A smart detection method and system based on rice growth stage

Abstract

This application relates to an intelligent detection method and system based on the rice growth stage, belonging to the field of crop detection technology. The method includes acquiring an overall image of the farmland; performing feature recognition in the overall farmland image to determine rice panicle segment features, and defining the endpoints of the rice panicle segment features as rice panicle boundary points; counting rice panicle segment features at the same boundary point to determine the common number of boundaries, and defining rice panicle boundary points with a common number greater than a preset required common number as planting points; constructing a farmland planting distribution map based on all planting points, and constructing an avoidance detection path based on the planting points in the farmland planting distribution map; controlling a near-ground robot to move along the avoidance detection path to acquire near-ground detection images of each planting point, and combining the near-ground detection images with corresponding local farmland images of each planting point to determine a plant detection image. This application improves the intelligence of near-ground robots in detecting rice.

Description

Technical Field

[0001] This application relates to the field of crop detection technology, and in particular to an intelligent detection method and system based on the rice growth stage. Background Technology

[0002] As a vital global food crop, accurate monitoring of rice's growth stage is crucial for precision agricultural management, directly impacting water and fertilizer regulation, pest and disease control, and yield prediction. Traditional rice growth stage monitoring relies primarily on manual field surveys, which are inefficient, subjective, and difficult to implement for large-scale real-time monitoring. With the development of smart agriculture technologies, intelligent detection methods based on computer vision and mobile platforms have gradually become a research hotspot, mainly including unmanned aerial vehicle (UAV) remote sensing detection and near-ground robot detection.

[0003] While drones can quickly acquire large-scale images of the rice canopy through high-altitude aerial photography, their detection perspective is limited, capturing only information from the top of the canopy and failing to comprehensively reflect key growth characteristics such as tillers and stems in the lower and middle parts of the plant. In contrast, near-ground robots can move autonomously in the field, acquiring a more comprehensive collection of rice canopy, stem, and tiller structures through multi-angle imaging, resulting in significantly higher detection accuracy than drones.

[0004] Among the aforementioned technologies, due to irregular farmland terrain, deviations in manual transplanting, or the influence of natural growth, the actual plant distribution is often not completely orderly in rows and columns. If a near-ground robot is used for inspection, staff need to input the movement path to reduce the possibility of the robot colliding with the stems and causing plant damage. Staff need to input the movement path in advance for different farmland environments, which has poor universality and is not convenient for near-ground robots to perform intelligent inspection of rice. There is still room for improvement. Summary of the Invention

[0005] To improve the intelligence of near-ground robots in detecting rice, this application provides an intelligent detection method and system based on the rice growth stage.

[0006] Firstly, this application provides an intelligent detection method based on the rice growth stage, employing the following technical solution:

[0007] A smart detection method based on rice growth stage includes:

[0008] Obtain the outline of farmland boundaries;

[0009] The flight path is generated based on the farmland boundary contour, and the UAV is controlled to move along the flight path to obtain local images of the farmland. All local images of the farmland are then merged to construct an overall image of the farmland.

[0010] Feature recognition is performed on the overall image of farmland to determine the features of rice spike line segments, and the endpoints of the rice spike line segments are defined as the boundary points of the rice spike.

[0011] The common number of boundaries is determined by counting the rice spike line segments at the same boundary point, and the boundary points of rice spikes with a common number of boundaries greater than the preset required common number are defined as planting points.

[0012] A farmland planting distribution map is constructed based on all planting points, and an avoidance detection path is constructed based on the planting points in the farmland planting distribution map.

[0013] The near-ground robot is controlled to move along the avoidance detection path to obtain near-ground detection images of each planting point, and the near-ground detection images and the corresponding local farmland images of each planting point are combined to determine the plant detection images.

[0014] Optionally, after the planting site is determined, intelligent detection methods based on the rice growth stage also include:

[0015] Connect any two planting points to construct a straight line for planting coverage;

[0016] The distance between each planting point and the planting cover line is determined, and the planting point whose distance between the point and the line is less than the preset benchmark distance is defined as the extension point of the planting cover line.

[0017] The number of expansion points is determined by counting, and planting cover lines with an expansion number less than the preset baseline arrangement number are removed.

[0018] Project all expansion points onto the corresponding planting cover line to determine the sequential projection points, and determine the planting distance based on the planting points corresponding to adjacent sequential projection points.

[0019] Select one planting distance from all planting distances as the primary planting distance, and define the remaining planting distances as secondary planting distances;

[0020] The balance anomaly parameter is determined by calculating the key distance and all secondary distances, and a flag anomaly signal is output when the balance anomaly parameter is greater than the preset benchmark anomaly parameter.

[0021] Optionally, after the abnormal signal is marked and output, the intelligent detection method based on the rice growth stage also includes:

[0022] The distance between key points corresponding to the marked abnormal signals is defined as the abnormal distance, and the distance between the remaining key points is defined as the normal distance.

[0023] The mean of all normal intervals is calculated to construct the normal mean distance, and it is determined whether the abnormal intervals are greater than the normal mean distance.

[0024] If the abnormal distance is not greater than the normal average distance, a short distance marker is added to the planting point corresponding to the current abnormal distance, and the number of short distances is determined based on the short distance markers. When the number of short distances is consistent with the preset number of times the target is met, the corresponding planting point is canceled.

[0025] If the abnormal interval is greater than the normal average interval, the relative distance multiple is determined by calculation based on the abnormal interval and the normal average interval.

[0026] Determine whether the relative distance multiple is within a reasonable range of preset multiples;

[0027] If the relative distance multiple is within a reasonable range, then output a point missing signal;

[0028] If the relative distance multiple is not within a reasonable range, a waiting analysis signal will be output based on the planting point corresponding to the abnormal distance.

[0029] Optionally, after the signal for missing data points is output, the intelligent detection method based on the rice growth stage also includes:

[0030] The number of missing points is determined based on the relative multiples of the distance.

[0031] The missing supplementary areas are delineated based on the expansion points corresponding to the abnormal spacing and the corresponding planting cover lines;

[0032] In the missing supplementation area, simulated supplementation points are randomly generated based on the number of missing points, and a simulated supplementation scheme is constructed based on the generated simulated supplementation points;

[0033] The expansion point is redefined and the equilibrium anomaly parameter is calculated based on the simulation supplement scheme. When the equilibrium anomaly parameter is less than the baseline anomaly parameter, the simulation supplement scheme is defined as a permissible supplement scheme.

[0034] The actual supplementary plan is determined in the licensed supplementary plan, and the simulated supplementary points corresponding to the actual supplementary plan are defined as planting points.

[0035] Optionally, the steps for determining the actual supplementary scheme in the licensing supplement scheme include:

[0036] Based on the simulated supplementary points and the preset similar distances, the similar range of points is defined, and the boundary points of rice ears within the similar range are defined as internal points;

[0037] The internal spacing distance is determined by calculating based on the internal points and simulated supplementary points, and the internal mean distance is determined by averaging all internal spacing distances.

[0038] The distance deviation parameter is determined by calculating the mean internal distance and all internal intervals.

[0039] The number of internal points is determined by counting the internal points, and the appropriate parameters for a single point are determined by calculating the number of internal points, the average distance between internal points, and the distance deviation parameter.

[0040] The appropriate parameters for the scheme are determined by calculation based on all single-point appropriate parameters, and the permitted supplementary scheme corresponding to the appropriate parameter of the scheme with the largest value is determined as the actual supplementary scheme.

[0041] Optionally, after partially eliminating the planting cover line, intelligent detection methods based on rice growth stages also include:

[0042] Determine whether there are any planting points that are not defined as extension points of any planting cover line;

[0043] If there is no planting point that is not defined as an extension point of any planting cover line, then output the complete analysis signal;

[0044] If there is a planting point that is not defined as an extension point of any planting cover line, then the planting point is defined as a suspicious point, and the remaining planting points are defined as mature points;

[0045] Determine the ripening interval between any two ripening points, and define the minimum ripening interval as the representative interval between the corresponding ripening points;

[0046] The average distance between all representatives is calculated to determine the average distance between them, and the reasonable distance range is determined based on the average distance between the representatives and the preset reasonable deviation distance.

[0047] The suspicious distances are determined based on the suspicious points and each mature point, and the smallest suspicious distance is defined as the suspicious representative distance of the corresponding suspicious point.

[0048] Determine whether the distance to the suspected representative is within a reasonable range;

[0049] If the distance to the suspected point is within a reasonable range, the planting point currently defined as a suspected point will be maintained.

[0050] If the distance to a suspected point is not within a reasonable range, the planting point defined as a suspected point will be cancelled.

[0051] Secondly, this application provides an intelligent detection system based on the rice growth period, employing the following technical solution:

[0052] A smart detection system based on the rice growth stage includes:

[0053] The acquisition module is used to acquire the outline of farmland boundaries;

[0054] The processing module, connected to the acquisition module, is used for information storage and processing;

[0055] The processing module generates a flight path based on the farmland boundary contour and controls the UAV to move along the flight path so that the acquisition module can acquire local farmland images. All local farmland images are then merged to construct a holistic farmland image.

[0056] The processing module performs feature recognition in the overall image of the farmland to determine the features of rice ear segments, and defines the endpoints of the rice ear segment features as rice ear boundary points;

[0057] The processing module counts the rice spike line segments at the same rice spike boundary point to determine the common number of boundaries, and defines the rice spike boundary points with a common number of boundaries greater than the preset required common number as planting points;

[0058] The processing module constructs a farmland planting distribution map based on all planting points, and constructs an avoidance detection path based on the planting points in the farmland planting distribution map;

[0059] The processing module controls the near-ground robot to move along the avoidance detection path so that the acquisition module can acquire near-ground detection images of each planting point, and combines the near-ground detection images with the corresponding local farmland images of each planting point to determine the plant detection images.

[0060] Thirdly, this application provides a computer storage medium capable of storing corresponding programs, which enhances the intelligence of near-ground robots in detecting rice, and adopts the following technical solution:

[0061] A computer-readable storage medium storing a computer program that can be loaded by a processor and executed for any of the above-described intelligent detection methods based on rice growth stages.

[0062] In summary, this application includes at least one of the following beneficial technical effects:

[0063] Before using near-ground robots for inspection, drones are used to fly and inspect the rice plants to determine the distribution of planting sites. This facilitates the intelligent planning of the near-ground robot's inspection path, thereby improving the intelligence of the near-ground robot in inspecting rice.

[0064] During the process of determining planting sites, some falsely detected and missed planting sites can be identified, thereby improving the accuracy of data analysis. Attached Figure Description

[0065] Figure 1 This is a flowchart of an intelligent detection method based on the rice growth period.

[0066] Figure 2 This is a schematic diagram of binarization processing of an overall image of farmland.

[0067] Figure 3 This is a diagram illustrating the missing and supplementary areas.

[0068] Figure 4 This is a flowchart of a module based on an intelligent detection method for rice growth stages. Detailed Implementation

[0069] To make the purpose, technical solution, and advantages of this application clearer, the following is combined with Figures 1-4 The present application will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the application.

[0070] The embodiments of this application will now be described in further detail with reference to the accompanying drawings.

[0071] This application discloses an intelligent detection method based on the rice growth stage, referring to... Figure 1 The method flow of the intelligent detection method based on rice growth stage includes the following steps:

[0072] Step S100: Obtain the farmland boundary outline.

[0073] The farmland boundary outline is the outer boundary outline of the farmland where the rice to be detected is located. It can be obtained by identifying the outline of the field ridge.

[0074] Step S101: Generate a flight path based on the farmland boundary contour, control the UAV to move along the flight path to obtain local farmland images, and fuse all local farmland images to construct an overall farmland image.

[0075] The flight path is the path taken by the UAV to completely detect the area enclosed by the farmland boundary outline. Generally, it is a zigzag path, that is, a boundary point is determined on the farmland boundary outline as the starting point, and then a suitable direction is determined to divide the path. When other outline points are detected, the path is controlled to translate and turn until the UAV can detect all farmland areas. The method of generating the flight path is a conventional technique for those skilled in the art and is not an innovation of this application, so it will not be described in detail here. The local farmland image is the image of the farmland area acquired by the UAV at a single time point when it moves along the flight path. The overall farmland image is the total image of all farmland areas obtained after performing point cloud fusion operation on all local farmland images.

[0076] Step S102: Perform feature recognition in the overall image of the farmland to determine the features of rice ear segments, and define the endpoints of the rice ear segment features as rice ear boundary points.

[0077] The Otsu adaptive thresholding method is selected to binarize the overall farmland image to obtain a binarized image. In this binarized image, white pixel areas represent rice paddies, and black pixel areas represent background areas. Therefore, the rice ear line segment feature is the independent white region in the binarized image, which can represent the rice ear. Figure 2 The feature recognition method for rice ear line segments can be to obtain a recognition database in advance through deep learning of neural networks, and then input the current overall image of farmland into the recognition database to obtain rice ear line segment features; the rice ear boundary points are the two endpoints of the rice ear line segment features. Since multiple rice ears grow from the rice stalks, the definition of the rice ear boundary points can better determine the stalks.

[0078] Step S103: Count the rice ear line segments on the same rice ear boundary point to determine the common number of boundaries, and define the rice ear boundary points with a common number of boundaries greater than the preset required common number as planting points.

[0079] The common boundary quantity refers to the number of rice spike line segments with the same boundary point as their endpoints. The required common quantity is the minimum common boundary quantity set by the staff to indicate that the boundary point of the rice spike is highly likely to be a stem. When the common boundary quantity is greater than the required common quantity, it means that the endpoints of multiple rice spike line segments are at the same point, which conforms to the natural law between rice spikes and stems. Therefore, the probability that the boundary point of the rice spike is the location of the stem is high. Planting points are defined to distinguish different location points for the convenience of subsequent analysis.

[0080] Step S104: Construct a farmland planting distribution map based on all planting points, and construct an avoidance detection path based on the planting points in the farmland planting distribution map.

[0081] The farmland planting distribution map reflects the distribution of rice planting locations in the current farmland area. The avoidance detection path is a movement path that allows the near-ground robot to effectively detect rice at each planting point without colliding with rice stalks during movement. It can be formed by constructing a detection point at the same distance in the same direction at each planting point, and then connecting all detection points in series according to the shortest path rule. The specific path construction method can be set and stored in advance by the staff.

[0082] Step S105: Control the near-ground robot to move along the avoidance detection path to obtain near-ground detection images of each planting point, and combine the near-ground detection images with the corresponding local farmland images of each planting point to determine the plant detection images.

[0083] The near-ground detection image is the image obtained after the near-ground robot detects the rice. At this time, the near-ground robot and the drone acquire images of the rice from different angles. The corresponding images are then combined to form a plant detection image that reflects the overall condition of the plant. This plant detection image can then be used to detect the growth stage of the rice.

[0084] After the planting site is determined, intelligent detection methods based on the rice growth stage also include:

[0085] Step S200: Connect any two planting points to construct a planting coverage line.

[0086] The planting cover line is a straight line that passes through any two planting points. Since rice is generally planted in rows and columns, the distribution of planting points under the straight line condition can be used to analyze the determined planting points.

[0087] Step S201: Determine the distance between each planting point and the planting cover line, and define the planting point whose distance between the planting point and the line is less than the preset benchmark positioning distance as the extension point of the planting cover line.

[0088] The distance between the dots and lines is the vertical distance between the planting point and the planting cover line. The baseline distance is the maximum distance between the dots and lines allowed when the planting point is located around the planting cover line, as set by the staff. When the distance between the dots and lines is less than the baseline distance, it means that the planting point is arranged in a row with the planting points that make up the planting cover line, which meets the requirement of orderly rice planting rows. Therefore, extension points are defined to identify different planting points for subsequent analysis.

[0089] Step S202: Count the expansion points to determine the expansion quantity, and remove planting cover lines whose expansion quantity is less than the preset baseline arrangement quantity.

[0090] The expansion quantity is the total number of expansion points determined by a single planting cover line. The baseline arrangement quantity is the minimum expansion quantity required when the corresponding planting points are arranged in an orderly manner according to the row and column requirements set by the staff. When the expansion quantity is less than the baseline arrangement quantity, it means that the planting cover line at this time cannot reflect the planting arrangement of rice, so it is removed to reduce analysis error.

[0091] Step S203: Project all expansion points onto the corresponding planting cover lines to determine the sequential projection points, and determine the planting distance based on the planting points corresponding to adjacent sequential projection points.

[0092] The sequential projection point is the point obtained by vertically projecting the extension point onto the planting cover line, and the planting interval distance is the straight-line distance between the planting points corresponding to adjacent sequential projection points on the planting cover line.

[0093] Step S204: Select one planting distance from all planting distances as the primary planting distance, and define the remaining planting distances as secondary planting distances.

[0094] Define key and secondary planting distances to differentiate between different planting distances, which will facilitate subsequent analysis.

[0095] Step S205: Calculate and determine the balance anomaly parameter based on the key distance and all secondary distances, and output a marker anomaly signal when there is a balance anomaly parameter that is greater than the preset benchmark anomaly parameter.

[0096] The balance anomaly parameter reflects whether the spacing between planting points meets the requirements for rice row planting. The larger the value, the greater the possibility of anomalies in the planting point location among the identified expansion points. It is determined by subtracting the key spacing from all secondary spacings and averaging the absolute values. The baseline anomaly parameter is the minimum balance anomaly parameter set by the staff when the corresponding planting point location is identified as abnormal. Anomaly signals are defined to identify abnormal situations at planting points, enabling staff to intervene in a timely manner or conduct subsequent data analysis based on the situation.

[0097] Following the output of the marked abnormal signal, the intelligent detection method based on the rice growth stage also includes:

[0098] Step S300: Define the key interval distance corresponding to the marked abnormal signal as the abnormal interval distance, and define the remaining key interval distances as the normal interval distance.

[0099] Define abnormal and normal intervals to distinguish different key intervals, which will facilitate subsequent analysis.

[0100] Step S301: Calculate the mean of all normal intervals to construct the normal mean distance, and determine whether the abnormal intervals are greater than the normal mean distance.

[0101] The normal mean distance is the average of all normal intervals. The purpose of this judgment is to determine whether the current distance is too large or too small.

[0102] Step S3011: If the abnormal distance is not greater than the normal average distance, add a short distance marker to the planting point corresponding to the current abnormal distance, count according to the short distance marker to determine the number of short distances, and cancel the corresponding planting point when the number of short distances is consistent with the preset number of times the target is met.

[0103] When the abnormal distance is not greater than the normal average distance, it indicates that the current distance is too small, meaning that some points that are not actual planting locations are defined as planting points. Therefore, short distance markers are added to distinguish different planting points. The short distance count is the number of times the same planting point is marked with a short distance marker. The target count is the number of short distances that staff set to be required for a planting point to be considered an abnormal planting location. For planting points in the middle, the target count is 2, and for planting points at the side endpoints, the target count is 1. Planting points with the same short distance count and target count are canceled to screen for abnormal planting points.

[0104] Step S3012: If the abnormal interval distance is greater than the normal average distance, then calculate the relative distance multiple based on the abnormal interval distance and the normal average distance.

[0105] When the abnormal distance is greater than the normal average distance, it indicates that the distance is too large, and there may be planting points in the middle that have not been marked, requiring further analysis; the relative distance multiple is the value obtained by dividing the abnormal distance by the normal average distance.

[0106] Step S302: Determine whether the relative distance multiple is within the preset reasonable range.

[0107] The reasonable range of the multiple is the range that the staff set for the relative multiple of the distance when there is a missing planting point, such as 1.8-2.2, 2.8-3.2, etc. The specific range is determined in advance by the staff. The purpose of the judgment is to find out whether there is a missing planting point and to automatically fill it.

[0108] Step S3021: If the relative distance multiple is within a reasonable range, output a point missing signal.

[0109] When the relative distance multiple is within a reasonable range, it indicates that there are missing planting points and that these points can be automatically filled in through data analysis. At this time, a missing point signal is output to identify the situation for subsequent analysis.

[0110] Step S3022: If the relative distance multiple is not within a reasonable range, output a waiting analysis signal based on the planting point corresponding to the abnormal distance.

[0111] When the relative distance multiple is not within a reasonable range, it indicates that there are missing planting points. However, since the distance cannot be automatically used to fill in the missing planting points, a waiting analysis signal is output so that staff can intervene in a timely manner.

[0112] Following the output of missing location signals, intelligent detection methods based on rice growth stages also include:

[0113] Step S400: Determine the number of missing points based on the relative distance multiple.

[0114] The number of missing planting points is determined by subtracting one from the relative distance and rounding according to the rules.

[0115] Step S401: Delineate the missing supplementary area based on the expansion points corresponding to the abnormal spacing and the corresponding planting cover lines.

[0116] The missing patch area is the region where the missing planting point should theoretically be located, i.e., the area to be analyzed when supplementing the missing planting point. This area is rectangular in shape, with its length being the abnormal spacing distance and its width being twice the baseline positioning distance. The expansion point corresponding to the abnormal spacing distance is located at the midpoint of the width boundary line. For specific methods of defining the missing patch area, please refer to [reference needed]. Figure 3 .

[0117] Step S402: Randomly generate simulated supplementary points in the missing supplementary area based on the number of missing points, and construct a simulated supplementary scheme based on the generated simulated supplementary points.

[0118] By randomly generating simulated supplementary points, missing planting points can be simulated and analyzed to generate corresponding simulated supplementary plans and analyze the simulated supplementary situation.

[0119] Step S403: Based on the simulation supplementary scheme, redetermine the expansion point and calculate the equilibrium anomaly parameter, and define the simulation supplementary scheme as a permissible supplementary scheme when the equilibrium anomaly parameter is less than the baseline anomaly parameter.

[0120] Recalculating the balance anomaly parameters can reflect the rationality of the supplementary point setting. Therefore, we define permissible supplementary schemes to identify schemes with more reasonable supplementary situations, which will facilitate subsequent analysis.

[0121] Step S404: Determine the actual supplementary plan in the permitted supplementary plan, and define the simulated supplementary point corresponding to the actual supplementary plan as the planting point.

[0122] The actual supplementary scheme is a licensed supplementary scheme for use. It can be randomly selected or selected through the method of steps S500-S504. By defining the corresponding simulated supplementary points as planting points, the missing planting points are supplemented, thereby improving the accuracy of data analysis.

[0123] The steps for determining the actual supplementary scheme in the license supplement scheme include:

[0124] Step S500: Determine the range of similar points based on the simulated supplementary points and the preset similar distances, and define the boundary points of rice ears within the range of similar points as internal points.

[0125] The "close distance" refers to the maximum allowable interval between points set by the staff. The "close distance" range is defined with the simulated supplementary point as the midpoint and the "close distance" as the radius. Internal points are defined to mark the boundary points of rice ears around the simulated supplementary point, which facilitates subsequent analysis.

[0126] Step S501: Calculate the internal distance based on the internal points and the simulated supplementary points, and calculate the average internal distance based on the average of all internal distances.

[0127] The internal interval distance is the straight-line distance between the internal points and the simulated supplementary points, and the internal mean distance is the average of all internal interval distances.

[0128] Step S502: Calculate and determine the distance deviation parameter based on the internal mean distance and all internal intervals.

[0129] The distance deviation parameter reflects the deviation between the simulated supplementary point and each internal point. The larger the value, the greater the deviation. It is determined by calculating the difference between each internal distance and the internal mean distance, and then averaging the absolute values.

[0130] Step S503: Count the internal points to determine the internal quantity, and calculate the appropriate parameters for a single point based on the internal quantity, the internal mean distance, and the distance deviation parameter.

[0131] The internal quantity refers to the total number of internal points within the defined proximity range. The single-point suitability parameter indicates whether the currently set single simulated supplementary point is suitable for its corresponding position. The larger this value, the more suitable the corresponding position. The formula for calculating the single-point suitability parameter is as follows: ,in For suitable parameters at a single point, For internal quantity, The internal mean distance, This is the distance deviation parameter. , as well as These are preset exponential parameters used to control the nonlinear effects of various parameters. , as well as These are preset weighting coefficients used to adjust the relative importance of each parameter. It is a very small constant to prevent the denominator from being 0.

[0132] Step S504: Calculate the appropriate parameters of the scheme based on all single-point appropriate parameters to determine the appropriate parameters of the scheme, and determine the permitted supplementary scheme corresponding to the appropriate parameter of the scheme with the largest value as the actual supplementary scheme.

[0133] The optimal parameter for the scheme is the average of the optimal parameters of each individual point in the simulated supplementary scheme. The scheme with the largest optimal parameter indicates that the corresponding supplementary scheme is the most reasonable, and therefore it can be determined as the actual supplementary scheme.

[0134] After partially eliminating the planting cover line, intelligent detection methods based on rice growth stages also include:

[0135] Step S600: Determine whether there are any planting points that are not defined as extension points of any planting cover line.

[0136] The purpose of the determination is to determine whether there are independent planting points that are not located on any planting cover line.

[0137] Step S6001: If there is no planting point that is not defined as an extension point of any planting cover line, then output the complete analysis signal.

[0138] If there is no planting point that is not defined as an extension point of any planting coverage line, it means that all planting points have been analyzed well. In this case, a complete analysis signal can be output to identify the situation.

[0139] Step S6002: If there is a planting point that is not defined as an extension point of any planting coverage line, then define the planting point as a suspicious point and define the remaining planting points as mature points.

[0140] When there is a planting point that is not defined as an extension point of any planting coverage line, it indicates that there is an independent planting point. In this case, it is necessary to further analyze the reliability of the planting point; define suspicious points and mature points to distinguish different planting points, which will facilitate subsequent analysis.

[0141] Step S601: Determine the ripening interval between any two ripening points, and define the minimum ripening interval as the representative interval between the corresponding ripening points.

[0142] The maturity interval is the straight-line distance between two maturity points, representing the shortest interval between individual rice plants and the rest of the rice plants.

[0143] Step S602: Calculate the average distance between all representatives based on the distance between them, and calculate the reasonable distance range based on the average distance between the representatives and the preset reasonable deviation distance.

[0144] The representative mean distance is the average of all representative distances, which reflects the size of the rice planting interval in the current farmland. The reasonable deviation distance is the maximum value set by the staff that allows for deviations in the size of the planting interval. The reasonable distance range is the range of distances that the minimum interval between rice and other rice should be when rice is planted normally. It is determined by adding and subtracting the reasonable deviation distance from the representative mean distance.

[0145] Step S603: Determine the suspicious distance based on the suspicious points and each mature point, and define the smallest suspicious distance as the suspicious representative distance of the corresponding suspicious point.

[0146] The suspicious distance is the straight-line distance between the suspicious point and the mature point. By defining it, it can represent the distance to determine the closest distance between the current suspicious point and the mature point, which is convenient for subsequent analysis.

[0147] Step S604: Determine whether the distance to the suspected representative is within a reasonable range.

[0148] The purpose of the assessment is to determine whether the currently analyzed suspicious points have a high probability of being actual planting sites.

[0149] Step S6041: If the distance to the suspected point is within a reasonable range, then maintain the planting point currently defined as a suspected point.

[0150] When the distance to the suspected point is within a reasonable range, it indicates that the current suspected point is likely to be the actual planting point, so its corresponding definition label can be maintained.

[0151] Step S6042: If the distance to the suspected point is not within a reasonable range, the planting point defined as a suspected point will be cancelled.

[0152] When the distance to a suspected point is not within a reasonable range, it indicates that the current suspected point is less likely to be an actual planting point. Therefore, its definition is cancelled to improve the accuracy of subsequent data analysis.

[0153] Reference Figure 4 Based on the same inventive concept, embodiments of the present invention provide an intelligent detection system based on the rice growth stage, comprising:

[0154] The acquisition module is used to acquire the outline of farmland boundaries;

[0155] The processing module, connected to the acquisition module, is used for information storage and processing;

[0156] The processing module generates a flight path based on the farmland boundary contour and controls the UAV to move along the flight path so that the acquisition module can acquire local farmland images. All local farmland images are then merged to construct a holistic farmland image.

[0157] The processing module performs feature recognition in the overall image of the farmland to determine the features of rice ear segments, and defines the endpoints of the rice ear segment features as rice ear boundary points;

[0158] The processing module counts the rice spike line segments at the same rice spike boundary point to determine the common number of boundaries, and defines the rice spike boundary points with a common number of boundaries greater than the preset required common number as planting points;

[0159] The processing module constructs a farmland planting distribution map based on all planting points, and constructs an avoidance detection path based on the planting points in the farmland planting distribution map;

[0160] The processing module controls the near-ground robot to move along the avoidance detection path so that the acquisition module can acquire near-ground detection images of each planting point, and combines the near-ground detection images with the corresponding local farmland images of each planting point to determine the plant detection images.

[0161] The anomaly analysis module analyzes anomalies in planting point markings.

[0162] The redundancy analysis module analyzes cases where planting points are marked redundantly.

[0163] The marker missing analysis module analyzes situations where planting point markers are missing.

[0164] The actual supplementary scheme determination module is used to determine a unique actual supplementary scheme from multiple licensing supplementary schemes;

[0165] The planting site analysis module analyzes and processes data from individual planting sites.

[0166] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, device, and unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0167] This invention provides a computer-readable storage medium storing a computer program that can be loaded and executed by a processor, which is a smart detection method based on the rice growth stage.

[0168] Computer storage media include, for example, USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, optical disks, and other media that can store program code.

Claims

1. A smart detection method based on rice growth stage, characterized in that, include: Obtain the outline of farmland boundaries; The flight path is generated based on the farmland boundary contour, and the UAV is controlled to move along the flight path to obtain local images of the farmland. All local images of the farmland are then merged to construct an overall image of the farmland. Feature recognition is performed on the overall image of farmland to determine the features of rice spike line segments, and the endpoints of the rice spike line segments are defined as the boundary points of the rice spike. The common number of boundaries is determined by counting the rice spike line segments at the same boundary point, and the boundary points of rice spikes with a common number of boundaries greater than the preset required common number are defined as planting points. A farmland planting distribution map is constructed based on all planting points, and an avoidance detection path is constructed based on the planting points in the farmland planting distribution map. Control the near-ground robot to move along the avoidance detection path to obtain near-ground detection images of each planting point, and combine the near-ground detection images with the corresponding local farmland images of each planting point to determine the plant detection image; After the planting site is determined, intelligent detection methods based on the rice growth stage also include: Connect any two planting points to construct a straight line for planting coverage; The distance between each planting point and the planting cover line is determined, and the planting point whose distance between the point and the line is less than the preset benchmark distance is defined as the extension point of the planting cover line. The number of expansion points is determined by counting, and planting cover lines with an expansion number less than the preset baseline arrangement number are removed. Project all expansion points onto the corresponding planting cover line to determine the sequential projection points, and determine the planting distance based on the planting points corresponding to adjacent sequential projection points. Select one planting distance from all planting distances as the primary planting distance, and define the remaining planting distances as secondary planting distances; The balance anomaly parameter is determined by calculating the key distance and all secondary distances, and a flag anomaly signal is output when the balance anomaly parameter is greater than the preset benchmark anomaly parameter.

2. The intelligent detection method based on rice growth stage according to claim 1, characterized in that, Following the output of the marked abnormal signal, the intelligent detection method based on the rice growth stage also includes: The distance between key points corresponding to the marked abnormal signals is defined as the abnormal distance, and the distance between the remaining key points is defined as the normal distance. The mean of all normal intervals is calculated to construct the normal mean distance, and it is determined whether the abnormal intervals are greater than the normal mean distance. If the abnormal distance is not greater than the normal average distance, a short distance marker is added to the planting point corresponding to the current abnormal distance, and the number of short distances is determined based on the short distance markers. When the number of short distances is consistent with the preset number of times the target is met, the corresponding planting point is canceled. If the abnormal interval is greater than the normal average interval, the relative distance multiple is determined by calculation based on the abnormal interval and the normal average interval. Determine whether the relative distance multiple is within a reasonable range of preset multiples; If the relative distance multiple is within a reasonable range, then output a point missing signal; If the relative distance multiple is not within a reasonable range, a waiting analysis signal will be output based on the planting point corresponding to the abnormal distance.

3. The intelligent detection method based on rice growth stage according to claim 2, characterized in that, Following the output of missing location signals, intelligent detection methods based on rice growth stages also include: The number of missing points is determined based on the relative multiples of the distance. The missing supplementary areas are delineated based on the expansion points corresponding to the abnormal spacing and the corresponding planting cover lines; In the missing supplementation area, simulated supplementation points are randomly generated based on the number of missing points, and a simulated supplementation scheme is constructed based on the generated simulated supplementation points; The expansion point is redefined and the equilibrium anomaly parameter is calculated based on the simulation supplement scheme. When the equilibrium anomaly parameter is less than the baseline anomaly parameter, the simulation supplement scheme is defined as a permissible supplement scheme. The actual supplementary plan is determined in the licensed supplementary plan, and the simulated supplementary points corresponding to the actual supplementary plan are defined as planting points.

4. The intelligent detection method based on rice growth stage according to claim 3, characterized in that, The steps for determining the actual supplementary scheme in the license supplement scheme include: Based on the simulated supplementary points and the preset similar distances, the similar range of points is defined, and the boundary points of rice ears within the similar range are defined as internal points; The internal spacing distance is determined by calculating based on the internal points and simulated supplementary points, and the internal mean distance is determined by averaging all internal spacing distances. The distance deviation parameter is determined by calculating the mean internal distance and all internal intervals. The number of internal points is determined by counting the internal points, and the appropriate parameters for a single point are determined by calculating the number of internal points, the average distance between internal points, and the distance deviation parameter. The appropriate parameters for the scheme are determined by calculation based on all single-point appropriate parameters, and the permitted supplementary scheme corresponding to the appropriate parameter of the scheme with the largest value is determined as the actual supplementary scheme.

5. The intelligent detection method based on rice growth stage according to claim 1, characterized in that, After partially eliminating the planting cover line, intelligent detection methods based on rice growth stages also include: Determine whether there are any planting points that are not defined as extension points of any planting cover line; If there is no planting point that is not defined as an extension point of any planting cover line, then output the complete analysis signal; If there is a planting point that is not defined as an extension point of any planting cover line, then the planting point is defined as a suspicious point, and the remaining planting points are defined as mature points; Determine the ripening interval between any two ripening points, and define the minimum ripening interval as the representative interval between the corresponding ripening points; The average distance between all representatives is calculated to determine the average distance between them, and the reasonable distance range is determined based on the average distance between the representatives and the preset reasonable deviation distance. The suspicious distances are determined based on the suspicious points and each mature point, and the smallest suspicious distance is defined as the suspicious representative distance of the corresponding suspicious point. Determine whether the distance to the suspected representative is within a reasonable range; If the distance to the suspected point is within a reasonable range, the planting point currently defined as a suspected point will be maintained. If the distance to a suspected point is not within a reasonable range, the planting point defined as a suspected point will be cancelled.

6. An intelligent detection system based on rice growth stage, used to execute the intelligent detection method based on rice growth stage as described in any one of claims 1 to 5, characterized in that, include: The acquisition module is used to acquire the outline of farmland boundaries; The processing module, connected to the acquisition module, is used for information storage and processing; The processing module generates a flight path based on the farmland boundary contour and controls the UAV to move along the flight path so that the acquisition module can acquire local farmland images. All local farmland images are then merged to construct a holistic farmland image. The processing module performs feature recognition in the overall image of the farmland to determine the features of rice ear segments, and defines the endpoints of the rice ear segment features as rice ear boundary points; The processing module counts the rice spike line segments at the same rice spike boundary point to determine the common number of boundaries, and defines the rice spike boundary points with a common number of boundaries greater than the preset required common number as planting points; The processing module constructs a farmland planting distribution map based on all planting points, and constructs an avoidance detection path based on the planting points in the farmland planting distribution map; The processing module controls the near-ground robot to move along the avoidance detection path so that the acquisition module can acquire near-ground detection images of each planting point, and combines the near-ground detection images with the corresponding local farmland images of each planting point to determine the plant detection images.

7. A computer-readable storage medium, characterized in that, The system stores a computer program capable of being loaded by a processor and executing the intelligent detection method based on the rice growth period as described in any one of claims 1 to 5.

Images