Blade phenotype parameter measurement method and system based on image processing technology

Abstract

The invention is suitable for the technical field of plant phenotypic parameter measurement, and provides a leaf phenotypic parameter measurement method and system based on an image processing technology, and the method comprises the following steps: S1, collecting a lemon leaf image, and carrying out the preprocessing of the image, and obtaining a binary contour image of a leaf; s2, extracting the length and the width of the blade by adopting a minimum enclosing rectangle method or an end point method; s3, adopting a Freeman chain code to track the blade contour, and calculating the perimeter of the blade; s4, calculating the leaf area by adopting a pixel number method; s5, measuring a blade tip included angle and a blade base included angle of the blade; the method can rapidly and accurately measure the phenotypic parameters of the lemon leaves, provides powerful technical support for lemon germplasm resource research, variety improvement and breeding work, and has a wide application prospect.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of plant phenotype parameter measurement, in particular to a leaf phenotype parameter measurement method and system based on image processing technology. BACKGROUND

[0002] Leaf is an important organ for photosynthesis and transpiration of plants, and its phenotype parameters (such as length, width, circumference, area, leaf tip angle, leaf base angle, etc.) can reflect the growth condition, genetic characteristics and environmental adaptability of plants. In lemon breeding and germplasm research, accurate and efficient measurement of leaf phenotype parameters is of great significance for evaluating the quality of varieties and screening excellent strains.

[0003] Traditional leaf phenotype parameter measurement methods mainly rely on manual measurement, such as using a vernier caliper to measure length and width, and using a protractor to measure the angle. This method is not only time-consuming and labor-intensive, but also easily affected by human factors, with poor measurement accuracy and repeatability. With the development of image processing technology, image-based leaf phenotype parameter measurement methods have gradually become a research hotspot. However, existing methods still have deficiencies in measurement accuracy, adaptability and automation, for example, the traditional minimum circumscribed rectangle method has large error in measuring the length and width of irregular lemon leaves; the measurement method of leaf tip and leaf base angle is not perfect, and it is difficult to meet the actual application requirements.

[0004] Therefore, in view of the above status, it is urgent to provide a leaf phenotype parameter measurement method and system based on image processing technology to overcome the deficiencies in current practical applications. SUMMARY

[0005] The purpose of the present application is to provide a leaf phenotype parameter measurement method and system based on image processing technology, which effectively solves the problems in the background art.

[0006] The present application is implemented as follows: a leaf phenotype parameter measurement method based on image processing technology, the method comprising the following steps: Step S1: acquiring a lemon leaf image and pre-processing the image to obtain a binary contour image of the leaf; Step S2: extracting the length and width of the leaf using the minimum circumscribed rectangle method or the endpoint method; Step S3: tracking the leaf contour using the Freeman chain code and calculating the leaf circumference; Step S4: calculating the leaf area using the pixel number method; Step S5: measuring the leaf tip angle and leaf base angle.

[0007] As a further scheme of the present application: in step S1, the preprocessing comprises image graying, threshold segmentation and morphological processing to remove image noise and highlight the leaf blade profile.

[0008] As a further scheme of the present application: in step S2, the specific steps of extracting the length and width of the leaf blade by the end point method comprise: The positions of the leaf tip and leaf base are determined by searching the global minimum and maximum column coordinate values of the leaf blade profile pixel points, and the length of the line segment from the leaf tip to the leaf base is calculated as the length of the leaf blade; A perpendicular line is made to the line segment, and the two points of intersection with the leaf blade profile are taken, and the maximum value of the line segment length between the two points is taken as the width of the leaf blade; In the calculation of the length and width of the leaf blade, the following formula is used for size calibration: ; Wherein, L leaf represents the length or width of the leaf blade, L ref represents the actual length of the standard reference, P leaf represents the total number of pixels of the leaf blade length or width in the leaf blade region, and P ref represents the total number of pixels of the reference length.

[0009] As a further scheme of the present application: in step S3, the Freeman chain code is used to track the leaf blade profile, and the perimeter of the leaf blade is calculated, which specifically comprises: The leaf blade profile is tracked clockwise, the distance between the profile points is determined according to the chain code value, and the total number of pixels of the perimeter is accumulated to calculate the actual perimeter of the leaf blade in combination with the total number of pixels and the actual length of the standard reference.

[0010] As a further scheme of the present application: in step S3, when calculating the perimeter, for the inclined direction profile points with odd chain code values, the distance between the two points is 1 pixel unit; For the horizontal or vertical direction profile points with even chain code values, the distance between the two points is pixel units.

[0011] As a further scheme of the present application: in step S4, the pixel number method is used to calculate the area of the leaf blade, which specifically comprises: The total number of pixels with color value 1 in the leaf blade region in the binary profile image is counted, and the actual area of the leaf blade is calculated in combination with the total number of pixels and the actual area of the standard reference; The calculation formula of the actual area of the leaf blade is: ; Wherein, S leaf represents the area of the leaf blade, S ref represents the actual area of the standard reference, P leafP represents the total number of pixels in the blade area ref P represents the total number of pixels in the blade area

[0012] As a further aspect of the present application: in step S5, the tip and base angles of the blade are measured, specifically comprising: Determine the relative positions of the tip and base and unify the orientation; Extract the upper and lower edge contour coordinates at one-third of the blade, and perform 2D linear fitting to obtain two straight line equations; Calculate the intersection of the two straight lines, and take one point on each of the upper and lower contour lines, and calculate the tip and base angles by the cosine law.

[0013] As a further aspect of the present application: in step S5, the method for determining the relative positions of the tip and base is: taking the left and right end points of the blade contour as the coordinate zero points, finding the regions at one-tenth of the left and right ends of the image, calculating the blade pixel areas in the two regions and comparing them; if the left end area is greater than the right end area, it is determined that the tip is to the right, and the image is vertically flipped to make the tip to the left; if the left end area is less than the right end area, it is determined that the tip is to the left.

[0014] As a further aspect of the present application: the method further comprises precision evaluation of the measurement results, and the evaluation indexes include mean absolute error (MAE), root mean square error (RMSE) and average determination coefficient (R²).

[0015] The lemon leaf phenotype parameter measurement system based on image processing technology runs the method as described above, and the system comprises: An image acquisition module for acquiring lemon leaf images; A preprocessing module for preprocessing the acquired images to obtain binary contour images of the leaves; A parameter extraction module for extracting the length, width, perimeter, area, tip angle and base angle of the leaves; A data storage and display module for storing and displaying the extracted leaf phenotype parameters.

[0016] Compared with the prior art, the present application has the following advantages: High measurement efficiency: the image processing technology is used to realize automatic extraction of leaf phenotype parameters, without the need for manual measurement one by one, greatly improving the measurement efficiency.

[0017] High measurement accuracy: through the combination of various methods such as endpoint method, Freeman chain code tracking and pixel number method, as well as strict size calibration and error analysis, the high accuracy of the measurement results is ensured, and good consistency with the true value is achieved.

[0018] Strong adaptability: can adapt to different shapes and sizes of lemon leaves, and can accurately measure the phenotype parameters of irregularly shaped leaves.

[0019] High degree of automation: from image acquisition to parameter extraction, storage and display, the whole process automation is realized, human intervention is reduced, and the objectivity and repeatability of measurement are improved. BRIEF DESCRIPTION OF DRAWINGS

[0020] In order to more clearly illustrate the specific embodiments of the present application or the technical solutions in the prior art, the drawings needed in the specific embodiments or the prior art description will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application, and those skilled in the art can obtain other drawings according to these drawings without creative labor.

[0021] Figure 1 It is a part of the original image of the lemon leaf sample.

[0022] Figure 2 It is a schematic diagram of extracting the length and width of the leaf by the minimum circumscribed rectangle method.

[0023] Figure 3 It is a schematic diagram of extracting the length and width of the leaf by the end point method.

[0024] Figure 4 It is a schematic diagram of comparing the calculation errors of the minimum circumscribed rectangle method and the end point method.

[0025] Figure 5 It is a schematic diagram of comparing the calculated values and the true values of the leaf parameters.

[0026] Figure 6 It is a contour tracking schematic diagram.

[0027] Figure 7 It is a schematic diagram of comparing the calculated values and the true values of the leaf parameters.

[0028] Figure 8 It is a schematic diagram of calculating the angle between the leaf tip and the leaf base.

[0029] Figure 9 It is a schematic diagram of comparing the calculated values and the true values of the leaf parameters. DETAILED DESCRIPTION

[0030] The technical solutions of the present application will be described below in conjunction with the drawings. Obviously, the described embodiments are part of the embodiments of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0031] In the present application, the test material is a hybrid F1 population of Eureka x Beijing Lemon, with a total of 80 plants. As shown in Figure 1 The hybrid population includes various leaf shapes, which generally represent most of the lemon leaf shapes available at the study site. Four fully expanded mature spring shoot leaves of different sizes were randomly selected from different crown layers and different orientations of each tree to achieve a wide variation in morphology. Fresh leaves were carefully separated and placed in plastic bags, then directly sent to the laboratory for image capture, and then manually measured according to the definitions of leaf length, width, leaf tip angle and leaf base angle. In the present application, the manual measurement of leaf length and width is done by vernier caliper, and the manual measurement of leaf tip and leaf base angle is done by protractor. The manually measured leaf length and width, leaf tip and leaf base angle, and the leaf perimeter and area calculated by ImageJ software are used as true values. To reduce errors, the same index of the same leaf is measured three times during manual measurement, and the average of the three measurements is used as the true value.

[0032] The present application will be further explained and described in conjunction with specific embodiments.

[0033] Referring to Figures 1-9 , the present application provides a leaf phenotype parameter measurement method and system based on image processing technology. The leaf phenotype parameter measurement method based on image processing technology comprises the following steps: Step S1: Collecting lemon leaf images and pre-processing the images to obtain binary contour images of the leaves; Step S2: Extracting the length and width of the leaves using the minimum circumscribed rectangle method or the endpoint method; Step S3: Tracking the leaf contour using the Freeman chain code and calculating the leaf perimeter; Step S4: Calculating the leaf area using the pixel number method; Step S5: Measuring the leaf tip angle and leaf base angle.

[0034] In the present embodiment, the leaf phenotype parameter measurement method based on image processing technology specifically comprises the following steps: 1. Measurement of leaf length and width According to the NY / T2930-2016 standard, the length of a lemon leaf is defined as the length of a normally growing representative mature spring shoot leaf from the base to the tip, and the width of a lemon leaf is defined as the distance between the widest parts of a normally growing representative mature spring shoot leaf. Generally, the methods for calculating the length and width of a leaf in an image mainly include the minimum circumscribed rectangle method and the endpoint method.

[0035] (1) Minimum circumscribed rectangle method The minimum bounding rectangle (MBR) of a plant leaf has uniqueness, which reflects some characteristics of the plant leaf to some extent, such as length, width, rectangularity, etc. The specific process of obtaining the minimum bounding rectangle of the leaf is as follows: first, the contour of the leaf is found after image preprocessing, then the convex hull containing all the pixel points of the leaf edge is constructed by using the Graham algorithm, and finally the minimum bounding rectangle of the leaf is obtained based on the convex hull chain according to the rotation search method. The length of the minimum bounding rectangle of the extracted leaf is taken as the length of the leaf, and the width of the minimum bounding rectangle is taken as the width of the leaf. The image of part of the lemon leaf for minimum bounding rectangle extraction is shown in Figure 2 .

[0036] As can be seen from Figure 2 , when the leaf is not highly symmetrical or the length and width of the leaf are relatively close, the length and width of the minimum bounding rectangle of the leaf cannot completely fit the actual length and width of the leaf.

[0037] (2) Endpoint method The definition of the length of the leaf is the length from the leaf base to the leaf tip of a representative mature spring shoot leaf in normal growth, and the definition of the width of the leaf is the distance at the widest part of the leaf body of a representative mature spring shoot leaf in normal growth. According to this definition, the endpoint method is adopted to obtain the length and width of the leaf, and the specific process is as follows: first, all the coordinates of the contour of the leaf are found after image preprocessing, then the positions of the leaf tip and the leaf end are determined by searching the global minimum and maximum column coordinate values of the pixel points of the leaf contour, and the length of the line segment between the leaf tip and the leaf end is the length of the leaf. Finally, a perpendicular line is made to the obtained leaf length line segment, and intersects with the leaf contour at two points, the length of the line segment between the two points is calculated, and the maximum value of the length of the line segment is taken as the width of the leaf, and the two intersection points with the contour at this time are the upper and lower endpoints. The image of part of the lemon leaf for minimum bounding rectangle extraction is shown in Figure 3 .

[0038] As can be seen from Figure 3 , the endpoint method has a good fitting degree for determining the length and width of the leaf, and can also well fit the length and width of the line segment for the leaf with poor symmetry or similar length and width.

[0039] After obtaining the pixel values of the length and width of the leaf, size calibration is performed. The length and width of the leaf are calculated by using the regional pixel counting and template calibration method, and the pixel value and actual length of the standard reference object in the image are combined to obtain the length and width of the leaf, and the calculation formula is as follows: (4-1) In formula 4-1, L leaf represents the length or width of the leaf, L ref represents the actual length of the standard reference object, and Pleaf P represents the total number of pixels in the blade area ref P represents the total number of pixels in the reference length.

[0040] To further verify the accuracy of the minimum enclosing rectangle method and the endpoint method in calculating the length and width of the leaf, the present application randomly selects 10 lemon leaf samples to measure the length and width using the two methods, and compares them with the true leaf length and width data.

[0041] Table 1 Comparison of lemon leaf length measurement methods (unit: cm)

[0042] Table 2 Comparison of lemon leaf width measurement methods (unit: cm)

[0043] As shown in Tables 1 and 2, the absolute error mean of the minimum enclosing rectangle method and the endpoint method in measuring the length of the leaf is 3.90 mm and 1.83 mm, and the absolute error mean of the minimum enclosing rectangle method and the endpoint method in measuring the width of the leaf is 1.80 mm and 1.63 mm, respectively. The measurement error of the endpoint method is smaller than that of the minimum enclosing rectangle method. Figure 4 As shown in Table 3, the error is represented by a curve, and compared with each other, the maximum value of the leaf length error is 1.501, which appears in the minimum enclosing rectangle, and the maximum value of the leaf width error is 0.253, which also appears in the minimum enclosing rectangle. The leaf length and width error measured by the endpoint method is relatively stable, and the mean is smaller, which indicates that the calculation result of this method is not affected by the shape of the leaf, and the method is universal and accurate. Therefore, the endpoint method is selected as the method for extracting the length and width of the leaf.

[0044] In order to quantitatively evaluate the proposed method, three evaluation indexes, including mean absolute error (MAE), root mean square error (RMSE) and average determination coefficient (R 2 ), are used, and their related formulas are as follows: (4-2) (4-3) (4-4) In the formula: is the measurement value of the sample according to the algorithm of the present application; is the average value of the value; is the true value of the sample ; is the average value of the value; is the number of samples.

[0045] The repeatability and accuracy of the end-point method were verified experimentally. 320 lemon leaf blades were used as experimental objects, and the artificial measurement results and the leaf blade data obtained by the algorithm were analyzed and compared, and the results are shown in Figure 5 .

[0046] It can be seen that there is a clear linear correlation between the calculated value and the true value, and the calculated value and the true value are close to the 1:1 line, indicating that the calculated value is close to the true value. The MAE between the calculated value and the true value of leaf length and leaf width is 0.233 cm and 0.131 cm, respectively, the RMSE value is 0.252 cm and 0.150 cm, respectively, and the R 2 value is 0.98 and 0.97, respectively. The MAE value, RMSE value and R 2 value show that the results of the end-point method for calculating the length and width of the lemon leaf blade are close to the true value, and the accuracy of the calculation results is high.

[0047] 2. Measurement of leaf blade perimeter and area The perimeter of the leaf blade is the length of the shape contour of the leaf blade. Before calculating the perimeter of the target region of the image, it is usually necessary to track the contour edge of the target region in the binary image using chain code. The present application uses Freeman chain code

[126] to track the contour clockwise. For the i-th contour point, Ci is the chain code of the point, and according to the direction relationship between the point and the next contour point, the value of Ci is in the range of [0, 7], as shown in Figure X. Scan the binary image from top to bottom and from left to right, take the first target point scanned as the starting point, and track the contour clockwise to return to the starting point, and the contour chain code sequence is 0706644322. After obtaining the contour chain code sequence, the region perimeter can be calculated.

[0048] The pixel is regarded as a point. When it is in the starting point, if the edge curve direction is inclined (the chain code value is odd), the perimeter is increased by 1; if the edge curve direction is horizontal or vertical (the chain code value is even), the perimeter is increased by . After obtaining the number of pixels of the perimeter, the actual perimeter of the leaf blade can be calculated according to the actual size of each pixel. The actual perimeter calculation formula is: (4-5) In formula (4-3), C leaf represents the perimeter of the leaf blade, L ref represents the actual length of the standard reference, P leaf represents the total number of pixels of the leaf blade perimeter in the leaf blade region, P ref represents the total number of pixels of the reference length.

[0049] Generally, the area of the target region in the image is calculated by three methods, i.e. the pixel number method, the boundary chain code method and the Green formula method. The pixel number method is relatively simple and has high precision. After image segmentation, the total number of pixels in the internal region of the leaf boundary in the binary image is counted as the area, as shown in formula (4-4).

[0050] As = Σ (x,y)∈sf(x, y)(4-6) wherein S is the target connected region, and the color value f(x, y) in the binary image is equal to 1 is accumulated area.

[0051] Then, the actual area of the leaf is calculated by combining the area of the standard reference in the image, as shown in formula (4-5).

[0052] (4-7) In formula (4-5), S leaf represents the leaf area, S ref represents the actual area of the standard reference, P leaf represents the total number of pixels of the leaf area in the leaf region, P ref represents the total number of pixels of the reference area.

[0053] In order to evaluate the reliability of the method proposed in the present application, 320 lemon leaves are obtained, and the perimeter and area of the leaves are counted by using the ImageJ software, which are taken as the true values of the measured leaves, and compared with the leaf data obtained by the algorithm. The results are shown in Table 1. Figure 7 As can be seen from Table 1, the calculated values of the leaf perimeter and the leaf area are close to the 1:1 line distribution, indicating that the difference between the measured values and the calculated values is small, and the calculated values can reflect the changes of the true lemon leaf perimeter and area. The MAE of the leaf perimeter and the leaf area is 0.329 cm and 0.111 cm 2 , respectively, the RMSE is 0.406 cm and 0.155 cm 2 , respectively, and the R 2 is 0.99, the small RMSE value and the R 2 value close to 1 indicate that the calculation result is reliable.

[0054] 3. Measurement of the leaf tip angle and the leaf base angle In Figure 1 , it can be found that the leaf shape of the lemon leaves is quite different, and the classification of the leaf shape is closely related to the angles of the leaf tip and the leaf base. Therefore, the present application proposes a method for detecting the angles of the leaf tip and the leaf base of the lemon leaves based on image processing. The specific process of the method is as follows: (1) Since the leaf tips and bases of different leaves do not face the same direction when the leaf photos are taken, some leaves point to the right while others point to the left, making it impossible to directly calculate the angle between the leaf tip and base. Therefore, it is not difficult to observe the image and find that the area of ​​the leaf tip region is smaller than that of the leaf base region. In view of this, the following method can be used for calculation: First, find the leaf outline, and take the left and right endpoints of the leaf outline as the coordinate zero points respectively, find the area of ​​one-tenth of the left and right ends of the image, and then calculate the leaf pixel area in these two areas and compare them. This gives the relative position and area ratio of the leaf tip and base, which helps in the subsequent angle calculation. If the area of ​​the left end is greater than that of the right end, it is determined that the tobacco leaf tip is facing right and the leaf base is facing left, and the image needs to be vertically flipped. If the area of ​​the left end is less than that of the right end, it is determined that the tobacco leaf tip is facing left and the leaf base is facing right. By using this method, all leaf tips in the leaf image are facing left and all leaf bases are facing right.

[0055] (2) After the leaf image is morphologically processed, the contour is extracted at one-third of the leaf image according to the screen coordinate system.

[0056] (3) Find all the upper and lower contour coordinates in the contour image respectively, and use 2D linear fitting to obtain two straight line fitting equations. The calculation formulas are shown in equations (4-8) and (4-9).

[0057] y1 = k1x1 + b1 (4-8) y2 = k2x2 + b2 (4-9) The intersection of two fitted lines can be obtained by fitting the equation, and the calculation formula is shown in (4-10).

[0058] (4-10) Meanwhile, take two points on the upper and lower contour lines respectively, as shown in equations (4-11) and (4-12).

[0059] (4-11) (4-12) Then, connect the three coordinate points to obtain the three sides of the triangle, and the calculation formula is as follows: (4-13) (4-14) (4-15) In the triangle formed, the angle between the intersection of the two fitted equations is the angle ∠θ between the tip and base of the lemon leaf, as shown below. Figure 8 As shown, the angle between the tip and base of the lemon leaf is finally obtained by using the law of cosines, and the calculation formula is shown in (4-16).

[0060] (4-16) In order to measure the accuracy of the measurement algorithm proposed above, the lemon leaf sample is selected for error analysis. The total amount of leaf samples is 320, and the tip and base angle of each leaf is measured, compared and analyzed. In order to facilitate comparison, the unit of the measurement parameter is unified as degree, accurate to one decimal place. For the input lemon leaf image, the tip angle and base angle of the leaf are calculated according to the algorithm proposed in the application, and the relationship between the true value and the calculated value is as shown in Figure 9 The actual measurement value of the tip and base angle of the lemon leaf and the program calculation value have a significant positive correlation, and the linear correlation coefficients R 2 are 0.92 and 0.90, respectively; the MAE between the calculated value and the true value of the tip and base angle is 1.7°, and the RMSE is 2.0° and 2.1°, respectively. Compared with the measured data, the error of the tip angle and base angle of the leaf identified according to the provided algorithm is smaller.

[0061] Please refer to Figures 1-9 , the lemon leaf phenotype parameter measurement system based on image processing technology provided by the embodiment of the application runs the method described above, and the system comprises: An image acquisition module is configured to acquire a lemon leaf image. A preprocessing module is configured to preprocess the acquired image to obtain a binary profile image of the leaf. A parameter extraction module is configured to extract the length, width, circumference, area, tip angle and base angle of the leaf. A data storage and display module is configured to store and display the extracted leaf phenotype parameters.

[0062] In summary, the present application firstly uses the minimum circumscribed rectangle method and the endpoint method to extract the length and width of the lemon leaf, compares the extraction results of the two methods, and performs precision analysis on the length and width of the lemon leaf extracted by using the endpoint method. The correlation determination coefficients between the calculated values and the actual measurement values of the length and width of the leaf calculated by the algorithm are 0.98 and 0.97, respectively, the MAE is 0.233 cm and 0.131 cm, respectively, and the RMSE value is 0.252 cm and 0.150 cm, respectively. Next, the detection algorithm of the two-dimensional geometric parameters of the lemon leaf, such as the circumference, area, tip angle and base angle, is described, and the correlation determination coefficients obtained by data statistical analysis are 0.99, 0.99, 0.92 and 0.90, respectively, the MAE is 0.329 cm, 0.111 cm 2 , 1.7° and 1.7°, respectively, and the RMSE is 0.406 cm, 0.155 cm 2 , 2.0° and 2.1°, respectively.

[0063] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, and are not intended to limit the present application; although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions recorded in the above embodiments can still be modified, or some or all of the technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for measuring blade phenotypic parameters based on image processing technology, characterized in that, The method includes the following steps: Step S1: Acquire images of lemon leaves and preprocess the images to obtain a binarized contour image of the leaves; Step S2: Extract the length and width of the blade using the minimum bounding rectangle method or the endpoint method; Step S3: Use Freeman chain code to trace the blade profile and calculate the blade circumference; Step S4: Calculate the blade area using the pixel count method; Step S5: Measure the leaf tip angle and leaf base angle of the leaf.

2. The method for measuring blade phenotypic parameters based on image processing technology according to claim 1, characterized in that, In step S1, the preprocessing includes image grayscale conversion, threshold segmentation, and morphological processing to remove image noise and highlight the leaf contours.

3. The method for measuring blade phenotypic parameters based on image processing technology according to claim 1, characterized in that, In step S2, the specific steps for extracting the length and width of the blade using the endpoint method include: The positions of the leaf tip and leaf base are determined by searching the global minimum and maximum column coordinate values ​​of the leaf outline pixels, and the length of the line segment from the leaf tip to the leaf base is calculated as the leaf length. Draw a perpendicular line to the line segment, intersecting the blade profile at two points, and take the maximum value of the line segment length between the two points as the blade width; When calculating the blade length and width, the dimensions are determined using the following formula: ； Among them, L leaf L represents the length or width of the blade. ref P represents the actual length of the standard reference object. leaf P represents the total number of pixels representing the length or width of the leaf in the leaf region. ref The total number of pixels representing the length of the reference object.

4. The method for measuring blade phenotypic parameters based on image processing technology according to claim 1, characterized in that, In step S3, Freeman chain code is used to trace the blade profile and calculate the blade circumference, specifically including: The blade profile is tracked clockwise, the distance between profile points is determined based on the chain code value, the total number of pixels for the perimeter is accumulated, and the actual perimeter of the blade is calculated by combining the total number of pixels of the standard reference object and the actual length.

5. The method for measuring blade phenotypic parameters based on image processing technology according to claim 4, characterized in that, In step S3, when calculating the perimeter, for contour points in the inclined direction with odd chain code values, the distance between two points is 1 pixel unit. For horizontal or vertical contour points with even chain code values, the distance between the two points is... Each pixel unit.

6. The method for measuring blade phenotypic parameters based on image processing technology according to claim 1, characterized in that, In step S4, the blade area is calculated using the pixel count method, specifically including: The total number of pixels with a color value of 1 in the leaf region of the binarized contour image is counted, and the actual area of ​​the leaf is calculated by combining the total number of pixels of the standard reference object and the actual area. The formula for calculating the actual area of ​​a blade is: ； Among them, S leaf S represents the blade area. ref P represents the actual area of ​​the standard reference object. leaf P represents the total number of pixels representing the leaf area in the leaf region. ref The total number of pixels representing the area of ​​the reference object.

7. The method for measuring blade phenotypic parameters based on image processing technology according to claim 1, characterized in that, In step S5, the leaf tip angle and leaf base angle are measured, specifically including: Determine the relative positions of the leaf tip and leaf base and align them in the same direction; Extract the upper and lower edge contour coordinates at one-third of the blade and perform 2D linear fitting to obtain two straight line equations; Calculate the intersection of the two straight lines, take one point on each of the upper and lower contour lines, and calculate the tip angle and base angle using the law of cosines.

8. The method for measuring blade phenotypic parameters based on image processing technology according to claim 7, characterized in that, In step S5, the method for determining the relative position of the leaf tip and leaf base is as follows: taking the left and right endpoints of the leaf outline as the coordinate zero points, find the area of ​​one-tenth of the left and right ends of the image, calculate the leaf pixel area in the two areas and compare them; if the area of ​​the left end is greater than the area of ​​the right end, it is determined that the leaf tip is facing right, and the image is vertically flipped so that the leaf tip is facing left; if the area of ​​the left end is less than the area of ​​the right end, it is determined that the leaf tip is facing left.

9. The method for measuring blade phenotypic parameters based on image processing technology according to claim 1, characterized in that, The method also includes evaluating the accuracy of the measurement results, with evaluation indicators including mean absolute error, root mean square error, and mean coefficient of determination.

10. A lemon leaf phenotypic parameter measurement system based on image processing technology, operating the method as described in any one of claims 1-9, characterized in that, The system includes: The image acquisition module is used to acquire images of lemon leaves; The preprocessing module is used to preprocess the acquired images to obtain the binarized contour image of the blades; The parameter extraction module is used to extract the length, width, perimeter, area, leaf tip angle, and leaf base angle of the leaf. The data storage and display module is used to store and display the extracted leaf phenotypic parameters.

Images