A method for counting wheat scab spores based on contour angle ratio

Abstract

The present invention relates to a method for counting wheat fusarium spores based on contour angle ratio, which solves the defect that it is difficult to count sickle-shaped wheat fusarium spores compared with the prior art. The present invention comprises the following steps: acquisition of wheat fusarium spores; pre-processing of wheat fusarium spores; spore impurity removal; concave point search; contour angle ratio screening; counting of wheat fusarium spores. The present invention processes the spore image by using mean shift, Sobel operator edge detection, and shape feature factor screening to improve the quality of the fusarium spore image taken under a microscope, and then screens out the fusarium spores according to the shape feature factor, and finally segments the fusarium spores based on the contour angle ratio of the sickle-shaped fusarium spores, thereby completing the counting of wheat fusarium spores quickly and accurately.

Description

Technical Field

[0001] The invention relates to the technical field of spore segmentation, in particular to a method for counting wheat fusarium spores based on profile angle ratio. Background Art

[0002] Wheat fusarium head blight is a typical spore disease. The pathogen lives in the form of hyphae, conidia or ascocarps on wheat straw during the summer and winter. After maturing, it spreads directly through wind and rain to infect wheat. Generally, when the number of fusarium head blight spores exceeds a certain threshold, it will cause an outbreak of fusarium head blight. If the fusarium head blight spores on wheat can be counted quickly in real time, effective prevention and control decisions can be made according to the situation to achieve early prevention of wheat fusarium head blight.

[0003] At present, the methods for counting wheat fusarium spores are mainly divided into field estimation method and laboratory detection method. The former is that a dedicated person observes the state of wheat in the field, estimates the number of pathogenic spores based on experience, and then calculates the disease index. Although this method is simple, it has disadvantages such as strong subjectivity and easy misjudgment; the latter refers to the test personnel manually collecting wheat samples, counting the number of fusarium spores under a microscope, and realizing the graded diagnosis of fusarium spores. Although this method improves the accuracy of spore counting, because the spores are small, have many impurities, are large in number and are sticky under the microscope, this method requires a lot of time and energy. Therefore, there is an urgent need for an automatic, fast and accurate method for counting wheat fusarium spores to improve efficiency.

[0004] The key to counting wheat fusarium spores lies in the effective segmentation of adherent spores. Currently, there are few studies on spore counting of plant diseases at home and abroad, and most of them are aimed at spores of pathogens with regular shapes such as stripe rust and powdery mildew. There is basically no research on fusarium spores, which are narrow and have multiple septa in the body. The oval shape of fusarium spores makes it more difficult to segment them.

[0005] Lei Yu et al. counted 90 summer spore images of wheat stripe rust based on the improved watershed method and obtained an accuracy rate of 92.6%; Su et al. designed a spore segmentation and counting algorithm based on integral images based on microscopic images of late blight spores and obtained an accuracy rate of 99.2%, but the disadvantage is that the image adhesion is too simple and there are few impurities; Gao Xing et al. combined the five-point angle method with the chain code difference method to improve the accuracy of concave point positioning, thereby improving the segmentation accuracy of adhesion targets; Qi Long et al. proposed a spore segmentation and counting algorithm based on Gaussian filtering and distance variation. The improved watershed algorithm obtained an average detection rate of 98.5% in 100 test images of rice blast fungus spores, but the spore adhesion situation was too simple and not suitable for complex situations; Liang et al. used the improved U-net to segment and count wheat powdery mildew spores, achieving an accuracy of 91.4%, but the number of spores in the image was too small and not suitable for counting multiple spore adhesions; Wang et al. performed segmentation and counting of adherent anthrax spores based on a method combining watershed and topological maps. The results showed that the algorithm was better than the traditional watershed algorithm.

[0006] The spores of wheat fusarium wilt are sickle-shaped with multiple septa running through the middle, which makes them more difficult to divide and count than spores of other diseases. At the same time, because the fusarium wilt spores are only tens of microns, the spore samples obtained in the field are extremely sticky under the microscope, and the spore microscopic images usually contain dust, pollen, and spores of other diseases. These factors greatly increase the difficulty of counting.

[0007] In addition, to obtain good quality microscopic images of wheat scab, plant protection personnel are generally required to be equipped with expensive microscopes, otherwise the image quality is too low and will interfere with the counting. In particular, due to the many characteristics of scab spores, such as narrow and long body, multiple channels across the cell membrane, and random protrusions on the edge of the contour, these make it more difficult to count spores than wheat stripe rust, white powder and other diseases.

[0008] Therefore, how to count the adhesion spores of wheat fusarium spores has become a technical problem that needs to be solved urgently. Summary of the invention

[0009] The purpose of the present invention is to solve the defect that it is difficult to count sickle-shaped wheat fusarium spores in the prior art, and to provide a method for counting wheat fusarium spores based on profile angle ratio to solve the above problem.

[0010] In order to achieve the above object, the technical solution of the present invention is as follows:

[0011] A method for counting wheat scab spores based on contour angle ratio comprises the following steps:

[0012] Acquisition of wheat fusarium spore adhesion image: Acquisition of wheat fusarium spore adhesion image;

[0013] Preprocessing of wheat fusarium spore adhesion images: image mean shift, sobel operator edge processing, and morphological processing are performed on wheat fusarium spore adhesion images;

[0014] Spore impurity removal: The area, length, aspect ratio of the circumscribed rectangle, and ellipse fitting eccentricity characteristic factors are selected to remove impurities from the wheat fusarium spore adhesion image;

[0015] Concave point search process: the spore connector is subjected to concave point search process;

[0016] Contour angle ratio screening: Use the custom screening factor of contour angle ratio to find the true concave points of the adhesion spores and perform concave point matching and segmentation;

[0017] Counting of wheat fusarium spores: Counting of the spores that have completed the concave point matching segmentation to obtain the counting results of wheat fusarium spores.

[0018] The image mean shift comprises the following steps:

[0019] The calculation formula for setting the offset mean is as follows

[0020]

[0021] Where: S h is a high-dimensional sphere region with x as the center and radius h; i It belongs to S h The point in the range of S h The number of points in the range;

[0022] Get the image of wheat fusarium spores after pretreatment;

[0023] The initial iteration point is randomly selected, and the P0 point is randomly selected in the wheat scab adhesion spore image;

[0024] Color vector accumulation: taking P0 as the center point, according to the selected area M, the color vectors of all points from point P0 to area M are accumulated on the adhesion spore image to obtain a new iteration point P1;

[0025] Iterate until convergence, repeat step 24) with P1 as the center point until convergence to a fixed point Pk. At this time, pk is the place with the largest probability density in the adhesion spore image area M, which is called a clustering point;

[0026] Traverse all areas of the adhesion spore image: By iterating the above steps, the original value of the image is continuously updated to find the area with the densest sample distribution, that is, to complete the image mean shift filtering process.

[0027] The spore impurity removal process comprises the following steps:

[0028] Obtain the image of wheat fusarium spore adhesion after filtering;

[0029] Calculate the characteristic factors: For all the targets in the wheat fusarium spore adhesion image, use the function to calculate the values ​​of four characteristic factors: contour area S, length L, circumscribed rectangle aspect ratio AR, and ellipse fitting eccentricity e. The calculation formula is as follows:

[0030] F(α,X)=X·α=ax 2 +bxy+cy 2 +dx+ey+f=0 (1)

[0031]

[0032] Where F(α,X) is the least squares elliptic curve fitted by the spore contour, where α = [abcdef] T is the coefficient of the ellipse equation, X=[x 2 xy y 2 xy 1]x, y are the horizontal and vertical coordinates of the point on the curve; e is the distance D between the two foci of the ellipse c With the major axis length L a The ratio of

[0033] Determine the parameter threshold of the spore characteristic factor: randomly select 50 wheat scab spore images, perform statistical calculations on the above characteristic factors for all spores, and obtain the parameter threshold;

[0034] Here, let S = 50, L = 60, AR = 1.2, e = 0.9;

[0035] Impurity removal: By screening the parameter threshold, all targets that meet the parameter threshold are rewritten into the new image.

[0036] The pit finding process comprises the following steps:

[0037] Acquire images of wheat fusarium spore impurities removal;

[0038] Generate convex hull: Select any spore target in the adhesion spore image, and generate a minimum convex polygon that can contain the spore target according to its outermost contour point set, that is, the convex hull;

[0039] Find the concave area: calculate the relative offset between the convex hull and the target body contour to obtain all the concave areas;

[0040] Return key information of the concave area: select any concave area, traverse and compare the distances from all contour points in the concave area to the convex hull, and return a set of key information based on the results, including the coordinates of the concave starting point, the end point, the coordinates of the farthest point from the convex hull, and the distance from the farthest point to the convex hull. The expressions are as follows:

[0041]

[0042] Where u is the vector from the spore contour point to the starting point of the depression, v is the vector obtained by rotating the vector from the starting point of the depression to the end point by 90°, and u′ is the vector projected from u to v;

[0043] Locating concave points: Mark the concave points in the spore target body according to the coordinates of the farthest point from the convex hull in the key information of the concave area;

[0044] Traverse all spore targets in the adhesion image: traverse all spore targets to obtain all concave points of wheat fusarium adhesion spores.

[0045] The contour angle ratio screening comprises the following steps:

[0046] For the image after finding the concave point of wheat fusarium spore adhesion, the contour angle distance ratio is defined. The contour angle distance ratio is defined as the ratio of the distance from the concave point to the connecting line of the starting point and the end point of the concave in a concave area of ​​a spore target body to the angle of the concave;

[0047] Statistical analysis of the contour angle distance ratio of the adhesion spores: 50 images of adhesion spores of wheat fusarium spores were randomly selected, and the contour angle distance ratio of the adhesion spores was statistically analyzed, with the threshold value range being 0.7-3.3;

[0048] ADF=θ / D

[0049]

[0050]

[0051] Where: θ is the angle between the concave point and the left and right contour points, is the vector formed by the concave point and its left and right contour points; D is the shortest vertical distance from the concave point to the spore envelope, (x 0 ,y 0 ) is the concave point, A, B, C are the parameters of the straight line equation formed by the starting point and the end point of the concave;

[0052] Removing false concave points: Based on the statistically obtained threshold range of the angular distance ratio of the adhesion spore contour, the false concave points in the adhesion image are removed, that is, the true concave points that can correctly segment the adhesion spore target body are screened out;

[0053] Traverse all spore target bodies: traverse all spore target bodies to obtain all correctly segmented true concave points of wheat fusarium adhesion spores;

[0054] Segmentation by concave point matching strategy: According to the number of true concave points of each spore target, the concave point matching strategy is used for segmentation;

[0055] The concave point matching strategy is as follows: when there is only one adhesion concave point, find the center of gravity of the adhesion spore at this time, connect the concave point with the center of gravity, and then directly segment the spore; if there are two adhesion concave points, directly connect them; if there are three or more concave points, use the contour segment ellipse fitting method for segmentation.

[0056] Beneficial Effects

[0057] The method for counting wheat fusarium spores based on contour angle ratio of the present invention uses mean shift, Sobel operator edge detection and shape feature factor screening to process spore images compared with the prior art, thereby improving the quality of fusarium spore images taken under a microscope, and then screening out fusarium spores according to shape feature factors, and finally segmenting the fusarium spores based on the contour angle ratio of sickle-shaped fusarium spores, thereby quickly and accurately completing the counting of wheat fusarium spores.

[0058] The test results show that the present invention achieved an accuracy rate of 93.1% in 313 simulated outdoor wheat fusarium spore microscopic images, which is 7.9% higher than the traditional machine learning clustering segmentation counting algorithm, effectively improving the counting efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 is a method sequence diagram of the present invention;

[0060] Figure 2 It is a wheat scab spore diagram in the prior art;

[0061] Figure 3 The mean shift graph after the processing involved in the present invention;

[0062] Figure 4 This is a diagram of the removal of spore impurities after the treatment involved in the present invention;

[0063] Figure 5 Finding a schematic diagram for the concave points involved in the present invention;

[0064] Figure 6a A single concave point matching graph of the concave point matching strategy involved in the present invention;

[0065] Figure 6b It is a double concave point matching graph of the concave point matching strategy involved in the present invention;

[0066] Figure 6c It is a multi-concave point matching graph of the concave point matching strategy involved in the present invention;

[0067] Figure 7a It is the segmentation map of the traditional watershed algorithm;

[0068] Figure 7b Segmentation map to improve the watershed algorithm;

[0069] Figure 7c It is the segmentation graph of the traditional clustering algorithm;

[0070] Figure 7d This is a segmentation diagram using the method described in the present invention. DETAILED DESCRIPTION

[0071] In order to have a further understanding and recognition of the structural features and the effects achieved by the present invention, a preferred embodiment and accompanying drawings are used for detailed description as follows:

[0072] like Figure 1 As shown, the method for counting wheat fusarium spores based on profile angle ratio of the present invention comprises the following steps:

[0073] The first step is to obtain the image of the adhesion spores of wheat fusarium rust: obtain the image of the adhesion spores of wheat fusarium rust.

[0074] The main equipment used to take microscopic images of spores is Leica MC190 HD (Leica microscope digital camera). When taking images of wheat scab spores, Leica MC190 HD is used in combination with the software Leica application Suite (LAS) to facilitate visualization. The specific operation is: use a sterile dropper to absorb a small amount of spore suspension, drop it on the center of a clean slide, cover it with a clean and dry cover glass, place it under a microscope, select 10*40 times magnification, adjust the lens focus, find a clear group of wheat scab spores in the LAS software visualization interface, and finally take pictures.

[0075] In order to fit the actual situation in the field as much as possible, when shooting spore images, we tried to select spore groups with complex backgrounds, many impurities, and many adhesions. Finally, after selection, 313 spore images of fusarium spores were obtained, each with a size of 512×512 pixels. The microscopic images of wheat fusarium spore adhesions are shown in Figure 2. Figure 2 shown.

[0076] from Figure 2It can be seen that wheat fusarium spores are different from traditional stripe rust, white powder and other spores. They are long and narrow, and have multiple transverse cell membranes in their bodies. At the same time, because wheat fusarium spores have multiple growth states, their size is not stable, and there are often random protrusions on the edges of the spore outlines, which makes counting very difficult. In addition, wheat fusarium spores are sickle-shaped and have uncertain growth, which makes the problems of spore adhesion and overlap extremely complex and diverse. Compared with spores of regular shape, the difficulty of counting them increases.

[0077] The second step is the preprocessing of the wheat fusarium spore image: the wheat fusarium spore image is processed by using traditional techniques such as image mean shift, sobel operator edge processing, and morphological processing.

[0078] Image mean shift filtering: The wheat fusarium spore adhesion image is filtered using the mean shift algorithm. In the wheat fusarium spore image, since it simulates the actual field situation, the image has a lot of noise and impurities, so preprocessing is necessary. The mean shift algorithm is essentially a clustering algorithm. Its basic principle is to continuously update the original value by offsetting the mean, and find the area with the densest sample distribution to achieve the filtering effect. The specific steps of filtering are as follows:

[0079] (1) Set the offset mean calculation formula as follows

[0080]

[0081] Where: S h is a high-dimensional sphere region with x as the center and radius h; i It belongs to S h The point in the range of S h The number of points in the range.

[0082] (2) Obtain images of wheat fusarium spores after pretreatment.

[0083] (3) Randomly select the initial iteration point and randomly select the P0 point in the wheat fusarium spore adhesion image.

[0084] (4) Color vector accumulation: Taking P0 as the center point, according to the selected area M, the color vectors of all points from point P0 to the M area class are accumulated on the adhesion spore image to obtain a new iteration point P1.

[0085] (5) Continue to iterate until convergence, repeat step (4) with P1 as the center point until convergence to a fixed point Pk. At this time, pk is the place with the largest probability density in the adhesion spore image area M, which is called the clustering point.

[0086] (6) Traversing all areas of the adhesion spore image: By iterating the above steps, the original value of the image is continuously updated to find the area with the densest sample distribution, thus completing the filtering operation.

[0087] The image is mean shifted, and the processing results are as follows Figure 3 As shown. Figure 3 It can be seen that the mean shift algorithm can filter the image at the color level and neutralize the colors with similar color distribution in the image. In other words, the mean shift can eliminate noise and impurities in low-quality spore images to achieve the effect of improving data quality.

[0088] The third step is to remove spore impurities: the impurities of the wheat fusarium spore image are removed by selecting characteristic factors such as area, length, aspect ratio of the circumscribed rectangle, and eccentricity of the ellipse fitting.

[0089] Considering the sickle-like shape of wheat fusarium spores and their distinguishability from impurities, shape feature factors such as area S, length L, aspect ratio AR of the circumscribed rectangle, and eccentricity e of the ellipse fitting were used to remove factors that interfere with spore counting, such as hyphae and cell impurities in the microscopic image. 50 wheat fusarium spore images were randomly selected, and the above features were statistically analyzed. Finally, the parameter thresholds were determined as S = 50 Pixel, L = 60 Pixel, AR = 1.2, and e = 0.9. The processing results are shown in the figure. Figure 4 As shown in the figure, it can be seen that after the above processing, the impurities and holes in the spore image have been removed, and the spore edge has been smoothed.

[0090] The specific steps for removing spore impurities are as follows:

[0091] (1) Obtain the image of wheat fusarium spores after filtering.

[0092] (2) Calculation of characteristic factors: For all target objects in the image of wheat fusarium spore adhesion, four characteristic factor values ​​are calculated using a function: contour area S, length L, aspect ratio of the circumscribed rectangle AR, and ellipse fitting eccentricity e. The calculation formula is as follows:

[0093] F(α,X)=X·α=ax 2 +bxy+cy 2 +dx+ey+f=0 (1)

[0094]

[0095] Where F(α,X) is the least squares elliptic curve fitted by the spore contour, where α = [abcdef] T is the coefficient of the ellipse equation, X=[x 2 xy y 2xy 1]x, y are the horizontal and vertical coordinates of the point on the curve; e is the distance D between the two foci of the ellipse c With the major axis length L a ratio.

[0096] (3) Determine the parameter threshold of the spore characteristic factor: randomly select 50 wheat fusarium spore images, perform statistical calculations on the above characteristic factors for all spores, and obtain the parameter threshold;

[0097] Here, it is assumed that S=50, L=60, AR=1.2, and e=0.9.

[0098] (4) Impurity removal: By screening the parameter threshold, all targets that meet the parameter threshold are rewritten into the new image.

[0099] Step 4: Find concave points: Find concave points on the spore connector. When multiple spores stick together, their contours will appear concave, and the contour curvature will change significantly. At this time, the point with the maximum local curvature of the contour is defined as a concave point. First, select any spore target in the image, and generate a minimum convex polygon that can contain all point sets based on its outermost contour point set; then, based on the convex polygon generated in the previous step, all concave areas of the sticky spores can be obtained, and the distances from all contour points to the convex hull can be compared to obtain the concave points. The detailed process is as follows: Figure 5 As shown in the figure, assuming that there are three spore contour points C[0], C[1], and C[2] in the concave area, the vector Rotate 90 degrees, and then project the vectors formed by connecting these three points with H[0] onto its unit vector, and you can get the distances D1, D2, and D3 from the three points to the convex hull (orange line). Finally, compare the sizes of these three distances to get the location of the concave points. Finally, traverse all spores in the image and repeat the above two steps to get the index coordinates of all depressions and all concave points in the image.

[0100] The specific steps of concave point finding processing are as follows:

[0101] (1) Obtain an image of the removal of wheat fusarium spore impurities.

[0102] (2) Convex hull generation: Select any spore target in the adhesion spore image and generate a minimum convex polygon that can contain the spore target based on its outermost contour point set, i.e., the convex hull.

[0103] (3) Find the concave area: Calculate the relative offset between the convex hull and the target contour to obtain all the concave areas.

[0104] (4) Return key information of the concave area: select a concave area, traverse and compare the distances from all contour points of the concave area to the convex hull, and return a set of key information based on the results, including the coordinates of the concave starting point, the end point, the coordinates of the farthest point from the convex hull, and the distance from the farthest point to the convex hull;

[0105]

[0106] Where u is the vector from the spore contour point to the starting point of the depression, v is the vector from the starting point of the depression to the end point rotated 90°, and u′ is the vector projected from u to v.

[0107] (5) Locating concave points: Mark the concave points in the spore target body according to the coordinates of the farthest point from the convex hull in the key information of the concave area.

[0108] (6) Traversing all spore targets in the adhesion image: By traversing all spore targets, all the concave points of wheat fusarium adhesion spores can be obtained.

[0109] The fifth step is contour angle distance ratio screening: Use the contour angle distance ratio custom screening factor to find the true concave points of the adhesion spores and perform concave point matching segmentation. Since wheat scab spores have through-membranes and the contour edge has irregular growth, adhesion spores may have false concave points. Therefore, contour angle distance ratio ADF is proposed as a screening factor. 50 wheat scab spore images are randomly selected, and the screening factor ADF is statistically analyzed to determine the key and segmentable adhesion concave points of wheat scab adhesion spores, which range from 0.7 to 3.3.

[0110] The specific steps of contour angle ratio screening are as follows:

[0111] (1) For the image after finding the concave point of wheat fusarium spore adhesion, the contour angle distance ratio is defined. The contour angle distance ratio is defined as the ratio of the distance from the concave point to the line connecting the starting point and the end point of the concave area of ​​a spore target body to the angle of the concave.

[0112] (2) Counting the contour angle distance ratio of the adhesion spores: 50 images of wheat fusarium spore adhesion were randomly selected, and the contour angle distance ratio of the adhesion spores was counted, with the threshold range set to 0.7-3.3;

[0113] ADF=θ / D

[0114]

[0115]

[0116] Where: θ is the angle between the concave point and the left and right contour points, is the vector formed by the concave point and its left and right contour points; D is the shortest vertical distance from the concave point to the spore envelope, (x0 ,y 0 ) is the concave point, and A, B, and C are the parameters of the straight line equation formed by the starting point and the end point of the concave.

[0117] (3) Removing false concave points: Based on the statistically obtained threshold range of the angular distance ratio of the adhesion spore contour, the false concave points in the adhesion image are removed, that is, the true concave points that can correctly segment the adhesion spore target are screened out.

[0118] (4) Traversing all spore targets: Traversing all spore targets, all true concave points of wheat fusarium adhesion spores that can be correctly segmented are obtained.

[0119] (5) Segmentation by concave point matching strategy: According to the number of true concave points of each spore target, different strategies are selected for segmentation;

[0120] The concave point matching strategy is as follows: when there is only one adhesion concave point, we only need to find the center of gravity of the adhesion spore at this time, connect the concave point with the center of gravity, and then we can directly segment the spore; if there are two adhesion concave points, we can directly connect them; if there are three or more concave points, we can use the contour segment ellipse fitting method to segment.

[0121] like Figure 6a , Figure 6b , Figure 6c As shown in the figure, when there is only one adhesion concave point, we only need to find the center of gravity of the adhesion spore at this time, connect the concave point with the center of gravity, and then we can directly segment the spore; if there are two adhesion concave points, we can directly connect them; if there are three or more concave points, we can use the contour segment ellipse fitting method to segment.

[0122] Step 6, counting of wheat fusarium spores: counting of the spores that have completed the concave point matching segmentation to obtain the counting result of wheat fusarium spores.

[0123] In order to verify the accuracy of the wheat fusarium adhesion spore counting algorithm of the present invention, 313 wheat fusarium spore images simulating outdoor field conditions were tested using PyCharm Community Edition 2020.

[0124] Through careful and accurate manual counting, the counting result is taken as the actual number of spores. In addition, the number of spores in the kth spore image manually counted is recorded as M. k The total number of spore divisions obtained by the algorithm is recorded as T k , the number of correct segmentations is recorded as N k , accuracy is defined as follows:

[0125]

[0126] Average accuracy is defined as:

[0127]

[0128] At the same time, in order to verify the effectiveness of the counting algorithm proposed in the present invention, the algorithm of the present invention is compared with the traditional watershed algorithm, the improved watershed algorithm and the clustering segmentation algorithm in traditional machine learning. The counting statistical results obtained by the above four algorithms are shown in Table 1.

[0129] Table 1 Spore counting results

[0130] Table 1 Results of species counting

[0131]

[0132]

[0133] As shown in Table 1, in 313 wheat fusarium spore images, the total number of spores counted manually is 3698, the total number of segmentations of the algorithm proposed in the present invention is 3798, the number of correct segmentations is 3441, and the average accuracy is about 93.1%. Compared with the traditional watershed and improved watershed, as well as the counting accuracy of the clustering segmentation algorithm, it has increased by at least ten percentage points, which can improve the counting efficiency of wheat fusarium spore microscopic images.

[0134] In addition, specific spore count images such as Figure 7d As shown, from the comparison Figure 7a , Figure 7b and Figure 7c It can be found that the algorithm proposed in the present invention pays more attention to the tiny details in the microscopic image than other algorithms, and some small spores are basically found. In addition, the algorithm has achieved a certain balance between under-segmentation and over-segmentation of the adhesion spores of Fusarium fusarium.

[0135] Figure 7a , Figure 7b and Figure 7cAs a comparative algorithm, it is better than the algorithm proposed in the present invention for specific adhesion spores, but it has a relatively large defect for the adhesion spores as a whole. For example, the watershed algorithm has a very serious problem in "under-segmentation". Some adhesion algorithms can find and segment, but most adhesion spore algorithms cannot handle it correctly, and even single spores cannot be found from the graph. Therefore, the number of correct spore segmentations is very low; there is also a traditional clustering segmentation algorithm, which has a better effect on the segmentation of some spores, but because the spores of Fusarium spores have the characteristic of crossing the diaphragm, its defect lies in "over-segmentation". In other words, a spore algorithm may directly segment two spores; three adhesion spores may directly segment 6 spores, so the traditional segmentation algorithm will also have very few correct spore segmentations. Compared with the methods of the prior art, the algorithm of the present invention achieves a balance between under-segmentation and over-segmentation, and can meet the basic counting requirements.

[0136] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions only describe the principles of the present invention. The present invention may be subject to various changes and improvements without departing from the spirit and scope of the present invention. These changes and improvements fall within the scope of the present invention. The scope of protection claimed by the present invention is defined by the attached claims and their equivalents.

Claims

1. A method for counting wheat scab spores based on contour angle ratio. It is characterized in that The following steps are involved: 11) Acquisition of an image of wheat fusarium spores: Acquiring an image of wheat fusarium spores; 12) Preprocessing of wheat fusarium spore adhesion images: performing image mean shift, Sobel operator edge processing, and morphological processing on wheat fusarium spore adhesion images; 13) Spore impurity removal: The area, length, aspect ratio of the circumscribed rectangle, and ellipse fitting eccentricity characteristic factors are selected to remove impurities from the wheat fusarium spore adhesion image; 14) Concave point finding process: performing concave point finding process on the spore connector; 15) Contour angle ratio screening: Use the contour angle ratio customized screening factor to find the true concave points of the adhesion spores and perform concave point matching and segmentation; The contour angle ratio screening comprises the following steps: 151) For the image after finding the concave point of wheat fusarium spore adhesion, define the contour angle distance ratio, which is defined as the ratio of the distance from the concave point to the connecting line of the starting point and the end point of the concave in a concave area of ​​a spore target body to the concave angle; 152) Counting the contour angle distance ratio of the adhesion spores: arbitrarily select 50 images of wheat scab adhesion spores, and count the contour angle distance ratio of the adhesion spores, setting the threshold range to 0.7-3.3; ADF=θ / D Where: θ is the angle between the concave point and the left and right contour points, is the vector formed by the concave point and its left and right contour points, D is the shortest vertical distance from the concave point to the spore envelope, (x 0 ,y 0 ) is the concave point, A, B, C are the parameters of the straight line equation formed by the starting point and the end point of the concave; 153) Removing false concave points: based on the statistically obtained threshold range of the angular distance ratio of the adhesion spore contour, removing the false concave points in the adhesion image, that is, screening out the true concave points that can correctly segment the adhesion spore target body; 154) Traversing all spore target bodies: traversing all spore target bodies to obtain all correctly segmented true concave points of wheat fusarium adhesion spores; 155) Segmentation by concave point matching strategy: according to the number of true concave points of each spore target body, the concave point matching strategy is selected for segmentation; The concave point matching strategy is as follows: when there is only one concave point, find the center of gravity of the concave point and connect it to the center of gravity, and then directly segment the spores; if there are two concave points, connect them directly; if there are three or more concave points, use the contour segment ellipse fitting method to segment. 16) Counting of wheat fusarium spores: Counting of the spores that have completed the concave point matching segmentation to obtain the counting result of wheat fusarium spores.

2. A method for counting wheat scab spores based on profile angle ratio according to claim 1, It is characterized in that The image mean shift comprises the following steps: 21) Set the offset mean calculation formula as follows Where: S h is a high-dimensional sphere region with x as the center and radius h; i It belongs to S h The point in the range of S h The number of points in the range; 22) Obtaining images of wheat fusarium spores after pretreatment; 23) randomly selecting an initial iteration point, and randomly selecting a P0 point in the wheat scab adhesion spore image; 24) Accumulating color vectors: taking P0 as the center point, according to the selected area M, accumulating the color vectors of all points from point P0 to the M area class on the adhesion spore image to obtain a new iteration point P1; 25) Iterate until convergence, repeat step 24) with P1 as the center point until convergence to a fixed point Pk, at which point Pk is the place with the largest probability density in the adhesion spore image region M, called a clustering point; 26) Traversing all regions of the adhesion spore image: By iterating the above steps, the original value of the image is continuously updated to find the area with the densest sample distribution, that is, completing the image mean shift filtering process.

3. A method for counting wheat scab spores based on profile angle ratio according to claim 1, It is characterized in that The spore impurity removal process comprises the following steps: 31) obtaining a filtered image of wheat fusarium spores; 32) Calculate characteristic factors: For all target bodies in the wheat scab adhesion spore image, use the function to calculate the values ​​of four characteristic factors: contour area S, length L, circumscribed rectangle aspect ratio AR, and ellipse fitting eccentricity e. The calculation formula is as follows: F(α,X)=X·α=ax 2 +bxy+cy 2 +dx+ey+f=0 (1) Where F(α,X) is the least squares elliptic curve fitted by the spore contour, where α = [abcdef] T is the coefficient of the ellipse equation, X=[x 2 xy y 2 xy 1], x, y are the horizontal and vertical coordinates of the point on the curve, e is the distance D between the two foci of the ellipse c With the major axis length L a The ratio of 33) Determine the parameter threshold of the spore characteristic factor: randomly select 50 wheat scab spore images, perform statistical calculation of the above characteristic factors on all spores, and obtain the parameter threshold; Here, let S = 50, L = 60, AR = 1.2, e = 0.9; 34) Impurity removal: By screening the parameter threshold, all targets that meet the parameter threshold are rewritten into the new image.

4. A method for counting wheat scab spores based on profile angle ratio according to claim 1, It is characterized in that The pit finding process comprises the following steps: 41) Obtaining an image of wheat fusarium spore impurities removal; 42) Generate convex hull: Select any spore target in the adhesion spore image, and generate a minimum convex polygon that can contain the spore target according to its outermost contour point set, i.e., convex hull; 43) Find the concave area: calculate the relative offset between the convex hull and the target body contour to obtain all the concave areas; 44) Return key information of the concave area: select any concave area, traverse and compare the distances from all contour points of the concave area to the convex hull, and return a set of key information based on the results, including the coordinates of the concave starting point, the end point, the coordinates of the farthest point from the convex hull, and the distance from the farthest point to the convex hull. The expressions are as follows: Where u is the vector from the spore contour point to the starting point of the depression, v is the vector obtained by rotating the vector from the starting point of the depression to the end point by 90°, and u′ is the vector projected from u to v; 45) Locating concave points: marking the concave points in the spore target body according to the coordinates of the farthest point from the convex hull in the key information of the concave area; 46) Traversing all spore targets in the adhesion image: traversing all spore targets to obtain all concave points of wheat fusarium adhesion spores.

Images