Method and system for detecting selfing affinity of brassica campestris

Abstract

The invention discloses a method and a system for detecting selfing affinity of brassica campestris, and relates to the technical field of biological detection. The method comprises the following steps: collecting data of brassica campestris flowers; performing quantitative judgment on the data of the brassica campestris flowers; collecting rape flower structure characteristic data; and performing quantitative judgment on the structural feature data of the rape flowers. The method comprises the steps of obtaining an original rape flower image, quantizing image quality to obtain an image detection quantized value, judging whether to carry out image quality optimization according to the quantized value, if so, obtaining rape flower structure feature data according to an optimized result, and quantitatively judging whether to carry out structure feature optimization based on the rape flower structure feature data. The accuracy of selfing affinity detection is improved, and the problem that the detection accuracy of rape selfing affinity is low due to the difference of rape flower image quality and flower structure characteristics in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the field of biological detection technology, in particular to a method and system for detecting self-compatibility of Brassica rapa. Background Art

[0002] Self-compatibility is a key agronomic trait in plant genetics and breeding. For Brassica rapa, understanding its self-compatibility helps breeders select appropriate inbred lines, thereby improving breeding efficiency. In hybrid breeding, understanding the self-compatibility of the parents helps predict the performance of hybrid offspring and select the optimal hybrid combination. Self-compatibility influences crop propagation methods and field management strategies. Understanding self-compatibility helps develop rational planting and harvesting plans, improving agricultural production efficiency. Self-compatibility is also related to a plant's ecological adaptability. Under different environmental conditions, self-compatibility can affect a plant's reproductive success rate, thereby influencing its distribution and survival in the natural environment.

[0003] The method for detecting rapeseed self-compatibility is to pollinate anthers and stigmas in vitro through in vitro pollination, observe the germination of pollen on the stigma and the growth of pollen tubes; use fluorescent dyes to label pollen or stigma, and observe the interaction between pollen and stigma under a fluorescence microscope; use genome sequencing technology to analyze genomic regions or gene expression profiles related to self-compatibility.

[0004] For example, the invention patent with publication number CN119023638A discloses a method for detecting the self-fertility of macadamia nuts, which includes: first, arranging the inflorescence of self-pollinated pistils, then bagging the inflorescence of the macadamia nut variety to be tested, removing the inflorescence from the bag after 7 to 9 days, immersing all the pistils attached to the inflorescence axis together with the inflorescence axis in FAA fixative, washing and softening, cutting the ovary and the lower half of the style horizontally and retaining it, then cutting the retained part longitudinally and staining it, observing the growth of pollen tubes in the ovary and the lower half of the style after staining under a fluorescence microscope, and calculating the PLS.

[0005] However, in the process of implementing the technical solutions of the embodiments of the present application, the present application discovered that the above technology has at least the following technical problems:

[0006] In the microscopic images of rapeseed flowers, problems such as blurred boundaries, reflections, and overlaps between stigmas and pollen grains result in low rapeseed flower image quality and difficulty in effectively distinguishing different target areas, resulting in low rapeseed flower image segmentation accuracy. There are significant differences in the distribution of rapeseed flower image data of different varieties in the feature space. Existing feature extraction methods may not be able to capture the difference characteristics across varieties, resulting in an imbalanced rapeseed sample problem and increasing the complexity of cross-variety data distribution differences. There is a problem of low detection accuracy of rapeseed self-compatibility due to differences in rapeseed flower image quality and flower structural characteristics. Summary of the Invention

[0007] The embodiments of the present application provide a method and system for detecting the self-compatibility of rapeseed, thereby solving the problem in the prior art of low detection accuracy of rapeseed self-compatibility due to differences in rapeseed flower image quality and flower structural characteristics, and improving the accuracy of self-compatibility detection.

[0008] The embodiment of the present application provides a method for detecting the self-compatibility of Brassica rapa, comprising the following steps: obtaining an original image of a Brassica rapa flower corresponding to the Brassica rapa to be tested and performing image data extraction to obtain Brassica rapa flower data; obtaining an image detection quantization value by quantifying the quality corresponding to the Brassica rapa flower data; determining whether to perform image quality optimization based on the obtained image detection quantization value and a preset image detection threshold in a self-compatibility detection database, wherein image quality optimization means improving the quality by processing the original image of the Brassica rapa flower; if image quality optimization is performed, then performing image quality optimization based on the optimized Brassica rapa flower. The original image is subjected to feature extraction, otherwise feature extraction is directly performed to obtain rapeseed flower structural feature data; the structural features corresponding to the rapeseed flower structural feature data are quantified to obtain a structural feature quantization value; based on the obtained structural feature quantization value and a preset structural feature threshold in a self-compatibility detection database, it is determined whether to perform structural feature optimization; if structural feature optimization is performed, the self-compatibility of the rapeseed type is determined based on the structural feature optimization; otherwise, the self-compatibility is directly determined to obtain structural feature optimization; structural feature optimization means that the reliability of rapeseed pollen grain identification is improved by processing the rapeseed flower structural feature extraction.

[0009] The embodiment of the present application provides a system for detecting the self-compatibility of Brassica rapa, including a Brassica rapa flower data acquisition module, a Brassica rapa flower data quantification judgment module, a Brassica rapa flower structural feature data acquisition module, and a Brassica rapa flower structural feature data quantification judgment module: the Brassica rapa flower data acquisition module is used to obtain the original image of the Brassica rapa flower corresponding to the Brassica rapa to be tested and perform image data extraction to obtain the Brassica rapa flower data; the Brassica rapa flower data quantification judgment module is used to obtain an image detection quantization value based on the quality corresponding to the Brassica rapa flower data quantification, and determine whether to perform image quality optimization based on the obtained image detection quantization value and a preset image detection threshold in the self-compatibility detection database. Image quality optimization means that the original image of the Brassica rapa flower is processed to improve the image quality. Quality; Rapeseed flower structural feature data acquisition module: used for performing feature extraction based on the optimized original image of the Brassica rapa flower if image quality optimization is performed, otherwise directly performing feature extraction to obtain rapeseed flower structural feature data; Rapeseed flower structural feature data quantification judgment module: used for obtaining structural feature quantization values by quantifying the corresponding structural features of the rapeseed flower structural feature data, and judging whether to perform structural feature optimization based on the obtained structural feature quantization values and the preset structural feature thresholds in the self-compatibility detection database; if structural feature optimization is performed, the self-compatibility of the Brassica rapa is determined based on the structural feature optimization, otherwise directly performing self-compatibility determination to obtain structural feature optimization, which means that the reliability of rapeseed pollen grain identification is improved by processing the rapeseed flower structural feature extraction.

[0010] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0011] 1. After self-pollination of Brassica rapa, the original image of the rapeseed flower is obtained and the image quality is quantified to obtain an image detection quantization value. Based on the quantization value, it is determined whether to perform image quality optimization. If so, the rapeseed flower structural feature data is obtained according to the optimized result. Based on the quantitative determination of the rapeseed flower structural feature data, it is determined whether to perform structural feature optimization. This improves the accuracy of self-compatibility detection and solves the problem of low detection accuracy of rapeseed self-compatibility in the prior art due to differences in rapeseed flower image quality and flower structural characteristics.

[0012] 2. By quantifying image quality and flower structure, we obtain image detection and structural feature quantification values, thereby providing accurate Brassica rapa flower data. This helps to more accurately analyze the image quality and structural characteristics of flowers, and then facilitates comparison of image quality and structural characteristics of different flowers or under different processing conditions.

[0013] 3. By determining whether to optimize self-compatibility detection based on the quantitative results, and if optimization is performed, judging self-compatibility based on the optimized results, improvements can be made to deficiencies in image quality and structural features, thereby improving the accuracy of self-compatibility detection. Image quality optimization can improve the reliability of pollen grain identification, reduce false detections and missed detections, and thus increase detection flexibility. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 This is a flow chart of a method for detecting self-compatibility of Brassica rapa provided in an embodiment of the present application;

[0015] Figure 2 This is a flowchart of image quality optimization for the method for detecting self-compatibility of Brassica rapa provided in an embodiment of the present application;

[0016] Figure 3 A flowchart of structural feature optimization for the method for detecting self-compatibility of Brassica rapa provided in an embodiment of the present application;

[0017] Figure 4 This is a schematic diagram of the structure of a system for detecting self-compatibility of Brassica rapa provided in an embodiment of the present application. DETAILED DESCRIPTION

[0018] The present invention provides a method and system for detecting self-compatibility in Brassica rapa, addressing the prior art issue of low accuracy in detecting self-compatibility due to differences in rapeseed flower image quality and flower structural characteristics. The system obtains a quantitative image quality value by quantifying the image quality of the original rapeseed flower image. Based on the quantitative value, a determination is made as to whether image quality optimization should be performed. If so, rapeseed flower structural characteristic data is obtained based on the optimized result. Based on the quantitative data, a determination is made as to whether structural characteristic optimization should be performed, thereby improving the accuracy of self-compatibility detection.

[0019] The technical solution in the embodiments of the present application is to solve the problem of low accuracy in detecting rapeseed self-compatibility due to differences in rapeseed flower image quality and flower structural characteristics. The overall idea is as follows:

[0020] The method obtains an original image of a rapeseed flower, pre-processes the original image of the rapeseed flower, including operations such as denoising, to improve the image quality of the rapeseed flower, quantizes the image quality, obtains an image detection quantization value, and judges whether image quality optimization is needed based on the quantization result. If image quality optimization is needed, feature extraction is performed on the optimized original image of the Brassica rapa flower; otherwise, feature extraction is performed directly to obtain rapeseed flower structural feature data. Quantization is performed based on the rapeseed flower structural feature data to obtain a structural feature quantization value. Based on the obtained structural feature quantization value and a preset structural feature threshold in a self-compatibility detection database, it is judged whether structural feature optimization is needed. If structural feature optimization is needed, self-compatibility of the Brassica rapa flower is determined based on the structural feature optimization; otherwise, self-compatibility determination is performed directly, thereby improving the efficiency of self-compatibility detection and thereby improving the accuracy of self-compatibility detection.

[0021] In order to better understand the above technical solution, the above technical solution will be described in detail below with reference to the accompanying drawings and specific implementation methods.

[0022] like Figure 1 As shown, it is a flow chart of a method for detecting self-compatibility of Brassica rapa provided by an embodiment of the present application. The method is used in a system for detecting self-compatibility of Brassica rapa, and the method comprises the following steps: obtaining an original image of a Brassica rapa flower corresponding to the Brassica rapa to be tested and performing image data extraction to obtain Brassica rapa flower data; obtaining an image detection quantization value by quantifying the quality corresponding to the Brassica rapa flower data; judging whether to perform image quality optimization based on the obtained image detection quantization value and a preset image detection threshold in a self-compatibility detection database; image quality optimization means improving the quality by processing the original image of the Brassica rapa flower; if image quality optimization is performed, If the image quality is optimized, feature extraction is performed based on the optimized original image of the Brassica rapa flower, otherwise feature extraction is performed directly to obtain the structural feature data of the rapeseed flower; the structural feature corresponding to the rapeseed flower structural feature data is quantified to obtain a structural feature quantization value, and whether to perform structural feature optimization is determined based on the obtained structural feature quantization value and the preset structural feature threshold in the self-compatibility detection database. If structural feature optimization is performed, the self-compatibility of the Brassica rapa flower is determined based on the structural feature optimization, otherwise the self-compatibility determination is performed directly. Structural feature optimization means that the reliability of rapeseed pollen grain identification is improved by processing the structural feature extraction of the rapeseed flower.

[0023] It should be understood that before obtaining the original image of the cabbage-type rapeseed flower, the process also includes: removing the stamens from the cabbage-type rapeseed flower; after removing the stamens from the flower, covering the cabbage-type rapeseed flower with an isolation bag; and when the cabbage-type rapeseed flower matures, shaking the isolation bag to transfer pollen from the same cabbage-type rapeseed flower to the stigma to achieve self-pollination.

[0024] In this embodiment, the stamens in the rapeseed flower need to be removed, usually when the rapeseed flower has not yet fully opened to prevent interference from foreign pollen. Fine tweezers or a emasculation tool can be used to remove the stamens. After removing the stamens, the processed rapeseed flower is enclosed in an isolation bag. The isolation bag is a transparent plastic bag or paper bag. Its purpose is to prevent the entry of foreign pollen and ensure that the subsequent pollination process only uses the rapeseed flower's own pollen. The rapeseed flower is allowed to mature naturally. When the rapeseed flower matures, the isolation bag is shaken so that pollen from the same rapeseed flower falls onto the stigma of the pistil, achieving self-pollination. The quality of the original image of the rapeseed flower is uneven due to image contrast, resulting in light and shade differences in different areas of the original image of the rapeseed flower, which in turn affects subsequent analysis. Since the structural characteristics of rapeseed flowers include but are not limited to the shape, size, color, texture of the petals, and the morphology of the stigma and pollen grains, these differences in flower structural characteristics increase the difficulty of image recognition, resulting in low accuracy in detecting rapeseed self-compatibility. Through the image quality optimization of the present application, the original image can be processed to improve the image quality. Through structural feature optimization, the flower structural feature extraction can be processed to improve the reliability of pollen grain identification, ultimately improving the accuracy of self-compatibility detection.

[0025] Specifically, the rapeseed flower data includes rapeseed flower image contrast, rapeseed flower image pixel intensity, rapeseed flower image artifacts and image noise intensity; the rapeseed flower structural feature data includes the maximum mean difference of rapeseed flower varieties, overlapping target missegmentation rate, image detection quantization value and image noise second intensity; the maximum mean difference of rapeseed flower varieties is used to measure the distribution distance of the rapeseed flower structure in the feature space; the overlapping target missegmentation rate represents the proportion of incorrect segmentation of the intersection area of the adhering pollen grains and anthers in the original image of the rapeseed flower.

[0026] It should be added that the image detection quantization value is obtained by quantifying the corresponding quality of the cabbage-type rapeseed flower data, which previously also included obtaining the image detection threshold, image detection correction, structural feature threshold and structural feature correction from the constructed self-compatibility detection database; the image detection threshold includes the rapeseed flower image contrast threshold, the rapeseed flower image pixel intensity standard value, the rapeseed flower image artifact threshold and the image noise intensity threshold; the image detection correction includes the rapeseed flower image contrast correction, the rapeseed flower image pixel intensity correction, the rapeseed flower image artifact correction and the image noise intensity correction; the structural feature threshold includes the maximum mean difference threshold of rapeseed flower varieties, the overlapping target missegmentation rate threshold and the image noise intensity second threshold; the structural feature correction includes the maximum mean difference correction of rapeseed flower varieties, the overlapping target missegmentation rate correction, the image detection correction and the image noise intensity second correction.

[0027] It should be added that before designing the frequency domain feature effectiveness optimization method, a self-affinity detection database for storing various types of data is established by the preset personnel. The self-affinity detection database includes but is not limited to rapeseed flower image contrast, rapeseed flower image pixel intensity, rapeseed flower image artifacts and overlapping target missegmentation rate, etc., among which various numerical values are directly set by preset professionals. For example, the rapeseed flower image contrast threshold is obtained by the preset staff based on the specific restriction expression of the image contrast threshold of rapeseed flower images in the historical database within the preset frequency range. The rapeseed flower image contrast threshold is obtained by the preset staff based on the specific restriction expression of the image contrast threshold of rapeseed flower images in the historical database within the preset frequency range. The mean operation is performed on the data set to obtain the rapeseed flower image contrast threshold, and it is pre-stored in the preset database.

[0028] Furthermore, the specific steps for obtaining the image detection quantization value are as follows: correcting the analysis result of the ratio of the rape flower image contrast and the rape flower image contrast threshold by using the rape flower image contrast correction amount to obtain the image detection contrast influence value; correcting the analysis result of the ratio of the rape flower image pixel intensity standard value and the rape flower image pixel intensity deviation degree by using the rape flower image pixel intensity correction amount to obtain the image detection pixel intensity influence value; correcting the analysis result of the ratio of the rape flower image artifact threshold and the rape flower image artifact by using the rape flower image artifact correction amount to obtain the image detection artifact influence value; correcting the analysis result of the image noise intensity threshold and the image noise intensity ratio by using the image noise intensity correction amount to obtain the image detection noise intensity influence value; coupling analysis of the image detection contrast influence value, the image detection pixel intensity influence value, the image detection artifact influence value and the image detection noise intensity influence value to obtain the image detection quantization value; the image detection quantization value represents the quantitative data of the degree of influence of the rape flower image contrast, the rape flower image pixel intensity, the rape flower image artifact and the image noise intensity on the rape flower image quality.

[0029] In this embodiment, the specific formula for obtaining the image detection quantization value is:

[0030]

[0031] δ represents the image detection quantization value. D represents the rape flower image contrast. The original image of the rape flower of the Chinese rape type is read using image processing software (such as OpenCV, etc.), and the difference between the maximum and minimum pixel intensity values in the original image of the rape flower of the Chinese rape type is recorded as the rape flower image contrast. D1 represents the rape flower image contrast threshold obtained from the self-compatibility detection database. Q represents the rape flower image pixel intensity. The rape flower image pixel intensity is quantified by calculating the histogram of the original image of the rape flower of the Chinese rape type, and the histogram of the rape flower image is further analyzed to calculate the average value of the rape flower image brightness value, that is, the rape flower image pixel intensity. Q1 represents the rape flower image pixel intensity standard value obtained from the self-compatibility detection database. W1 represents the rape flower image artifact threshold obtained from the self-compatibility detection database. W represents rapeseed flower image artifacts. Artifact detection algorithms (such as edge detection and morphological operations) are used to identify and quantify artifacts in rapeseed flower images. The gradient strength and direction of the original rapeseed flower image are calculated. The gradient amplitude is refined. Edges in the original rapeseed flower image are determined, edge pixels are connected, and the number of detected edge pixels, edge length, and edge density (the ratio of the number of edge pixels to the total image area) are calculated. Decision fusion techniques are used to determine rapeseed flower image artifacts. Z1 represents the image noise intensity threshold obtained from the self-compatibility detection database. Z represents the image noise intensity. Image processing software (such as the Python OpenCV library) is used to extract the noise component from the original rapeseed flower image and obtain the image noise intensity.

[0032] is the preset rapeseed flower image contrast correction value obtained from the self-compatibility detection database. is the preset rapeseed flower image pixel intensity correction value obtained from the self-compatibility detection database.

[0033] is the preset rapeseed flower image artifact correction value obtained from the self-compatibility detection database.

[0034] is the preset image noise intensity correction value obtained from the self-compatibility detection database.

[0035] A mapping table of correction values is obtained from a self-compatibility test database. For example, a table of correction values is constructed based on historical rapeseed flower image contrast, rapeseed flower image pixel intensity, rapeseed flower image artifacts, and image noise intensity, along with corresponding correction values. This table defines a clear set of association rules that converts specific values for rapeseed flower image contrast, rapeseed flower image pixel intensity, rapeseed flower image artifacts, and image noise intensity into their corresponding correction values. This mechanism effectively enables dynamic acquisition of correction values, whether achieving a one-to-one exact match or aggregating multiple parameters into a many-to-one relationship.

[0036] The higher the rapeseed flower image artifact, the more it will interfere with the normal brightness distribution of the image, making the target and background that should be clearly distinguished in the image become blurred, and the rapeseed flower image contrast is lower; the higher the image noise intensity, the more random brightness fluctuations will be introduced, which will blur the boundaries between different areas in the image, making the image look more blurred, and the rapeseed flower image contrast is lower; the larger the rapeseed flower image artifact, the more image noise will increase, and the higher the image noise intensity.

[0037] There is a positive correlation between the contrast of rapeseed flower images and the image detection quantization value. The higher the contrast of rapeseed flower images, the clearer the image can show the structural features of rapeseed flowers, and the higher the image detection quantization value. There is a negative correlation between the absolute value of the difference between the pixel intensity of rapeseed flower images and the standard value of the pixel intensity of rapeseed flower images and the image detection quantization value. The larger the absolute value of the difference between the pixel intensity of rapeseed flower images and the standard value of the pixel intensity of rapeseed flower images, the more details in the highlight area will be lost, and the lower the image detection quantization value. There is a negative correlation between the artifacts of rapeseed flower images and the image detection quantization value. The larger the artifacts of rapeseed flower images, the more they will interfere with the real information of the image, making feature extraction difficult, and the smaller the image detection quantization value. There is a negative correlation between the image noise intensity and the image detection quantization value. The higher the image noise intensity, the more random brightness fluctuations will be introduced. These fluctuations will interfere with the accuracy of feature extraction, and the smaller the image detection quantization value.

[0038] By analyzing the correlation between rapeseed flower image contrast, rapeseed flower image pixel intensity, rapeseed flower image artifacts and image noise intensity and image detection quantization values, we can more accurately evaluate rapeseed flower image detection, optimize image detection methods, and enhance image detection accuracy.

[0039] After performing image detection contrast optimization and image channel optimization, the image quality of the rapeseed flower is evaluated again. If the image detection quantization value is greater than or equal to the preset image detection threshold obtained from the self-compatibility detection database, it indicates that the rapeseed flower image quality is qualified and the next step of feature extraction is carried out. Otherwise, the rapeseed flower image data is re-acquired.

[0040] Furthermore, the specific steps for obtaining the structural feature quantization value are as follows: correcting the analysis result of the ratio of the maximum mean difference threshold of rapeseed flower varieties and the maximum mean difference of rapeseed flower varieties by the maximum mean difference correction amount of rapeseed flower varieties to obtain the structural feature maximum mean difference influence value; correcting the analysis result of the ratio of the overlapping target error segmentation rate and the overlapping target error segmentation rate threshold by the overlapping target error segmentation rate correction amount to obtain the structural feature error segmentation rate influence value; correcting the result of the multiplication inverse of the image detection quantization value by the image detection quantization value correction amount to obtain the image detection influence value; correcting the analysis result of the ratio of the second image noise intensity and the second image noise intensity threshold by the second image noise intensity correction amount to obtain the structural feature noise intensity second influence value; coupling the structural feature maximum mean difference influence value, the structural feature error segmentation rate influence value, the image detection influence value and the structural feature noise intensity second influence value to obtain the structural feature quantization value; the structural feature quantization value represents the quantitative data of the degree of influence of the maximum mean difference of rapeseed flower varieties, the overlapping target error segmentation rate, the image detection quantization value and the image noise intensity on the extraction of rapeseed flower structural features.

[0041] In this embodiment, the specific formula for obtaining the quantitative value of the structural feature through analysis is:

[0042]

[0043] ρ1+ρ2+ρ3+ρ4=1;

[0044] θ represents the quantified value of structural features. C represents the maximum mean difference between rapeseed flower varieties, which measures the distribution distance in feature space between the current rapeseed flower image and the pre-set self-compatible rapeseed flower images obtained from the self-compatibility database. A pre-trained model (e.g., ResNet) is used to extract deep features of the rapeseed flower image (e.g., stigma geometry, such as spots, stripes, and veins, and petal texture features, such as leaves and stems). A Gaussian kernel is applied to the deep features, and the distribution distance is calculated based on the kernel matrix to obtain the maximum mean difference between rapeseed flower varieties. C1 represents the maximum mean difference threshold for rapeseed flower varieties obtained from the self-compatibility database. F represents the overlapping object missegmentation rate, which quantifies the missegmentation rate of overlapping objects, such as adherent pollen grains and the intersection of stigma and anther. Using annotated image data, each object is annotated with its category and boundary. A segmentation algorithm (e.g., U-Net) is used to segment the image, and the overlap between the segmented result and the annotated data is calculated. F1 represents the overlapping object missegmentation rate threshold obtained from the self-compatibility database. δ represents the image detection quantization value. P represents the second image noise intensity, which refers to the image noise intensity of the segmented rapeseed flower image. This intensity is calculated by using a signal processing library such as SciPy to calculate the signal-to-noise ratio. P1 represents the second threshold of image noise intensity obtained from the self-compatibility detection database.

[0045] ρ1 is the maximum mean difference correction value of the preset rapeseed flower varieties obtained from the self-compatibility detection database.

[0046] ρ2 is the preset overlapping target mis-segmentation rate correction value obtained from the self-compatibility detection database.

[0047] ρ3 is a preset image detection quantization value correction value obtained from the self-compatibility detection database.

[0048] ρ4 is a second correction value of the preset image noise intensity obtained from the self-compatibility detection database.

[0049] A mapping table of correction values is obtained from the self-compatibility detection database. For example, a correction value mapping table is constructed based on historical maximum mean differences among rapeseed flower varieties, overlapping target missegmentation rates, image detection quantization values, and the second intensity of image noise, along with corresponding correction values. This table defines a clear set of association rules that converts specific values for the maximum mean differences among rapeseed flower varieties, overlapping target missegmentation rates, image detection quantization values, and the second intensity of image noise into their corresponding correction values. This mechanism effectively achieves dynamic acquisition of correction values, whether achieving a one-to-one precise match or converging multiple parameters into a many-to-one relationship.

[0050] The greater the maximum mean difference between rapeseed flower varieties, the easier it is to distinguish different varieties of flowers, and the lower the mis-segmentation rate of overlapping targets; the greater the second intensity of image noise, the more it will interfere with the accuracy of the image segmentation algorithm, making it difficult for the algorithm to correctly distinguish overlapping flowers, which will increase the mis-segmentation rate of overlapping targets.

[0051] There is a negative correlation between the maximum mean difference of rapeseed flower varieties and the quantization value of structural features. The greater the maximum mean difference of rapeseed flower varieties, the greater the difference between varieties, the clearer the structural features, and the smaller the quantization value of structural features. There is a positive correlation between the overlapping target missegmentation rate and the quantization value of structural features. The greater the overlapping target missegmentation rate, the unclear structural features or high similarity, and the larger the quantization value of structural features. There is a negative correlation between the image detection quantization value and the quantization value of structural features. The larger the image detection quantization value, the better the image quality and the smaller the quantization value of structural features. There is a positive correlation between the image noise intensity and the quantization value of structural features. The greater the image noise intensity, the more difficult it is to clearly capture the structural features of the flower, and the larger the quantization value of structural features.

[0052] By analyzing the maximum mean difference among rapeseed flower varieties, the overlapping target missegmentation rate, the image detection quantization value, and the correlation between the image noise intensity and the structural feature quantization value, the structural characteristics of rapeseed flowers can be evaluated more accurately, the rapeseed flower structural feature method can be optimized, and the rapeseed flower structural feature extraction can be enhanced.

[0053] Furthermore, whether to perform image quality optimization is determined based on the obtained image detection quantization value and the preset image detection threshold in the self-affinity detection database. The specific steps are: comparing the image detection quantization value with the preset image detection threshold obtained from the self-affinity detection database. If the image detection quantization value is greater than or equal to the preset image detection threshold obtained from the self-affinity detection database, image quality optimization is not performed; otherwise, image quality optimization is performed; if the structural feature quantization value is lower than or equal to the preset structural feature threshold obtained from the self-affinity detection database, structural feature optimization is not performed; otherwise, structural feature optimization is performed.

[0054] Specifically, the specific steps of image quality optimization are as follows: an image detection deviation value is obtained based on an image detection quantization value and a preset image detection threshold, and the image detection deviation value is used to measure the degree of quality deviation of the rapeseed flower image; if the image detection deviation value is lower than or equal to the safe image detection deviation threshold in the self-affinity detection database, the rapeseed flower image is recorded as a first-level sample; if the image detection deviation value is greater than the safe image detection deviation threshold and less than the preset image detection deviation threshold, the rapeseed flower image is recorded as a second-level sample, and image detection contrast optimization is performed on the second-level sample, where the image detection contrast optimization means increasing the brightness of areas with image pixel values lower than the preset pixel value threshold by the obtained image brightness adjustment amount, and decreasing the brightness of areas with image pixel values greater than the preset pixel value threshold by the obtained image brightness adjustment amount to improve the image contrast, and the image brightness adjustment amount is obtained by inputting the image detection quantization value into a preset self-affinity detection database for mapping; if the image detection deviation value is greater than or equal to the preset image detection deviation threshold, the rapeseed flower image is recorded as a third-level sample, and image channel optimization is performed on the third-level sample to improve the visual quality of the rapeseed flower image; image quality optimization includes image detection contrast optimization and image channel optimization.

[0055] In this embodiment, if Figure 2 The figure shows an image quality optimization flow chart for the method for detecting self-compatibility of Brassica rapa provided by an embodiment of the present application. The specific logic is as follows: If the image detection quantization value is lower than the preset image detection threshold, image quality optimization is performed: the image detection quantization value is compared with the preset image detection threshold to obtain an image deviation value. If the image deviation value is lower than the safe image deviation threshold, it indicates that the sample quality is high, and the rapeseed flower image is classified as a first-level sample, and no subsequent processing feedback is required, indicating that it is within a controllable range; if the image deviation value is greater than the safe image deviation threshold and less than the preset image deviation threshold, the rapeseed flower image is classified as a second-level sample; if the image deviation value is greater than the preset image deviation threshold, the rapeseed flower image is classified as a third-level sample; if the sample is a second-level sample, adjustment is performed through image contrast optimization. The specific process of image contrast optimization is: the dark and bright areas of the rapeseed flower image are processed separately through multi-scale gamma correction, and the areas with image pixel values below the preset pixel value threshold are classified as dark areas, and the areas with image pixel values greater than the preset pixel value threshold are classified as bright areas. The brightness of the dark and bright parts of the rapeseed flower image is adjusted according to the image detection quantization value.

[0056] For areas of the rapeseed flower image where the image pixel value is lower than the preset pixel value threshold obtained from the self-affinity detection database, the brightness of the dark area is adjusted according to the image detection quantization value, and the brightness of the dark area corresponding to the image detection quantization value preset in the database is obtained according to the current image detection quantization value as the adjustment value of the brightness increase. An adjustment value is matched according to the image detection quantization value, and the adjustment value is determined based on the image detection quantization value. This adjustment value is added to the brightness of the original dark area.

[0057] For areas of the rapeseed flower image where the image pixel value is greater than the preset pixel value threshold obtained from the self-affinity detection database, the brightness of the bright area of the image is adjusted according to the image detection quantization value, and the brightness of the bright area of the image corresponding to the preset image detection quantization value in the database is obtained according to the current image detection quantization value as the adjustment value of the brightness reduction amount. An adjustment value is matched according to the image detection quantization value, and the adjustment value is determined based on the image detection quantization value, and this adjustment value is added to the brightness of the original bright area of the image.

[0058] Furthermore, image channel optimization means stretching image channels with image contrast lower than a preset image contrast by obtaining the image red and blue channel variation, and compressing image channels with image contrast greater than a preset image contrast by obtaining the image red and blue channel variation to improve the accuracy of image detection; image contrast is used to reflect the degree of color difference in the original image of the cabbage-type rapeseed flower; channel variation is obtained by inputting the image detection quantization value into a preset self-compatibility detection database for mapping.

[0059] In this embodiment, if the sample is a three-level sample, the target rapeseed flower color parameters are set according to the color of the rapeseed flower (such as the RGB value of the yellow petals), and the color deviation caused by illumination is eliminated through multi-scale gamma correction to restore the true color of the white or neutral gray area. The mean of the RGB (Red, Green, Blue) channels is adjusted based on the image detection quantization value to achieve white balance, and the mean (R1, R2, R3) of the three channels R, G, and B of the rapeseed flower image are calculated respectively. Taking the G channel as the benchmark, the R and B channels of each pixel are scaled based on the image detection quantization value (scaling is applied to the petal area), and the G channel remains unchanged.

[0060] If the image contrast is lower than the preset image contrast, the distribution of the red and blue channels is stretched, and the stretching range is adjusted according to the image detection quantization value to expand the red and blue channel ranges. The expansion standard is to obtain the red and blue channel ranges corresponding to the preset image detection quantization values in the database according to the current image detection quantization value as the red and blue channel range expansion adjustment value, and match an adjustment value according to the image detection quantization value. The adjustment value is determined based on the image detection quantization value, and this adjustment value is added to the original red and blue channel range size.

[0061] If the image contrast is greater than the preset image contrast, the red and blue channels are compressed to reduce the deviation, and the compression range is adjusted according to the image detection quantization value. The red and blue channel ranges are compressed. The compression standard is to obtain the red and blue channel ranges corresponding to the preset image detection quantization values in the database according to the current image detection quantization value as the red and blue channel range compression adjustment values. An adjustment value is matched according to the image detection quantization value. The adjustment value is determined based on the image detection quantization value, and this adjustment value is added to the original red and blue channel range size.

[0062] Through the above steps, the dynamic range of the red and blue channels can be dynamically adjusted based on the image deviation value, and the image contrast and color accuracy can be targeted. Gamma correction is used to enhance the contrast of the G channel. The enhanced G channel is merged with the original R and B channels to make target areas such as pollen grains more clearly visible, thereby improving image quality.

[0063] Furthermore, the specific steps of structural feature optimization are: obtaining the pollen grain number density and comparing it with the corresponding preset pollen grain number density threshold; if the pollen grain number density is less than or equal to the preset pollen grain number density threshold, adjusting the pixel number interval of the pollen grain area, wherein the pixel number interval adjustment of the pollen grain area means expanding the initial pixel number interval of the pollen grain area by the obtained pixel number interval change to improve the accuracy of pollen grain identification, and the pixel number interval change is input into the preset self-compatibility detection database based on the structural feature quantization value for mapping; if the pollen grain number density is greater than the preset pollen grain number density threshold, optimizing the pixel number interval of the pollen grain area, wherein the pixel number interval optimization of the pollen grain area means narrowing the pixel number interval of the pollen grain area by the obtained pixel number interval adjustment to improve the accuracy of pollen grain identification, and the pixel number interval adjustment is input into the preset self-compatibility detection database based on the structural feature quantization value for mapping.

[0064] In this embodiment, if Figure 3As shown, this is a structural feature optimization flow chart of a method for detecting self-compatibility of Brassica rapa provided in an embodiment of the present application. Stigma morphological features and pollen grain distribution features are extracted from segmented rapeseed flower images. In image analysis, overlapping pollen grains lead to missegmentation, thereby affecting the accurate counting of the number of pollen grains. The higher the pollen grain number density, the greater the possibility of overlap, and the missegmentation rate will also increase accordingly.

[0065] Image processing software (such as ImageJ, Matlab, Python, etc.) was used to analyze the pollen images taken under a microscope. The pollen grains and non-pollen grain areas were distinguished by setting the pixel number interval of the pollen grain area. The minimum value in the pixel number interval of the pollen grain area represented the minimum number of pixels occupied by the pollen grain area, and the maximum value in the pixel number interval of the pollen grain area represented the maximum number of pixels occupied by the pollen grain area.

[0066] If the pollen grain number density is less than or equal to the preset pollen grain number density threshold, it means that the pollen grains are missed. By expanding the pixel number interval of the pollen grain area, the pollen grains are re-identified through the adjusted pixel number interval, and the missed parts are supplemented. The pixel number interval of the pollen grain area is adjusted according to the structural feature quantization value. The pixel number interval change corresponding to the structural feature quantization value preset in the database is obtained according to the current structural feature quantization value. A pixel number interval change is matched according to the structural feature quantization value. The pixel number interval change is determined based on the amount of the structural feature quantization value. This pixel number interval change is added to the maximum value of the pixel number interval of the original pollen grain area. The minimum value of the pixel number interval of the pollen grain area is subtracted from the pixel number interval change to obtain the adjusted pixel number interval of the pollen grain area.

[0067] If the pollen grain number density is greater than the preset pollen grain number density threshold, it means that the pollen grains are too dense. By narrowing the pixel number interval of the pollen grain area, reducing the small particles identified as pollen grains, establishing a spatial grid, counting the local density, filtering out pollen points in the overcrowded area, and adjusting the pixel number interval of the pollen grain area according to the structural feature quantization value, the pixel number interval adjustment amount corresponding to the structural feature quantization value preset in the database is obtained according to the current structural feature quantization value, and a pixel number interval adjustment amount is matched according to the structural feature quantization value. The pixel number interval adjustment amount is determined based on the amount of the structural feature quantization value, and this pixel number interval adjustment amount is added to the minimum value of the original pixel number interval of the pollen grain area. The pixel number interval adjustment amount is subtracted from the maximum value of the pixel number interval of the pollen grain area to obtain the adjusted pixel number interval of the pollen grain area.

[0068] Based on the above optimization and adjustment steps, fluorescence microscopy was used to observe, and fluorescent dyes were used to mark pollen and stigma. The interaction between pollen and stigma was observed under a fluorescence microscope to determine self-compatibility. The determination process was as follows: under a fluorescence microscope, pollen tubes were observed extending from pollen grains and growing along the stigma tissue, pollen successfully germinated on the stigma, and the pollen tubes were able to grow normally and penetrate the stigma tissue, indicating that the cabbage-type rapeseed flower variety was self-compatible; under a fluorescence microscope, pollen grains were observed to aggregate on the stigma, but the pollen tubes failed to grow normally or their growth was inhibited, the pollen could not germinate on the stigma, or the pollen tube growth was inhibited and could not penetrate the stigma tissue, indicating that the cabbage-type rapeseed flower variety was self-incompatible.

[0069] like Figure 4 As shown, it is a structural schematic diagram of the detection system for self-pollination affinity of Brassica rapa provided in an embodiment of the present application. The detection system for self-pollination affinity of Brassica rapa provided in an embodiment of the present application includes: a Brassica rapa self-pollination module, a Brassica rapa flower data acquisition module, a Brassica rapa flower data quantification judgment module, a Brassica rapa flower structural feature data acquisition module and a Brassica rapa flower structural feature data quantification judgment module: the Brassica rapa self-pollination module is used to remove the stamens from the Brassica rapa flower, after removing the stamens from the flower, cover the Brassica rapa flower with an isolation bag, and when the Brassica rapa flower matures, shake the isolation bag to transfer pollen from the same Brassica rapa flower to the stigma to achieve self-pollination; the Brassica rapa flower data acquisition module is used to obtain the original image of the Brassica rapa flower corresponding to the Brassica rapa to be tested and perform image data extraction to obtain the Brassica rapa flower data; the Brassica rapa flower data quantification judgment module is used to obtain the image detection quantity according to the quality corresponding to the quantification of the Brassica rapa flower data quantization value, and judging whether to perform image quality optimization based on the obtained image detection quantization value and the preset image detection threshold in the self-affinity detection database, where image quality optimization means processing the original image of the Brassica rapa flower to improve the quality; a rapeseed flower structural feature data acquisition module: for performing feature extraction based on the optimized original image of the Brassica rapa flower if image quality optimization is performed, otherwise directly performing feature extraction to obtain rapeseed flower structural feature data; a rapeseed flower structural feature data quantization judgment module: for obtaining structural feature quantization values corresponding to structural features quantified by the rapeseed flower structural feature data, judging whether to perform structural feature optimization based on the obtained structural feature quantization values and the preset structural feature threshold in the self-affinity detection database, and if structural feature optimization is performed, performing Brassica rapa self-affinity judgment based on the structural feature optimization, otherwise directly performing self-affinity judgment, where structural feature optimization means processing the extracted structural features of the rapeseed flower to improve the reliability of rapeseed pollen grain identification.

[0070] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0071] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0072] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0073] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0074] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0075] Obviously, those skilled in the art may make various modifications and variations to the present invention without departing from the spirit and scope of the present invention. Thus, if such modifications and variations fall within the scope of the claims and their equivalents, the present invention is intended to include such modifications and variations.

Claims

1. A method for detecting self-compatibility of Brassica rapa, characterized in that: The following steps are involved: Obtaining an original image of a Brassica rapa flower corresponding to the Brassica rapa to be tested and performing image data extraction to obtain Brassica rapa flower data; obtaining an image detection quantization value by quantifying the quality corresponding to the Brassica rapa flower data, and determining whether to perform image quality optimization based on the obtained image detection quantization value and a preset image detection threshold in a self-compatibility detection database, wherein the image quality optimization means processing the original Brassica rapa flower image to improve its quality; If image quality optimization is performed, feature extraction is performed based on the optimized original image of the rapeseed flower; otherwise, feature extraction is performed directly to obtain rapeseed flower structural feature data; The structural feature quantification value of the corresponding structural feature is obtained by quantifying the structural feature data of the rapeseed flower. Based on the obtained structural feature quantification value and the preset structural feature threshold in the self-compatibility detection database, it is determined whether to perform structural feature optimization. If structural feature optimization is performed, the self-compatibility of the rapeseed type is determined based on the structural feature optimization; otherwise, the self-compatibility is determined directly. The structural feature optimization means that the reliability of rapeseed pollen grain identification is improved by processing the structural features extracted from the rapeseed flower.

2. The method for detecting self-compatibility of Brassica rapa according to claim 1, wherein: The rapeseed flower data includes rapeseed flower image contrast, rapeseed flower image pixel intensity, rapeseed flower image artifacts and image noise intensity; The rape flower structural feature data includes the maximum mean difference of rape flower varieties, the overlapping target mis-segmentation rate, the image detection quantization value and the second intensity of image noise; The maximum mean difference of rapeseed flower varieties is used to measure the distribution distance of the rapeseed flower structure in the feature space; The overlapping target mis-segmentation rate represents the mis-segmentation ratio of the intersection area of the adhering pollen grains and anthers in the original image of the rapeseed flower.

3. The method for detecting self-compatibility of Brassica rapa according to claim 2, wherein: The method of obtaining an image detection quantization value by quantifying the quality of the rapeseed flower data may also include obtaining an image detection threshold, an image detection correction value, a structural feature threshold, and a structural feature correction value from a constructed self-compatibility detection database; The image detection thresholds include a rape flower image contrast threshold, a rape flower image pixel intensity standard value, a rape flower image artifact threshold, and an image noise intensity threshold; The image detection correction amount includes a rape flower image contrast correction amount, a rape flower image pixel intensity correction amount, a rape flower image artifact correction amount and an image noise intensity correction amount; The structural feature thresholds include a maximum mean difference threshold of rapeseed flower varieties, an overlapping target mis-segmentation rate threshold, and a second image noise intensity threshold; The structural feature correction amount includes a maximum mean difference correction amount of rapeseed flower varieties, an overlapping target mis-segmentation rate correction amount, an image detection correction amount and a second image noise intensity correction amount.

4. The method for detecting self-compatibility of Brassica rapa according to claim 3, wherein: The specific steps for obtaining the image detection quantization value are: The contrast correction value of the rape flower image is used to correct the ratio analysis result of the rape flower image contrast and the rape flower image contrast threshold, and the image detection contrast impact value is obtained; The pixel intensity correction value of the rape flower image is used to correct the percentage analysis result of the deviation between the standard value of the rape flower image pixel intensity and the standard value of the rape flower image pixel intensity, and the pixel intensity influence value of the image detection is obtained; The rape flower image artifact correction amount is used to correct the rape flower image artifact threshold and the rape flower image artifact ratio analysis result to obtain the image detection artifact impact value; The image noise intensity correction value is used to correct the image noise intensity threshold and the image noise intensity ratio analysis results to obtain the image detection noise intensity impact value; The image detection contrast influence value, the image detection pixel intensity influence value, the image detection artifact influence value and the image detection noise intensity influence value are coupled and analyzed to obtain the image detection quantization value; The image detection quantization value represents quantitative data of the degree of influence of rapeseed flower image contrast, rapeseed flower image pixel intensity, rapeseed flower image artifacts and image noise intensity on the rapeseed flower image quality.

5. The method for detecting self-compatibility of Brassica rapa according to claim 3, wherein: The specific steps for obtaining the quantitative value of the structural feature are: The maximum mean difference correction value of rapeseed flower varieties was used to correct the maximum mean difference threshold of rapeseed flower varieties and the analysis results of the proportion of the maximum mean difference of rapeseed flower varieties, and the maximum mean difference influence value of structural characteristics was obtained; The overlapping target mis-segmentation rate correction amount is used to correct the analysis result of the ratio of the overlapping target mis-segmentation rate to the overlapping target mis-segmentation rate threshold, and the influence value of the structural feature mis-segmentation rate is obtained; The result of multiplication inverse element of the image detection quantization value is corrected by the image detection quantization value correction amount to obtain the image detection influence value; Correcting the analysis result of the ratio of the second image noise intensity to the second image noise intensity threshold by using the second correction amount of the image noise intensity to obtain a second influence value of the structural feature noise intensity; The maximum mean difference influence value of the structural feature, the influence value of the structural feature mis-segmentation rate, the image detection influence value and the second influence value of the structural feature noise intensity are coupled and calculated to obtain the structural feature quantization value; The structural feature quantization value represents quantitative data of the degree of influence of the maximum mean difference of rapeseed flower varieties, overlapping target missegmentation rate, image detection quantization value and image noise intensity on the extraction of rapeseed flower structural features.

6. The method for detecting self-compatibility of Brassica rapa according to claim 1, wherein: The specific steps of determining whether to perform image quality optimization based on the obtained image detection quantization value and the preset image detection threshold in the self-affinity detection database are as follows: Comparing the image detection quantization value with the preset image detection threshold obtained from the self-affinity detection database, if the image detection quantization value is greater than or equal to the preset image detection threshold obtained from the self-affinity detection database, then no image quality optimization is performed, otherwise, image quality optimization is performed; If the quantified value of the structural feature is lower than or equal to the preset structural feature threshold value obtained from the self-compatibility detection database, the structural feature optimization is not performed; otherwise, the structural feature optimization is performed.

7. The method for detecting self-compatibility of Brassica rapa according to claim 6, wherein: The specific steps of image quality optimization are: Obtaining an image detection deviation value based on the image detection quantization value and a preset image detection threshold, wherein the image detection deviation value is used to measure the quality deviation degree of the rapeseed flower image; If the image detection deviation value is lower than or equal to the safety image detection deviation threshold in the self-compatibility detection database, the rapeseed flower image is recorded as a first-level sample; If the image detection deviation value is greater than the safe image detection deviation threshold and less than the preset image detection deviation threshold, the rape flower image is recorded as a secondary sample, and image detection contrast optimization is performed on the secondary sample. The image detection contrast optimization means increasing the brightness of areas where the image pixel values are lower than the preset pixel value threshold by the obtained image brightness adjustment amount, and decreasing the brightness of areas where the image pixel values are greater than the preset pixel value threshold by the obtained image brightness adjustment amount to improve the image contrast. The image brightness adjustment amount is obtained by inputting the image detection quantization value into a preset self-compatibility detection database for mapping; If the image detection deviation value is greater than or equal to the preset image detection deviation threshold, the rape flower image is recorded as a third-level sample, and the image channel optimization is performed on the third-level sample to improve the visual quality of the rape flower image; The image quality optimization includes image detection contrast optimization and image channel optimization.

8. The method for detecting self-compatibility of Brassica rapa according to claim 7, wherein: The image channel optimization means stretching the image channel whose image contrast is lower than the preset image contrast by obtaining the variation of the red and blue channels of the image, and compressing the image channel whose image contrast is greater than the preset image contrast by obtaining the variation of the red and blue channels of the image to improve the accuracy of image detection; The image contrast is used to reflect the degree of color difference in the original image of the Brassica rapa flower; The channel variation is obtained by inputting the image detection quantization value into a preset self-compatibility detection database for mapping.

9. The method for detecting self-compatibility of Brassica rapa according to claim 6, wherein: The specific steps of performing structural feature optimization are: Obtain the pollen grain number density and compare it with the corresponding preset pollen grain number density threshold; If the pollen grain number density is less than or equal to a preset pollen grain number density threshold, adjusting the pixel number interval of the pollen grain area, wherein the pixel number interval adjustment of the pollen grain area means expanding the initial pixel number interval of the pollen grain area by using the obtained pixel number interval variation to improve the accuracy of pollen grain recognition, wherein the pixel number interval variation is obtained by mapping the input of the structural feature quantization value into a preset self-compatibility detection database; If the pollen grain number density is greater than a preset pollen grain number density threshold, the pixel number interval of the pollen grain area is optimized. The pixel number interval optimization of the pollen grain area means reducing the pixel number interval of the pollen grain area by the obtained pixel number interval adjustment amount to improve the accuracy of pollen grain identification. The pixel number interval adjustment amount is input based on the quantized value of the structural feature and mapped into a preset self-compatibility detection database.

10. A system for detecting the self-compatibility of Brassica rapa, characterized in that: It includes a cabbage-type rapeseed flower data acquisition module, a cabbage-type rapeseed flower data quantification judgment module, a rapeseed flower structure feature data acquisition module and a rapeseed flower structure feature data quantification judgment module: The Brassica rapa flower data acquisition module is used to obtain the original image of the Brassica rapa flower corresponding to the Brassica rapa to be tested and perform image data extraction to obtain the Brassica rapa flower data; The Brassica rapa flower data quantification judgment module is configured to obtain an image detection quantization value based on the quality corresponding to the Brassica rapa flower data quantification, and determine whether to perform image quality optimization based on the obtained image detection quantization value and a preset image detection threshold in the self-compatibility detection database, wherein the image quality optimization means improving the quality of the Brassica rapa flower by processing the original image of the Brassica rapa flower; The rape flower structural feature data acquisition module is used to extract features based on the optimized original image of the rape flower if image quality optimization is performed, or directly extract features to obtain rape flower structural feature data if image quality optimization is performed; The rapeseed flower structural feature data quantification judgment module is used to obtain a structural feature quantification value by quantifying the structural feature corresponding to the rapeseed flower structural feature data, and judge whether to perform structural feature optimization based on the obtained structural feature quantification value and a preset structural feature threshold in the self-compatibility detection database. If structural feature optimization is performed, the self-compatibility of the rapeseed type is determined based on the structural feature optimization; otherwise, the self-compatibility is directly determined to obtain structural feature optimization. The structural feature optimization means that the reliability of rapeseed pollen grain identification is improved by processing the rapeseed flower structural feature extraction.

Images