Method and system for detecting seeding quality of automatic seeding machine based on image recognition

Abstract

The embodiment of the invention provides an automatic seeder seeding quality detection method based on image recognition, and belongs to the technical field of agricultural seeding. The detection method comprises the steps of obtaining a seed quality image; preprocessing the quality image; performing edge detection on the quality image by using a Canny operator to obtain an image of the contour of the seed; performing morphological processing on the image of the contour of the seed to remove noisy points; obtaining candidate resowing seed holes and resowing seed holes according to the distance between the center points of the contours of the adjacent seeds; according to the perimeter and the angular point of the contour of the seed, further obtaining a reseeding seed hole and a qualified seeding hole in the candidate reseeding seed holes; calculating the number of resown seed holes and the number of qualified seed holes according to the resown seed holes and the qualified seed holes; counting the number of missed seeding seed holes according to the number of reseeding seed holes and the number of qualified seeding holes, and calculating the qualified rate of seeding. According to the invention, by analyzing the image data of sowing, the detection of overlapped seeds, resown seeds and miss-sowing is realized at the same time.

Description

Technical Field

[0001] The present invention relates to the technical field of agricultural sowing, and in particular to a sowing quality detection method and system of an automatic seeder based on image recognition. Background Art

[0002] In recent years, my country's vegetable seedling technology has made significant progress. At present, most of the leafy vegetable seedlings are grown in seedbed trays. Precision seeding can achieve one-time seedling formation. The traditional seedling method is labor-intensive, occupies a lot of manpower, and has low efficiency. The reform of traditional seedling cultivation by factory seedling cultivation is an inevitable choice to improve agricultural development. Factory seedling cultivation requires one seed per hole to ensure the quality of seedlings and the uniformity of seedlings. In order to speed up the sowing speed and reduce people's labor intensity, the automatic precision sowing system came into being. The quality of the seeder directly affects the production efficiency of the seedlings, so real-time monitoring of the sowing quality of the automatic seeder is of great significance.

[0003] At present, there is no specific detection method for the sowing quality of hydroponic sponge seedlings. The automatic seeder works in an assembly line mode, and the sowing efficiency can reach one tray in a few seconds. The efficiency of traditional manual observation of sowing quality is difficult to match, so traditional manual observation cannot quickly evaluate the sowing quality. Since there are many situations in tray sowing, such as multiple seeds in one hole, multiple seeds overlapping, and multiple seeds independent, the general contour detection algorithm can only identify a single seed, so it cannot be directly used to detect the sowing pass rate of the automatic seeder. Summary of the invention

[0004] The purpose of the embodiments of the present invention is to provide a method and system for detecting the sowing quality of an automatic seeder based on image recognition, which solves the real-time monitoring and identification of the over-seeding and missed-seeding problems of the seeder.

[0005] In order to achieve the above object, an embodiment of the present invention provides a method for detecting sowing quality of an automatic seeder based on image recognition, the detection method comprising:

[0006] Obtain seed quality images;

[0007] Preprocessing the quality image;

[0008] Using the Canny operator to perform edge detection on the quality image to obtain an image of the outline of the seed;

[0009] performing morphological processing on the image of the outline of the seed to remove noise;

[0010] Acquire candidate reseeding seed holes and reseeding seed holes according to the distances between the center points of the contours of adjacent seeds;

[0011] Further obtaining reseeding seed holes and qualified seed holes in the candidate reseeding seed holes according to the perimeter and corner points of the outline of the seed;

[0012] Calculate the number of reseeding seed holes and the number of qualified seed holes according to the reseeding seed holes and qualified seed holes;

[0013] The number of missed seeding holes is counted according to the number of reseeded seeding holes and the number of qualified seeding holes, and the seeding qualified rate is calculated.

[0014] Optionally, obtain a seed quality image, including:

[0015] The hole tray is transferred to the position directly below the camera and the camera is fixed to ensure that the number of holes of the sponge and the coordinate position of each hole in the seed quality image obtained each time remain unchanged;

[0016] Setting the shooting interval of the camera according to the time required for sowing to determine the horizontal fixed position of the camera;

[0017] Obtain seed quality images at the sowing detection end through a camera;

[0018] The seed quality image is transmitted to a detector in real time for detection.

[0019] Optionally, preprocessing the quality image includes:

[0020] Cropping the quality image according to the rectangular area;

[0021] Performing grayscale processing on the quality image;

[0022] Gaussian filtering is used to reduce noise on the quality image.

[0023] Optionally, the image of the outline of the seed is subjected to morphological processing to remove noise, comprising:

[0024] Performing erosion and dilation operations on the image of the outline of the seed;

[0025] Filter the image of the outline of the seed with a smaller area.

[0026] Optionally, obtaining the number of candidate reseeding seed holes of a hole according to the spacing between the center points of the contours of adjacent seeds includes:

[0027] Get the geometric moment of the seed's outline;

[0028] The coordinates of the geometric centroid are obtained according to the geometric moment and stored.

[0029] Optionally, obtaining the candidate reseeding seed holes and the reseeding seed holes according to the distances between the center points of the contours of adjacent seeds comprises:

[0030] Calculating the distance between the center points of the contours of adjacent seeds;

[0031] Determining whether the distance is less than a first threshold;

[0032] When it is determined that the distance is less than the first threshold, determining that the hole where the seed is located is a reseeding hole;

[0033] When it is determined that the distance is not less than the first threshold, the hole where the seed is located is determined to be a candidate reseeding hole and the sowing state of the hole is further determined.

[0034] Optionally, further obtaining reseeding seed holes and qualified seeding holes in the candidate reseeding seed holes according to the perimeter and corner points of the outline of the seed comprises:

[0035] Calculating the perimeter of the outline of the seed for a single hole;

[0036] Determining whether the perimeter of the contour is greater than a second threshold;

[0037] When it is determined that the perimeter of the contour is greater than the second threshold, determining that the hole where the seed is located is a reseeding hole;

[0038] In the case where it is determined that the perimeter of the contour is not greater than the second threshold, determining whether the corner point threshold of the contour is greater than a third threshold;

[0039] When it is determined that the corner point threshold of the contour is greater than a third threshold, determining that the hole where the seed is located is a reseeding hole;

[0040] When it is determined that the corner point threshold of the seed is not greater than the third threshold, the hole where the seed is located is determined to be a qualified sowing hole.

[0041] Optionally, calculating the number of reseeding seed holes and the number of qualified seed holes according to the reseeding seed holes and qualified seed holes includes:

[0042] In the case where the hole is determined to be a reseeding hole, the number of the reseeding hole is increased by 1;

[0043] When it is determined that the hole is a qualified sowing hole, the number of the qualified sowing holes is increased by 1.

[0044] Optionally, counting the number of missed seeding holes according to the number of reseeded seeding holes and the number of qualified seeding holes, and calculating the seeding qualified rate, comprises:

[0045] Obtain the number of holes in the colonized sponge;

[0046] The number of holes minus the number of reseeding holes and the number of qualified seeding holes is used to obtain the number of missed seeding holes;

[0047] The sowing qualified rate is calculated according to the number of re-sowing seed holes, the number of qualified sowing holes and the number of missed sowing seed holes.

[0048] On the other hand, the present invention provides an automatic seeder sowing quality detection system based on image recognition, the system comprising a processor for executing any of the detection methods described above.

[0049] Through the above technical scheme, the present invention provides a method and system for detecting the sowing quality of an automatic seeder based on image recognition. First, the acquired seed quality image is analyzed, and the re-seeding seed holes, that is, the situation where there are two non-overlapping seeds in one hole, are preliminarily screened out through the spacing between the center points of the seed contours. Then, the remaining candidate re-seeding seed holes are screened out for the overlapping seed holes and qualified seed holes, and the situation where the single seed is misjudged as an overlapping seed due to its large size is eliminated by setting the perimeter and corner point thresholds. Compared with the prior art, the present invention can directly detect and distinguish the sowing states of multiple seeds in one hole, overlapping multiple seeds, and independent multiple seeds when the seeder sows, and finally calculate the sowing qualification rate of the seeder sowing, providing basic data for improving the uniformity and yield of crops.

[0050] Other features and advantages of the embodiments of the present invention will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the following specific implementations, they are used to explain the embodiments of the present invention, but do not constitute a limitation on the embodiments of the present invention. In the accompanying drawings:

[0052] Figure 1 is a flow chart of a detection method according to one embodiment of the invention;

[0053] Figure 2 is a flow chart of obtaining a seed quality image according to one embodiment of the invention;

[0054] Figure 3 is a front view of a baffle mechanism fixed on a seed drill frame according to an embodiment of the invention;

[0055] Figure 4 is a top view of a baffle mechanism fixed on a seed drill frame according to one embodiment of the invention;

[0056] Figure 5 It is a right view of a baffle mechanism fixed on a seed drill frame according to an embodiment of the invention

[0057] Figure 6 is a flow chart of obtaining a seed quality image according to one embodiment of the invention;

[0058] Figure 7 is a flow chart of morphological processing according to one embodiment of the invention;

[0059] Figure 8 is a flow chart of obtaining the center of a contour of a seed according to one embodiment of the invention;

[0060] Fig. 9 is a seed quality image captured within a selected rectangular area according to an embodiment of the invention;

[0061] Fig.10 is an image of an uneroded, dilated filtered contour according to one embodiment of the invention;

[0062] Fig.11 is an image of the contour after erosion and dilation filtering according to one embodiment of the invention;

[0063] Fig.12 is a flow chart of obtaining candidate reseeding seed holes and reseeding seed holes according to one embodiment of the invention;

[0064] Fig.13 is a flow chart of distinguishing overlapping seed holes and qualified seed holes in candidate reseeding seeds according to one embodiment of the invention;

[0065] Fig.14 According to one embodiment of the invention, the candidate reseeding seed holes and the image of the reseeding seed holes are obtained according to the center point of the contour;

[0066] Fig.15 is an image for distinguishing overlapping seed holes and qualified seed holes in candidate reseeding seeds according to the perimeter of the contour according to one embodiment of the invention;

[0067] Fig.16 This is an image finally screened by a sowing quality detection method according to an embodiment of the invention.

[0068] Fig.17 is a flow chart of hole count statistics according to one embodiment of the invention;

[0069] Fig.18 The present invention is a flowchart for calculating the seeding qualification rate according to one embodiment of the present invention. DETAILED DESCRIPTION

[0070] The specific implementation of the embodiment of the present invention is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described here is only used to illustrate and explain the embodiment of the present invention, and is not used to limit the embodiment of the present invention.

[0071] It should be noted that the acquisition, transmission, storage, use, and processing of data in the technical solution of this application are in compliance with the relevant provisions of national laws and regulations. In the embodiments of this application, some existing solutions in the industry such as certain software, components, and models may be mentioned, which should be considered as exemplary. Their purpose is only to illustrate the feasibility of implementing the technical solution of this application, but it does not mean that the applicant has or will necessarily use the solution.

[0072] Figure 1 is a flow chart of a detection method according to one embodiment of the invention. In the figure, the sowing method comprises:

[0073] In step S1, a seed quality image is acquired.

[0074] In step S2, the quality image is preprocessed.

[0075] In step S3, the Canny operator is used to perform edge detection on the quality image to obtain an image of the outline of the seed.

[0076] In step S4, morphological processing is performed on the image of the seed outline to remove noise.

[0077] In step S5, candidate reseeding seed holes and reseeding seed holes are obtained according to the distances between the center points of the contours of adjacent seeds.

[0078] In step S6, reseeding seed holes and qualified seed holes in the candidate reseeding seed holes are further obtained according to the perimeter and corner points of the outline of the seed.

[0079] In step S7, the number of reseeding seed holes and the number of qualified seed holes are calculated according to the reseeding seed holes and the qualified seed holes.

[0080] In step S8, the number of missed seeding holes is counted according to the number of reseeded seeding holes and the number of qualified seeding holes, and the seeding qualified rate is calculated.

[0081] The present invention first analyzes the acquired seed quality image, and preliminarily screens out the reseeded seed holes, i.e., the situation where there are two non-overlapping seeds in one hole, through the spacing between the center points of the seed contour, and then performs secondary screening on the remaining candidate reseeded seed holes to screen out overlapping seed holes and qualified seed holes, and excludes the situation where a single seed is misjudged as an overlapping seed due to its large size by setting the perimeter and corner point thresholds. Compared with the prior art, the present invention collects images of the sowing situation in a baffle shooting environment, combines computer vision and image processing technology, obtains the image of the seed contour through the Canny operator provided by OpenCV, and then analyzes the image of the seed contour, which can directly detect and distinguish the sowing states of multiple seeds in one hole, overlapping multiple seeds, and independent multiple seeds when the sowing machine is sowing, and finally calculates the sowing qualification rate of the sowing machine. The method is automated, efficient, and accurate, and provides basic data for improving the uniformity and yield of crops.

[0082] In this embodiment, the seed quality image can be obtained by various methods known to those skilled in the art. In one example of the present invention, the seed quality image is obtained by a camera. Specifically, Figure 2 As shown, it may include:

[0083] In step S11, the hole tray is transferred to the bottom of the camera and the position of the camera is fixed to ensure that the number of holes of the planting sponge and the coordinate position of each hole in the seed quality image obtained each time remain unchanged. The camera is fixed by a bracket, and the fixed height is adjustable to ensure that the entire seedling tray can be photographed to ensure that the shooting angle and height are fixed, and upper and lower hollow baffles are arranged around the camera, and the baffles are fixed on the bracket, which shall not interfere with the transmission of the seedling tray below on the conveyor belt, nor the arrangement of the camera above.

[0084] In step S12, the camera shooting interval is set according to the time required for sowing to determine the horizontal fixed position of the camera. The seeder completes the lowering and sowing actions on the front conveyor belt. The shooting time interval is set according to the time required for sowing. There is the same position interval between each seedling tray, and the horizontal fixed position of the camera is determined according to this interval.

[0085] In step S13, a seed quality image of the sowing detection end is obtained through a camera.

[0086] In step S14, the seed quality image is transmitted to the detector in real time for detection.

[0087] The present invention is applied to an automatic precision seeder for hydroponic sponge seedlings. The background of the identification object is a black sponge. Reflection will occur during shooting. A large amount of reflective noise will seriously affect the identification effect. Therefore, an adjustable baffle is arranged to avoid direct strong light and make the fixed camera adjustable. When adjusting the position of the camera, first move the four aluminum profile columns to a suitable position, fix the baffle on the four aluminum profile columns, and adjust the horizontal adjustment hole according to the stop position of the seedling tray to find a suitable detection position. Then adjust the height at which the baffle is fixed, so as to be able to completely shoot the entire sponge. It should be noted that the lower end of the baffle cannot interfere with the transmission of the seedling tray on the conveyor belt. Finally, adjust the longitudinal adjustment hole of the camera so that the shooting center is as close to the center line of the width direction of the seedling tray as possible. Among them, the baffle on the seeder frame can be as follows Figure 3 , 4 , as shown in Figure 5.

[0088] In this embodiment, in order to ensure that the subsequent image analysis and recognition are more accurate and reliable, the acquired quality image needs to be preprocessed. The specific steps of the preprocessing can be various steps known to those skilled in the art. In one example of the present invention, specifically, Figure 6 As shown, including:

[0089] In step S21, the quality image is cropped according to the rectangular area. A rectangular area is selected from the sowing quality image captured by the camera, so that the area contains the entire black sponge as much as possible and contains as little complex background as possible.

[0090] In step S22, grayscale processing is performed on the quality image.

[0091] In step S23, Gaussian filtering is used to reduce noise in the quality image. When shooting an image, the black sponge inevitably reflects light and generates noise, so Gaussian filtering is needed to perform preliminary noise reduction on the grayscale image.

[0092] In this embodiment, considering the need to further eliminate noise to improve image quality, after using the canny operator to extract features from the seed outline, the image of the seed outline can be morphologically processed. Specifically, Figure 7 As shown, it may include:

[0093] In step S41, the image of the seed outline is eroded and expanded. Since the sponge reflective points generally exist in the form of points, the erosion and expansion operations can be used to eliminate noise points and fill the outline holes.

[0094] In step S42, the image of the outline of the seed having a smaller area is filtered.

[0095] In this embodiment, the method for obtaining the center of the contour of the seed can be various methods known to those skilled in the art. In one example of the present invention, the center of the contour of the seed is obtained by using the cv2.findContours() function in OpenCV. Specifically, Figure 8 As shown, including:

[0096] In step S51, the geometric moments of the seed contour are obtained.

[0097] In step S52, the coordinates of the geometric centroid are obtained according to the geometric moment and stored.

[0098] The camera is fixed at a height of 30 cm from the sponge surface to shoot, and the image is cropped according to the rectangular area, such as Fig. 9 As shown in the figure, there may be no seeds, one seed, or multiple seeds in the sponge hole. According to the shooting environment, adjust and set the appropriate threshold. Fig.10 As shown in the figure, the contour detection image without corrosion, expansion and filtering has noise contours. Fig.11 As shown in FIG. 1 , after the erosion operation with the core number of 1 and the dilation operation with the core number of 5, the contours with an area less than the minimum threshold of 130 are filtered out. In order to preliminarily screen out the reseeded seed holes, that is, the case where there are two non-overlapping seeds in one hole, in this embodiment, the candidate reseeded seed holes and the reseeded seed holes are obtained by calculating the distance between the center points of the contours of the seeds. Specifically, Fig.12 As shown, including:

[0099] In step S53, the distances between the center points of the contours of adjacent seeds are calculated.

[0100] In step S54, it is determined whether the distance is less than a first threshold value, and if it is determined that the distance is less than the first threshold value, step S55 is executed, otherwise, step S56 is executed. The first threshold value is 120.

[0101] In step S55, the hole where the seed is located is determined to be a reseeding hole.

[0102] In step S56, the hole where the seed is located is determined to be a candidate hole for reseeding the seed, and the sowing state of the hole is further determined.

[0103] In this embodiment, since the reseeding seed holes selected by the preliminary screening are the cases where there are two non-overlapping seeds in one hole, there will be cases where multiple seeds overlap in one hole in the remaining candidate reseeding seed holes, which are also reseeding seed holes. Therefore, it is necessary to further judge the candidate reseeding seed holes to distinguish the overlapping seed holes and qualified seed holes in the candidate reseeding seeds. Specifically, Fig.13 As shown, including:

[0104] In step S61 , the perimeter of the outline of the seed of a single hole is calculated.

[0105] In step S62, it is determined whether the perimeter of the contour is greater than a second threshold value, and if it is determined that the perimeter of the contour is greater than the second threshold value, step S63 is executed, otherwise, step S64 is executed. The second threshold value is 70.

[0106] In step S63, the hole where the seed is located is determined to be a reseeding hole.

[0107] In step S64, it is determined whether the corner threshold of the contour is greater than the third threshold. If it is determined that the corner threshold of the contour is greater than the third threshold, step S65 is executed, otherwise, step S66 is executed. The third threshold is 8. Since there may be a single large seed, the perimeter of the single seed is longer and the misjudgment occurs, so it is necessary to screen again according to the corner threshold.

[0108] In step S65, the hole where the seed is located is determined to be a reseeding hole.

[0109] In step S66, the hole where the seed is located is determined to be a qualified sowing hole.

[0110] like Fig.14 As shown in the figure, the center point distance threshold is set to 120, and the contour is initially judged. Due to the overlapping seeds problem, the classification result is wrong. Fig.15 As shown in , the second threshold is set to 90, and those greater than the threshold are recorded as reseeding, but due to the existence of a single large seed, the sorting results are still wrong. Fig.16 As shown, the third threshold is set to 8 and the second threshold is set to 70, and the classification result is correct. Therefore, it is finally determined to set the third threshold to 8 and the second threshold to 70.

[0111] In this embodiment, it is necessary to count the number of reseeded and qualified seeding holes, specifically, Fig.17 As shown, including:

[0112] In step S71, when it is determined that the hole is a reseeding seed hole, the number of the reseeding seed holes is increased by 1.

[0113] In step S72, when it is determined that the hole is a qualified sowing hole, the number of qualified sowing holes is increased by 1.

[0114] In this embodiment, it is necessary to finally obtain the sowing qualification rate of the sowing machine. The specific steps for calculating the qualification rate can be various steps known to those skilled in the art. In one example of the present invention, for example Fig.18 As shown, it may include:

[0115] In step S81, the number of holes of the implantation sponge is obtained.

[0116] In step S82, the number of holes minus the number of reseeded holes and the number of qualified holes is used to obtain the number of missed seeding holes.

[0117] In step S83, the seeding qualification rate is calculated according to the number of reseeded seed holes, the number of qualified seeding holes and the number of missed seeding seed holes.

[0118] On the other hand, the present invention provides an automatic seeder sowing quality detection system based on image recognition, the system comprising a processor for executing any of the detection methods described above.

[0119] Through the above technical scheme, the present invention provides a method and system for detecting the sowing quality of an automatic seeder based on image recognition. First, the acquired seed quality image is analyzed, and the re-seeding seed holes, that is, the situation where there are two non-overlapping seeds in one hole, are preliminarily screened out through the spacing between the center points of the seed contours. Then, the remaining candidate re-seeding seed holes are screened out for the overlapping seed holes and qualified seed holes, and the situation where the single seed is misjudged as an overlapping seed due to its large size is eliminated by setting the perimeter and corner point thresholds. Compared with the prior art, the present invention can directly detect and distinguish the sowing states of multiple seeds in one hole, overlapping multiple seeds, and independent multiple seeds when the seeder sows, and finally calculate the sowing qualification rate of the seeder sowing, providing basic data for improving the uniformity and yield of crops.

[0120] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0121] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A method for detecting the sowing quality of an automatic seeder based on image recognition, characterized in that: The detection method comprises: Obtain seed quality images; Preprocessing the quality image; Using the Canny operator to perform edge detection on the quality image to obtain an image of the outline of the seed; performing morphological processing on the image of the outline of the seed to remove noise; Acquire candidate reseeding seed holes and reseeding seed holes according to the distances between the center points of the contours of adjacent seeds; Further obtaining reseeding seed holes and qualified seed holes in the candidate reseeding seed holes according to the perimeter and corner points of the outline of the seed; Calculate the number of reseeding seed holes and the number of qualified seed holes according to the reseeding seed holes and qualified seed holes; The number of missed seeding holes is counted according to the number of reseeded seeding holes and the number of qualified seeding holes, and the seeding qualified rate is calculated.

2. The quality inspection method according to claim 1, characterized in that: Get seed quality images including: The hole tray is transferred to the position directly below the camera and the camera is fixed to ensure that the number of holes of the sponge and the coordinate position of each hole in the seed quality image obtained each time remain unchanged; Setting the shooting interval of the camera according to the time required for sowing to determine the horizontal fixed position of the camera; Obtain seed quality images at the sowing detection end through a camera; The seed quality image is transmitted to a detector in real time for detection.

3. The quality inspection method according to claim 1, characterized in that: Preprocessing the quality image includes: Cropping the quality image according to the rectangular area; Performing grayscale processing on the quality image; Gaussian filtering is used to reduce noise on the quality image.

4. The quality inspection method according to claim 1, characterized in that: The image of the seed outline is subjected to morphological processing to remove noise, including: Performing erosion and dilation operations on the image of the outline of the seed; Filter the image of the outline of the seed with a smaller area.

5. The quality inspection method according to claim 1, characterized in that: Obtaining the number of candidate reseeding seed holes of the holes according to the spacing between the center points of the contours of adjacent seeds includes: Get the geometric moment of the seed's outline; The coordinates of the geometric centroid are obtained according to the geometric moment and stored.

6. The quality inspection method according to claim 1, characterized in that: Acquiring candidate reseeding seed holes and reseeding seed holes according to the distances between the center points of the contours of adjacent seeds, comprising: Calculating the distance between the center points of the contours of adjacent seeds; Determining whether the distance is less than a first threshold; When it is determined that the distance is less than the first threshold, determining that the hole where the seed is located is a reseeding hole; When it is determined that the distance is not less than the first threshold, the hole where the seed is located is determined to be a candidate reseeding hole and the sowing state of the hole is further determined.

7. The quality inspection method according to claim 6, characterized in that: Further obtaining reseeding seed holes and qualified seed holes in the candidate reseeding seed holes according to the perimeter and corner points of the outline of the seed, including: Calculating the perimeter of the outline of the seed of a single hole; Determining whether the perimeter of the contour is greater than a second threshold; When it is determined that the perimeter of the contour is greater than the second threshold, determining that the hole where the seed is located is a reseeding hole; In the case where it is determined that the perimeter of the contour is not greater than the second threshold, determining whether the corner point threshold of the contour is greater than a third threshold; When it is determined that the corner point threshold of the contour is greater than a third threshold, determining that the hole where the seed is located is a reseeding hole; When it is determined that the corner point threshold of the seed is not greater than the third threshold, the hole where the seed is located is determined to be a qualified sowing hole.

8. The quality inspection method according to claim 7, characterized in that: Calculating the number of reseeding seed holes and the number of qualified seed holes according to the reseeding seed holes and the qualified seed holes includes: In the case where the hole is determined to be a reseeding hole, the number of the reseeding hole is increased by 1; When it is determined that the hole is a qualified sowing hole, the number of the qualified sowing holes is increased by 1.

9. The quality inspection method according to claim 8, characterized in that: The number of missed sowing holes is counted according to the number of re-sowing holes and the number of qualified sowing holes, and the sowing qualified rate is calculated, including: Obtain the number of holes in the colonized sponge; The number of holes minus the number of reseeding holes and the number of qualified seeding holes is used to obtain the number of missed seeding holes; The sowing qualified rate is calculated according to the number of re-sowing seed holes, the number of qualified sowing holes and the number of missed sowing seed holes.

10. A sowing quality detection system for an automatic seeder based on image recognition, characterized in that: The system comprises a processor for executing the detection method according to any one of claims 1 to 9.

Images