Tobacco variety distinguishing method, electronic equipment and storage medium

Abstract

The invention relates to the technical field of data processing, in particular to a tobacco variety distinguishing method, electronic equipment and a storage medium, and the method comprises the steps: obtaining a tobacco image in a target area, performing tobacco identification and tobacco segmentation on the tobacco image, screening out a plurality of target tobacco objects and vector information of a grid image corresponding to each target tobacco object, and calculating to obtain a morphological feature data set corresponding to each target tobacco object according to the vector information of the grid image, according to a plurality of pre-obtained morphological characteristic data sets corresponding to a given tobacco variety, calculating the difference between the target tobacco object and the given tobacco variety, and determining whether the tobacco variety to which the target tobacco object belongs and the given tobacco variety are different according to the difference; according to the method, the digitalization of the morphological characteristics of the tobacco leaves is realized based on image acquisition and processing, the difference characteristics and the difference size can be accurately and reliably determined through calculation, and the efficiency and the accuracy of tobacco leaf variety distinguishing are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a tobacco variety distinguishing method, electronic equipment and storage medium. Background Art

[0002] In order to ensure regional tobacco variety control, ensure tobacco production and improved cultivation, and consolidate and improve regional tobacco quality, it is necessary to establish a way to distinguish between different tobacco varieties to ensure tobacco quality and safety. At present, tobacco planting styles vary from place to place, with distinct regional characteristics, but there are still some inferior varieties mixed together. Therefore, it is necessary to identify the characteristics of different tobacco varieties and distinguish them during the field period, and to discover non-specified varieties.

[0003] At present, the differentiation of tobacco leaf varieties still mainly relies on the empirical judgment of experienced tobacco technicians in the field. This places certain requirements on the tobacco technicians' skills in distinguishing different varieties of tobacco leaves. For inexperienced tobacco technicians, the distinction may be difficult and require actual field inspections, which will also affect the efficiency and accuracy of tobacco leaf judgment. Summary of the invention

[0004] In response to the above technical problems, the present invention provides a method for distinguishing tobacco leaf varieties, an electronic device and a storage medium, which realize the digitization of tobacco leaf morphological characteristics based on image acquisition and processing, and help to accurately and reliably determine the difference characteristics and difference sizes through calculation, thereby achieving efficiency and accuracy in distinguishing tobacco leaf varieties.

[0005] According to a first aspect of the present invention, a method for distinguishing tobacco varieties is provided, comprising the following steps: Based on the preset shooting height, the tobacco leaf images in the target area taken by the remote sensing UAV are obtained.

[0006] An image segmentation model is used to perform tobacco leaf recognition and tobacco leaf segmentation on the tobacco leaf image to obtain a tobacco leaf segmentation result, and based on the tobacco leaf segmentation result, a number of target tobacco leaf objects and vector information of a raster image corresponding to each target tobacco leaf object are screened out; the target tobacco leaf object refers to any tobacco leaf object whose degree of integrity meets the first preset requirement.

[0007] According to the vector information of the raster image corresponding to each target tobacco leaf object, a morphological feature data set corresponding to each target tobacco leaf object is calculated; the morphological feature data set corresponding to the target tobacco leaf object includes aspect ratio, asymmetry index, shape index, boundary index, density, compactness, roundness, rectangle fitting degree and brightness value.

[0008] According to the morphological feature data set corresponding to each target tobacco leaf object and several morphological feature data sets corresponding to a given tobacco leaf variety obtained in advance, the difference between each morphological feature corresponding to the tobacco leaf variety to which the target tobacco leaf object belongs and the given tobacco leaf variety is obtained.

[0009] When the difference in the morphological features is greater than a preset difference threshold, it is determined that the tobacco variety to which the target tobacco object belongs is different from the given tobacco variety.

[0010] According to a second aspect of the present invention, a non-transitory computer-readable storage medium is provided, in which at least one instruction or at least one program is stored, and the at least one instruction or the at least one program is loaded and executed by a processor to implement the above-mentioned method for distinguishing tobacco varieties.

[0011] According to a third aspect of the present invention, there is provided an electronic device comprising a processor and the above-mentioned non-transitory computer-readable storage medium.

[0012] The present invention has at least the following beneficial effects: The present invention provides a method for distinguishing tobacco leaf varieties. First, a tobacco leaf image in a target area is acquired, tobacco leaf recognition and tobacco leaf segmentation are performed on the tobacco leaf image, and vector information of a plurality of target tobacco leaf objects and a raster image corresponding to each target tobacco leaf object is screened out. Digitalization of tobacco leaf morphological features is achieved through image acquisition and processing. A morphological feature data set corresponding to each target tobacco leaf object is calculated according to the vector information of the raster image. Combined with a plurality of morphological feature data sets corresponding to a given tobacco leaf variety acquired in advance, a difference between the tobacco leaf variety to which the target tobacco leaf object belongs and the given tobacco leaf variety based on each morphological feature is calculated. Whether the tobacco leaf variety to which the target tobacco leaf object belongs and the given tobacco leaf variety are different tobacco leaf varieties is determined according to the difference. The difference features and difference sizes can be accurately and reliably determined through the calculation results, thereby achieving efficiency and accuracy in distinguishing tobacco leaf varieties. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0014] Figure 1 A flow chart of a method for distinguishing tobacco leaf varieties provided by an embodiment of the present invention; Figure 2 A schematic diagram of a segmented complete tobacco leaf object provided by an embodiment of the present invention; Figure 3A box diagram showing the asymmetry differences between Yunyan 87 and K326; Figure 4 Box diagram showing the difference in brightness values ​​between Yunyan 87 and K326. DETAILED DESCRIPTION

[0015] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0016] The embodiment of the present invention provides a method for distinguishing tobacco varieties, such as Figure 1 As shown, the method comprises the following steps: S100, based on a preset shooting height, obtaining an image of tobacco leaves in a target area photographed by a remote sensing drone.

[0017] Specifically, the target area refers to a pre-selected tobacco field of tobacco varieties to be distinguished. In a specific implementation, there may be differences in the growth of tobacco leaves in the field, and a plot with uniform growth is selected to represent the growth status of tobacco leaves in the tobacco field.

[0018] In a specific embodiment, the preset shooting height is selected through the following steps: S101, according to a plurality of preset tobacco leaf varieties, obtain remote sensing images of any tobacco field corresponding to each preset tobacco leaf variety at a plurality of given shooting heights. For example, the plurality of given shooting heights may be 8m, 10m, 12m, 15m, 20m, 30m, etc.

[0019] S102, sending all remote sensing images to the user end, and receiving the shooting height screening results based on the remote sensing images by the user end, so as to obtain the shooting height range corresponding to each preset tobacco field variety; it can be understood that: the user selects remote sensing images with tobacco leaf size that meets the requirements and high clarity according to the tobacco leaf size and clarity in the remote sensing images, and determines the shooting height range corresponding to each preset tobacco field variety according to the shooting heights corresponding to the selected remote sensing images.

[0020] S103, according to the shooting height range corresponding to each preset tobacco field variety, determine the target height range; the target height range is a range obtained according to the union of several shooting height ranges. For example, if the shooting height ranges corresponding to two preset tobacco field varieties are 10-15m and 12-16m respectively, then the target height range is 12-15m.

[0021] S104, for any shooting height in the target height range, based on the remote sensing image feature set corresponding to each preset tobacco field variety at the shooting height, obtain the similarity between every two remote sensing images, and calculate the average similarity value; the remote sensing image feature set includes the average tobacco leaf size and color saturation; it can be understood as: the tobacco leaf size and color saturation are respectively vectorized and combined into vector pairs, and the similarity between the vector pairs is calculated.

[0022] S105: Select the shooting height corresponding to the minimum average similarity value as the preset shooting height.

[0023] As mentioned above, by taking multiple preset tobacco field varieties as samples, a shooting height range with a wide application range can be selected, and based on the similarity between the remote sensing images corresponding to each preset tobacco field variety at the same shooting height, the shooting height with the minimum similarity average value is selected as the preset shooting height, which is conducive to distinguishing different tobacco leaf varieties at the same shooting height, ensuring the clarity of the image shooting, and improving the accuracy of distinguishing tobacco leaf varieties.

[0024] S200, using an image segmentation model to perform tobacco leaf recognition and tobacco leaf segmentation on the tobacco leaf image, obtaining a tobacco leaf segmentation result, and screening out a number of target tobacco leaf objects and vector information of a raster image corresponding to each target tobacco leaf object based on the tobacco leaf segmentation result.

[0025] Specifically, the target tobacco leaf object refers to any tobacco leaf object whose degree of integrity meets the first preset requirement. For example, the tobacco leaf object with intact leaves is selected, and the schematic diagram of the segmented intact tobacco leaf object is as follows: Figure 2 shown.

[0026] Furthermore, the image segmentation model may be a SAM segmentation model.

[0027] In a specific embodiment, several target tobacco leaf objects are screened out through the following steps: S201, obtaining the pixel size corresponding to each initial tobacco leaf object in the tobacco leaf segmentation result; it can be understood that: the initial tobacco leaf object refers to any tobacco leaf object obtained in the tobacco leaf segmentation result.

[0028] S202, selecting a plurality of first tobacco leaf objects from a plurality of initial tobacco leaf objects according to a preset pixel interval; the first tobacco leaf object refers to any initial tobacco leaf object whose corresponding pixel size is within the pixel interval.

[0029] S203, randomly selecting a first number of first tobacco leaf objects from a plurality of first tobacco leaf objects and sending them to a user terminal, so that a target user selects a plurality of second tobacco leaf objects whose completeness meets a first preset requirement from the first number of first tobacco leaf objects.

[0030] S204, receiving the screening feedback result from the user end, when the number of screened second tobacco leaf objects is greater than the preset second number, taking each second tobacco leaf object as a target tobacco leaf object; otherwise, expanding the first number according to the preset step length, and returning to the execution step to randomly screen out a first number of first tobacco leaf objects from a number of first tobacco leaf objects and sending them to the user end, until a number of target tobacco leaf objects are screened out.

[0031] As mentioned above, since there are many leaves in the captured images, the computational complexity can be reduced by screening the tobacco leaf objects. In the screening process, there is no need to screen all of them. First, a preliminary screening is performed based on the pixel size, and then a number of first tobacco leaf objects are randomly screened out from the preliminary screening results for the user to perform a secondary screening, thereby reducing the number of tobacco leaf objects during the secondary screening. The number of tobacco leaves will only be expanded when the number of complete tobacco leaf objects screened out by the user is small, which can improve the efficiency of manual screening.

[0032] S300, based on the vector information of the raster image corresponding to each target tobacco leaf object, calculate the morphological feature data set corresponding to each target tobacco leaf object; the morphological feature data set corresponding to the target tobacco leaf object includes aspect ratio, asymmetry index, shape index, boundary index, density, compactness, roundness, rectangle fit and brightness value.

[0033] Preferably, step S300 further includes the following steps: S301, for any target tobacco leaf object, the vector information of the raster image corresponding to the target tobacco leaf object is loaded into a given image analysis software and the raster image is subjected to chessboard segmentation processing to obtain a chessboard segmentation result; wherein the size of the chessboard segmentation is larger than the maximum value in the rows and columns of the raster image.

[0034] Furthermore, the image analysis software may be eCognition software.

[0035] S302, according to the chessboard segmentation result, respectively determine a number of parameter values ​​corresponding to each morphological feature to calculate each morphological feature data corresponding to the target tobacco leaf object; it can be understood that: a number of parameters corresponding to each morphological feature can be obtained from the chessboard segmentation result, for example, when the morphological feature is the aspect ratio, the number of parameters corresponding to the morphological feature are the length and width of the tobacco leaf object.

[0036] Specifically, the aspect ratio of the target tobacco leaf object refers to the ratio of the length to the width of the minimum circumscribed rectangle of the target tobacco leaf object.

[0037] Furthermore, the asymmetry index A of the target tobacco leaf object meets the following conditions: A=X / Y, where X is the short axis of the smallest ellipse surrounding the target tobacco leaf object, and Y is the long axis of the smallest ellipse surrounding the target tobacco leaf object.

[0038] Furthermore, the shape index G of the target tobacco leaf object meets the following conditions: G=L 0 / (4S 1 1 / 2 ), where L 0 is the boundary length of the target tobacco leaf object, S 1 is the area of ​​the target tobacco leaf object.

[0039] Specifically, the boundary index B of the target tobacco leaf object meets the following conditions: B=L 0 / 2(L 1 +W 1 ), where L 1 is the length of the minimum bounding rectangle of the target tobacco leaf object, W 1 The width of the minimum bounding rectangle of the target tobacco leaf object.

[0040] Specifically, the density D of the target tobacco leaf object meets the following conditions: D=P 1 / 2 / (1+Y), where P is the number of pixels contained in the target tobacco leaf object.

[0041] Furthermore, the compactness C of the target tobacco leaf object meets the following conditions: C=(L 1 ×W 1 ) / P.

[0042] Specifically, the roundness E of the target tobacco leaf object meets the following conditions: E=Z max -Z min , where Z max is the radius of the largest closed ellipse within the target tobacco leaf object, Z min The radius of the minimum ellipse that encloses the target tobacco leaf object.

[0043] Furthermore, the rectangular fitting degree F of the target tobacco leaf object meets the following conditions: F=(S 2 -S 3 ) / S 1 , where S 2 For S 1 The area of ​​the target rectangle of equal area, S 3 is the area of ​​the target tobacco leaf object outside the target rectangle: it can be understood as: S 2 =S 1 .

[0044] Specifically, the brightness value of the target tobacco leaf object refers to the average value of the intensities of several spectral bands contained in the target tobacco leaf object.

[0045] As mentioned above, by calculating the values ​​of several morphological characteristics, the digitization of tobacco leaf morphology is realized. Calculating different morphological characteristic parameters is conducive to the subsequent accurate screening of differential characteristics that can help to distinguish tobacco leaf varieties. In addition, a variety of morphological characteristics are introduced to cover the possible differential characteristics of tobacco leaf varieties, making the basis for judging tobacco leaf variety differentiation more comprehensive, and further improving the accuracy and reliability of tobacco leaf variety differentiation.

[0046] S400, based on the morphological feature data set corresponding to each target tobacco leaf object and several morphological feature data sets corresponding to the given tobacco leaf variety obtained in advance, obtain the difference size of each morphological feature between the tobacco leaf variety to which the target tobacco leaf object belongs and the given tobacco leaf variety; that is, for each morphological feature, calculate the difference size between the tobacco leaf variety of the target tobacco leaf object and the given tobacco leaf variety.

[0047] In a specific embodiment, step S400 includes the following steps: S401, for any morphological feature, according to the morphological feature data set corresponding to each target tobacco leaf object, calculate the first feature average value corresponding to the morphological feature. For example, for the asymmetry index, according to the value of the asymmetry index corresponding to each target tobacco leaf object, calculate an asymmetry index average value.

[0048] S402, calculating a second feature average value corresponding to the morphological feature according to a plurality of morphological feature data sets corresponding to a given tobacco leaf variety.

[0049] Specifically, the second feature average value is calculated in the same manner as the first feature average value.

[0050] S403, according to a number of morphological feature data corresponding to the morphological feature, the first feature average value and the second feature average value, an effect size calculation method is used to calculate the difference between the tobacco variety to which the target tobacco object belongs and the given tobacco variety based on the morphological feature.

[0051] Specifically, the plurality of morphological feature data corresponding to the morphological feature include a plurality of data corresponding to the same morphological feature of a plurality of target tobacco leaf objects and a plurality of data corresponding to the same morphological feature of a given tobacco leaf variety. For example, when the morphological feature is an asymmetry index, the plurality of morphological feature data corresponding to the morphological feature include an asymmetry index corresponding to each target tobacco leaf object and an asymmetry index corresponding to each given tobacco leaf object.

[0052] Furthermore, the difference K between the tobacco variety to which the target tobacco object belongs and the given tobacco variety based on morphological characteristics meets the following conditions: K = (λ 1 -λ 2 ) / (∑ m i=1 (J i -J 0 ) 2 / (m-1)) 1 / 2 , where λ 1 is the average value of the first feature corresponding to any morphological feature, λ 2 is the average value of the second feature corresponding to any morphological feature, m is the number of morphological feature data corresponding to any morphological feature, J i is the i-th morphological feature data corresponding to any morphological feature, J 0 is the mean of m morphological feature data.

[0053] For example, between different tobacco varieties, Yunyan 87 and K326, asymmetry is a significant feature, and the calculated difference is 1.17, while the brightness value of the leaf surface has no significant difference, and the difference is -0.05. At the same time, a box plot is used for data visualization, such as Figure 3 As shown in the figure, it can be found that in terms of asymmetry, Yunyan 87 is concentrated between 0.73-0.8, while K326 is concentrated between 0.65-0.74; in terms of brightness value characteristics, the two are concentrated in similar areas, and the feature distinction is not obvious, such as Figure 4 shown.

[0054] As mentioned above, by calculating the morphological characteristic data corresponding to the given tobacco variety and the target tobacco variety respectively, the morphological characteristics with significant differences can be screened out, and based on the calculation of the difference size, the degree of difference between the target tobacco variety and the given tobacco variety corresponding to each morphological characteristic can be reflected. According to the degree of difference, it is helpful to accurately and quickly determine whether the target tobacco variety being planted is the same tobacco variety as the given tobacco variety required to be planted, thereby improving the efficiency and accuracy of tobacco variety differentiation.

[0055] In another embodiment, the method obtains several morphological feature data sets corresponding to a given tobacco leaf variety through the following steps: S10, based on a preset shooting height, obtaining a shooting image corresponding to a given tobacco variety in a given area shot by a remote sensing drone.

[0056] S20, using an image segmentation model to perform tobacco leaf recognition and tobacco leaf segmentation on the captured image, and screening out a plurality of third tobacco leaf objects and obtaining vector information of a raster image corresponding to each third tobacco leaf object.

[0057] Specifically, the third tobacco leaf image refers to any tobacco leaf object whose degree of integrity meets the second preset requirement. For example, tobacco leaf objects with intact tobacco leaves are selected.

[0058] Furthermore, the screening method of the third tobacco leaf image is consistent with the screening method of the target tobacco leaf object.

[0059] S30, based on the vector information of the raster image corresponding to each third tobacco leaf object, calculate the morphological feature dataset corresponding to each third tobacco leaf object, and use the several morphological feature datasets corresponding to the several third tobacco leaf objects as the several morphological feature datasets corresponding to the given tobacco leaf variety; it can be understood that: the calculation method of the morphological feature dataset corresponding to the third tobacco leaf object is consistent with the morphological feature dataset corresponding to the target tobacco leaf object.

[0060] As mentioned above, for a given tobacco variety required to be planted in a certain area, a number of morphological characteristic data corresponding to the given tobacco variety are obtained in the same way as the target tobacco variety. By comparing the data obtained in the same way, it is helpful to improve the reliability of the calculated difference between the two tobacco varieties.

[0061] S500: When the difference in the morphological features is greater than a preset difference threshold, it is determined that the tobacco variety to which the target tobacco object belongs is different from the given tobacco variety. The technicians in this field can set the difference threshold according to actual needs.

[0062] In summary, the present invention provides a method for distinguishing tobacco leaf varieties, which first obtains a tobacco leaf image in a target area, performs tobacco leaf recognition and tobacco leaf segmentation on the tobacco leaf image, and screens out a number of target tobacco leaf objects and vector information of a raster image corresponding to each target tobacco leaf object, and realizes the digitization of tobacco leaf morphological features through image acquisition and processing, and calculates a morphological feature data set corresponding to each target tobacco leaf object according to the vector information of the raster image, and calculates the difference between the tobacco leaf variety to which the target tobacco leaf object belongs and the given tobacco leaf variety based on each morphological feature in combination with a number of morphological feature data sets corresponding to a given tobacco leaf variety obtained in advance, and determines whether the tobacco leaf variety to which the target tobacco leaf object belongs and the given tobacco leaf variety are different tobacco leaf varieties according to the difference size, and the difference features and difference sizes can be accurately and reliably determined through the calculation results, thereby achieving efficiency and accuracy in distinguishing tobacco leaf varieties.

[0063] An embodiment of the present invention also provides a non-transitory computer-readable storage medium, which can be set in an electronic device to store at least one instruction or at least one program related to implementing a method in a method embodiment. The at least one instruction or the at least one program is loaded and executed by the processor to implement the tobacco variety differentiation method provided in the above embodiment.

[0064] An embodiment of the present invention further provides an electronic device, comprising a processor and the aforementioned non-transitory computer-readable storage medium.

[0065] Although some specific embodiments of the present invention have been described in detail by way of example, it will be appreciated by those skilled in the art that the above examples are for illustration only and are not intended to limit the scope of the present invention. It will also be appreciated by those skilled in the art that various modifications may be made to the embodiments without departing from the scope and spirit of the present invention. The scope of the present invention is defined by the appended claims.

Claims

1. A method for distinguishing tobacco leaf varieties, characterized in that: The method comprises the following steps: Based on the preset shooting height, the tobacco leaf images in the target area taken by the remote sensing UAV are obtained; Using an image segmentation model to perform tobacco leaf recognition and tobacco leaf segmentation on the tobacco leaf image, obtaining a tobacco leaf segmentation result, and screening out a number of target tobacco leaf objects and vector information of a raster image corresponding to each target tobacco leaf object based on the tobacco leaf segmentation result; The target tobacco leaf object refers to any tobacco leaf object whose tobacco leaf integrity meets the first preset requirement; According to the vector information of the raster image corresponding to each target tobacco leaf object, a morphological feature data set corresponding to each target tobacco leaf object is calculated; the morphological feature data set corresponding to the target tobacco leaf object includes aspect ratio, asymmetry index, shape index, boundary index, density, compactness, roundness, rectangle fitting degree and brightness value; According to the morphological feature data set corresponding to each target tobacco leaf object and several morphological feature data sets corresponding to a given tobacco leaf variety obtained in advance, the difference between each morphological feature corresponding to the tobacco leaf variety to which the target tobacco leaf object belongs and the given tobacco leaf variety is obtained; When the difference in the morphological features is greater than a preset difference threshold, it is determined that the tobacco variety to which the target tobacco object belongs is different from the given tobacco variety.

2. The method for distinguishing tobacco leaf varieties according to claim 1, characterized in that: Select the preset shooting height by following the steps below: According to a number of preset tobacco leaf varieties, remote sensing images of any tobacco field corresponding to each preset tobacco leaf variety at a number of given shooting heights are obtained; Send all remote sensing images to the user end, and receive the shooting height screening results based on the remote sensing images from the user end to obtain the shooting height range corresponding to each preset tobacco field variety; According to the shooting height range corresponding to each preset tobacco field variety, a target height range is determined; the target height range is a range obtained according to the union of several shooting height ranges; For any shooting height within the target height range, based on the remote sensing image feature set corresponding to each preset tobacco field variety at the shooting height, the similarity between every two remote sensing images is obtained, and the average similarity value is calculated; the remote sensing image feature set includes the average tobacco leaf size and color saturation; The shooting height corresponding to the minimum similarity average value is selected as the preset shooting height.

3. The method for distinguishing tobacco leaf varieties according to claim 1, characterized in that: Several target tobacco leaf objects are screened out through the following steps: Obtain the pixel size corresponding to each initial tobacco leaf object in the tobacco leaf segmentation result; According to a preset pixel interval, a plurality of first tobacco leaf objects are selected from a plurality of initial tobacco leaf objects; the first tobacco leaf object refers to any initial tobacco leaf object whose corresponding pixel size is within the pixel interval; Randomly selecting a first number of first tobacco leaf objects from a plurality of first tobacco leaf objects and sending the objects to a user terminal, so that a target user selects a plurality of second tobacco leaf objects whose completeness meets a first preset requirement from the first number of first tobacco leaf objects; Receive the screening feedback result from the user end. When the number of screened second tobacco leaf objects is greater than the preset second number, each second tobacco leaf object is used as a target tobacco leaf object; otherwise, the first number is expanded according to the preset step size, and the execution step is returned to randomly screen out a first number of first tobacco leaf objects from a number of first tobacco leaf objects and send them to the user end until a number of target tobacco leaf objects are screened.

4. The method for distinguishing tobacco leaf varieties according to claim 1, characterized in that: The step of calculating the morphological feature data set corresponding to each target tobacco leaf object according to the vector information of the raster image corresponding to each target tobacco leaf object comprises the following steps: For any target tobacco leaf object, the vector information of the raster image corresponding to the target tobacco leaf object is loaded into a given image analysis software and the raster image is subjected to chessboard segmentation processing to obtain a chessboard segmentation result; wherein the size of the chessboard segmentation is larger than the maximum value in the rows and columns of the raster image; According to the chessboard segmentation result, several parameter values ​​corresponding to each morphological feature are determined respectively, so as to calculate each morphological feature data corresponding to the target tobacco leaf object.

5. The method for distinguishing tobacco leaf varieties according to claim 1, characterized in that: The difference between the target tobacco leaf object's tobacco leaf variety and the given tobacco leaf variety's corresponding morphological features is obtained through the following steps: For any morphological feature, according to the morphological feature data set corresponding to each target tobacco leaf object, the first feature average value corresponding to the morphological feature is calculated; According to a plurality of morphological characteristic data sets corresponding to a given tobacco leaf variety, a second characteristic average value corresponding to the morphological characteristic is calculated; According to several morphological feature data corresponding to the morphological feature, the first feature average value and the second feature average value, the effect size calculation method is used to calculate the difference between the tobacco variety to which the target tobacco object belongs and the given tobacco variety based on the morphological feature.

6. The method for distinguishing tobacco leaf varieties according to claim 5, characterized in that: The plurality of morphological feature data corresponding to the morphological feature include a plurality of data corresponding to the same morphological feature of a plurality of target tobacco leaf objects and a plurality of data corresponding to the same morphological feature of a given tobacco leaf variety.

7. The method for distinguishing tobacco leaf varieties according to claim 1, characterized in that: Obtain several morphological feature datasets corresponding to a given tobacco variety through the following steps: Based on a preset shooting height, obtaining a shooting image corresponding to a given tobacco variety in a given area shot by a remote sensing drone; An image segmentation model is used to perform tobacco leaf recognition and tobacco leaf segmentation on the captured image, and a plurality of third tobacco leaf objects are screened out and vector information of a raster image corresponding to each third tobacco leaf object is obtained; the third tobacco leaf image refers to any tobacco leaf object whose degree of integrity meets the second preset requirement; According to the vector information of the raster image corresponding to each third tobacco leaf object, the morphological feature data set corresponding to each third tobacco leaf object is calculated, and the several morphological feature data sets corresponding to several third tobacco leaf objects are used as the several morphological feature data sets corresponding to a given tobacco leaf variety.

8. A non-transitory computer-readable storage medium, wherein at least one instruction or at least one program is stored in the storage medium, characterized in that: The at least one instruction or the at least one program is loaded and executed by the processor to implement the tobacco variety differentiation method as described in any one of claims 1-7.

9. An electronic device, characterized in that: The invention comprises a processor and the non-transitory computer-readable storage medium as claimed in claim 8.

Images