is: where A T is the blade profile area corresponding to the current threshold, II is the indicator function, taking 1 when D(x, y) ≥ T and 0 otherwise; Step S53: Calculate the leaf area at different density values and determine the optimal density threshold T opt , expression: Where A T is the blade profile area corresponding to the current threshold, and A j is the area of the contour of the j-th blade image, is the value of T when the formula reaches the minimum value; Step S54: Reconstruct the representative blade profile for the blade area corresponding to the optimal density threshold T opt ​ 10. A method for generating a representative blade based on blade profile analysis according to claim 1, characterized in that, In the said Step S52, the adjustment method of the density value is that the change amount each time is 1 divided by the number of all leaf contour images.

Images