An intelligent evaluation method and operation system for water and fertilizer utilization efficiency

Abstract

The application relates to the technical field of water and fertilizer application evaluation, and particularly discloses an intelligent evaluation method and an operation system for water and fertilizer utilization efficiency, the method comprising the following steps: obtaining a crop image of crops corresponding to fertilization parameters; identifying the crop image, outputting a growth vigor evaluation report, and constructing a mapping relationship from the fertilization parameters to the growth vigor evaluation report, which is used for evaluating the fertilization parameters; the fertilization parameters are recorded, at the same time, the crop image is obtained through visual monitoring points, the crop image is identified, the growth vigor conditions before and after fertilization are determined, the growth vigor difference before and after fertilization is determined, and a mapping relationship from the fertilization parameters to the growth vigor difference is constructed; when facing a prediction demand, the growth vigor difference of the fertilization parameters can be predicted before the fertilization behavior, and the prediction accuracy is high.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of water and fertilizer application evaluation, and particularly relates to an intelligent evaluation method and operation system for water and fertilizer utilization efficiency. BACKGROUND

[0002] Intelligent agricultural water and fertilizer integration is an agricultural management mode in which modern information technology and agricultural scientific methods are deeply integrated. It determines water and fertilizer demand information by comprehensively considering the soil fertility conditions (such as the content of various nutrients in the soil, soil texture, etc.), the actual needs of crops at different growth stages (such as different requirements for water and nutrients at seedling stage, flowering stage, fruiting stage, etc.), and meteorological conditions (including factors such as air temperature, precipitation, and light duration). In the prior art, the correspondence between various conditions and water and fertilizer demand information is determined by big data, and the effectiveness is high.

[0003] However, in actual application, the staff often adjusts the water and fertilizer application process according to the actual situation, and the staff is difficult to know the effect brought by different water and fertilizer application information. Therefore, how to provide an evaluation scheme for the water and fertilizer application process to assist the staff in adjusting the water and fertilizer application process is a technical problem to be solved by the technical scheme of the present application. SUMMARY

[0004] The present application aims to provide an intelligent evaluation method and operation system for water and fertilizer utilization efficiency to solve the problems in the background art.

[0005] To achieve the above-mentioned purpose, the present application provides the following technical scheme:

[0006] An intelligent evaluation method for water and fertilizer utilization efficiency, the method comprising:

[0007] Obtaining crop distribution information, and selecting a visual monitoring point according to the crop distribution information;

[0008] Real-time recording of fertilization parameters, generating a collection instruction pointing to the visual monitoring point according to the fertilization parameters, and obtaining a crop image of the crop corresponding to the fertilization parameters; the fertilization parameters include fertilization point, fertilization type and fertilization amount;

[0009] Identifying the crop image, determining crop pulse drop, determining an initial detection point in real time based on the crop pulse drop, performing regional continuation based on the initial detection point, and determining a detection area;

[0010] Screening the determined detection area to obtain a regional connection graph, inputting the regional connection graph into a trained neural network recognition model, outputting a growth vigor evaluation report, constructing a mapping relationship from the fertilization parameters to the growth vigor evaluation report, and using the mapping relationship to evaluate the fertilization parameters.

[0011] As a further scheme of the present application: the step of acquiring crop distribution information and selecting visual monitoring points according to the crop distribution information comprises:

[0012] Acquiring a remote sensing image of the crop area, identifying the remote sensing image, and locating the crops;

[0013] Randomly selecting visual monitoring points in the crop area and synchronously determining the collection direction of the visual monitoring points; the collection direction is a horizontal direction, which adopts a preset collection height and a direction with the most crops in a preset collection angle;

[0014] In the crop area, the area corresponding to the collected crops is removed to obtain an updated crop area;

[0015] Randomly selecting visual monitoring points in the updated crop area and synchronously determining the collection direction of the visual monitoring points, and performing the cycle until each crop is collected by at least one visual monitoring point to obtain a selection scheme; wherein in the random selection process, the selection probability of each position in the crop area is determined by the distance between the position and the center of the area, and the selection probability is proportional to the distance;

[0016] The cycle is performed until the number of selection schemes reaches a preset number threshold, and the selection scheme with the least visual monitoring points is selected from all selection schemes as the final scheme.

[0017] As a further scheme of the present application: the step of recording the fertilization parameters in real time, generating a collection instruction pointing to the visual monitoring point according to the fertilization parameters, and acquiring the crop image of the crop corresponding to the fertilization parameters comprises:

[0018] When the fertilization behavior is monitored, record the fertilization point, the fertilization type and the fertilization amount;

[0019] Determine the influence area according to the fertilization point, the fertilization type and the fertilization amount;

[0020] Query the crops in the influence area as the collected crops;

[0021] Query the visual monitoring point corresponding to the collected crops, generate a collection instruction pointing to the visual monitoring point, and acquire the crop image of the crop corresponding to the fertilization parameters.

[0022] As a further scheme of the present application: the step of identifying the crop image, determining the crop pulse, determining the initial detection point in real time based on the crop pulse, and performing regional extension based on the initial detection point to determine the detection area comprises:

[0023] Acquire the planting duration of the crop, and acquire a standard image according to the planting duration;

[0024] Positioning the markers in the standard image and the crop image, calibrating the standard image and the crop image according to the markers; the markers are preset image features;

[0025] Calculating the difference matrix of the standard image and the crop image based on the same difference step, traversing the two difference matrices, determining and fitting the crop vein drop;

[0026] Determining the initial detection point based on the crop vein drop, performing regional continuation based on the initial detection point, and determining the detection area;

[0027] The calculation process of the difference matrix is as follows:

[0028] ;

[0029] In the formula, the difference between the square root value of the sum of squares of n and m and b is less than a preset threshold value.

[0030] As a further scheme of the application, the step of determining the initial detection point based on the crop vein drop and performing regional continuation based on the initial detection point to determine the detection area comprises:

[0031] According to a preset direction, traverse the pixel points in the crop vein drop, and when there is no next pixel point within a preset angle range, mark the current pixel point as the initial detection point;

[0032] Taking the initial detection point as the center, calculate the color value difference of adjacent pixel points in a preset direction in sequence;

[0033] When the color value difference is less than a preset color value threshold, mark the corresponding pixel point;

[0034] Statistically determine the pixel points to determine the continuation line;

[0035] Determine the center point according to the continuation line, and determine the detection area according to the center point.

[0036] As a further scheme of the application, the step of determining the center point according to the continuation line and determining the detection area according to the center point comprises:

[0037] Obtain the pixel length of each continuation line, and determine the direction weight according to the pixel weight;

[0038] Determine the center point direction according to the direction weight, and intercept the average length of the pixel length of all continuation lines in the center point direction to determine the center point;

[0039] Determine the detection area according to the center point, and calculate the color value mean and standard deviation in the detection area in real time; wherein the area size of the detection area is an incremental value, and the incremental direction is determined by the difference value of adjacent pixel points;

[0040] Compare the color value mean and standard deviation with preset numerical conditions to determine the detection area.

[0041] As a further scheme of the present application: the detection area determined by the screening is used to obtain a region connection graph, the region connection graph is input into a trained neural network recognition model, a growth vigor evaluation report is output, a mapping relationship of fertilization parameters to the growth vigor evaluation report is constructed, and the step of evaluating the fertilization parameters comprises:

[0042] The detection area is input into a preset contour recognition model to determine the effectiveness of the detection area; the effectiveness includes invalid and non-invalid;

[0043] The non-invalid detection area is selected and input into a preset statistical graph layer to obtain a region connection graph;

[0044] The region connection graph is input into a trained neural network recognition model to output a growth vigor evaluation report;

[0045] The growth vigor evaluation reports before and after the fertilization behavior are read, the growth vigor difference is calculated as a label of the corresponding water and fertilizer parameters of the crop, a sample set is constructed, a water and fertilizer efficiency evaluation model is trained based on the sample set, and the predicted growth vigor difference of the water and fertilizer parameters is obtained.

[0046] The present application also provides an intelligent evaluation operation system for water and fertilizer utilization efficiency, the system comprising:

[0047] A monitoring point selection module is used to obtain crop distribution information and select visual monitoring points according to the crop distribution information;

[0048] A crop image acquisition module is used to record fertilization parameters in real time, generate acquisition instructions pointing to the visual monitoring points according to the fertilization parameters, and obtain crop images of crops corresponding to the fertilization parameters; the fertilization parameters include fertilization points, fertilization types and fertilization amounts;

[0049] A detection area determination module is used to identify the crop images, determine crop growth, determine initial detection points in real time based on the crop growth, and determine detection areas based on the initial detection points;

[0050] A fertilization parameter evaluation module is used to screen the determined detection areas, obtain a region connection graph, input the region connection graph into a trained neural network recognition model, output a growth vigor evaluation report, and construct a mapping relationship of fertilization parameters to the growth vigor evaluation report for evaluating the fertilization parameters.

[0051] As a further scheme of the present application: the monitoring point selection module comprises:

[0052] A crop positioning unit is used to obtain remote sensing images of crop areas, identify the remote sensing images, and locate the crops;

[0053] The direction determining unit is used for randomly selecting a visual monitoring point in the crop area and synchronously determining a collection direction of the visual monitoring point; the collection direction is a horizontal direction, which adopts a preset collection height and a direction with the most crops in a collection wide angle;

[0054] The area updating unit is used for eliminating the area corresponding to the collected crops in the crop area, so as to obtain an updated crop area.

[0055] The selection scheme generating unit is used for randomly selecting a visual monitoring point in the updated crop area and synchronously determining a collection direction of the visual monitoring point, and the process is repeatedly executed until each crop is collected by at least one visual monitoring point, so as to obtain a selection scheme; in the random selection process, the selection probability of each position in the crop area is determined by the distance between the position and the center of the area, and the selection probability is proportional to the distance.

[0056] The selection scheme screening unit is used for repeatedly executing the process until the number of selection schemes reaches a preset number threshold, selecting a selection scheme with the least visual monitoring points from all selection schemes, and taking the selection scheme as a final scheme.

[0057] As a further scheme of the present application, the crop image acquisition module comprises:

[0058] The behavior monitoring unit is used for recording a fertilization point, a fertilization type and a fertilization amount when a fertilization behavior is monitored.

[0059] The influence area determining unit is used for determining an influence area according to the fertilization point, the fertilization type and the fertilization amount.

[0060] The crop querying unit is used for querying crops in the influence area as collected crops.

[0061] The instruction sending unit is used for querying a visual monitoring point corresponding to the collected crops, generating a collection instruction pointing to the visual monitoring point, and acquiring a crop image of the crop corresponding to the fertilization parameter.

[0062] Compared with the prior art, the present application has the beneficial effects that the present application records the fertilization parameter, at the same time, acquires the crop image through the visual monitoring point, identifies the crop image, determines the growth situation before and after the fertilization, further determines the growth difference before and after the fertilization, constructs the mapping relationship from the fertilization parameter to the growth difference, and when facing the prediction demand, the growth difference of the fertilization parameter can be predicted before the fertilization behavior, and the prediction accuracy is high. BRIEF DESCRIPTION OF DRAWINGS

[0063] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are only some embodiments of the present application.

[0064] Figure 1 A flow chart of an intelligent evaluation method for water and fertilizer utilization efficiency.

[0065] Figure 2 A block diagram of the composition structure of an intelligent evaluation operation system for water and fertilizer utilization efficiency. DETAILED DESCRIPTION

[0066] In order to make the technical problems to be solved by the present application, technical solutions and beneficial effects more clear, the present application will be further described in detail below in combination with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, and are not used to limit the present application.

[0067] Figure 1 A flow chart of an intelligent evaluation method for water and fertilizer utilization efficiency, in the embodiment of the present application, an intelligent evaluation method for water and fertilizer utilization efficiency, the method comprises:

[0068] Step S100: acquiring crop distribution information, and selecting a visual monitoring point according to the crop distribution information;

[0069] Crop distribution information in a water and fertilizer management area is acquired, the crop distribution information is which crops are in each position, a visual monitoring point is selected according to the crop distribution information, a camera is installed at the visual monitoring point, and is used for shooting crops.

[0070] Step S200: recording a fertilization parameter in real time, generating a collection instruction pointing to the visual monitoring point according to the fertilization parameter, and acquiring a crop image of a crop corresponding to the fertilization parameter; the fertilization parameter comprises a fertilization point, a fertilization type and a fertilization amount;

[0071] At each time of fertilization, a fertilization parameter is recorded, the fertilization parameter is which position is fertilized with which fertilizer, and the fertilization amount is how much, that is, the fertilization parameter comprises a fertilization point, a fertilization type and a fertilization amount, a related visual monitoring point is located according to the fertilization parameter, a collection instruction pointing to the visual monitoring point is generated, and an image collection device at the visual monitoring point can acquire a crop image; it is worth mentioning that the determination process of the related visual monitoring point is that a crop affected by the fertilization parameter is determined, and then a visual monitoring point capable of shooting the crop is acquired as the related visual monitoring point.

[0072] Step S300: identifying the crop image, determining a crop pulse, determining an initial detection point in real time based on the crop pulse, and determining a detection area based on the initial detection point;

[0073] The obtained crop image is identified, and the crop vein drop can be determined; the crop vein drop is an image feature corresponding to the stem in the growth process of the crop; generally, the crop vein drop in the flowering period and the fruiting period is easier to obtain; the crop vein drop is a standard tree structure, and the tail of each branch is used as a preliminary detection point, and then the preliminary detection point is used as the center to perform regional extension, so as to obtain a detection area; the detection area generally corresponds to a certain stage of the fruit, such as the image area corresponding to the flower in the flowering period, or the image area corresponding to the fruit in the fruiting period.

[0074] Step S400: screening the determined detection area to obtain a region connection graph, inputting the region connection graph into a trained neural network recognition model, outputting a growth vigor evaluation report, and constructing a mapping relationship of fertilization parameters to the growth vigor evaluation report for evaluating the fertilization parameters;

[0075] The regional extension process may have errors and the accuracy is low, at this time, the detection area needs to be screened first, and then the screened detection area is connected to obtain a region connection graph; the region connection graph is a graph structure, which corresponds to the distribution of the fruit; similarly, if it is the flowering period, it is the distribution of the flower, and if it is the fruiting period, it is the distribution of the fruit; a sample set is created by a worker in advance, a mapping model (neural network recognition model) of the region connection graph to the growth vigor evaluation report is established, and when the region connection graph is obtained, the region connection graph is input into the mapping model to obtain the growth vigor evaluation report; it should be noted that the environmental data and the crop image both have a time label, and each analysis process is performed under the same time correspondence.

[0076] Regarding step S100, the step of obtaining crop distribution information and selecting a visual monitoring point according to the crop distribution information includes:

[0077] Obtaining a remote sensing image of the crop area, identifying the remote sensing image, and locating the crop;

[0078] Randomly selecting a visual monitoring point in the crop area and synchronously determining the collection direction of the visual monitoring point; the collection direction is a horizontal direction, which adopts a preset collection height and a direction with the most crops in a preset collection angle;

[0079] In the crop area, the area corresponding to the crops that have been collected is removed to obtain an updated crop area;

[0080] Randomly selecting a visual monitoring point in the updated crop area and synchronously determining the collection direction of the visual monitoring point, and repeatedly performing until each crop is collected by at least one visual monitoring point to obtain a selection scheme; in the random selection process, the selection probability of each position in the crop area is determined by the distance between the position and the center of the region, and the selection probability is proportional to the distance;

[0081] The selecting is performed repeatedly until the number of the selected schemes reaches a preset number threshold, and a selected scheme with the least visual monitoring points is selected as a final scheme.

[0082] In an example of the technical scheme of the present application, the selection process of the visual monitoring points is described. Since the distribution of crops is generally regular, such as planting at equal intervals in rows, it is of course feasible to set visual monitoring points artificially, such as two in a row. However, this method is not necessarily the best way, and its regularity is relatively strong, but the cost utilization rate may be low. In fact, only a limited number of cameras that can capture all crops are needed.

[0083] The combination of cameras that can capture all crops is very large, and there are many feasible schemes. How to obtain the images of all crops with the least visual monitoring points is a very complex mathematical problem that cannot be solved by those skilled in the art. In fact, since the crops are not necessarily planted at strict equal intervals, the interval may change during growth. It is not very realistic to solve it as a mathematical problem. Therefore, the above content provides a relatively simple and relatively optimal scheme selection process, which is as follows:

[0084] A remote sensing image of the crop area is obtained, the remote sensing image is identified, the crops are located, visual monitoring points are randomly selected in the crop area, and the collection direction of the visual monitoring points is determined synchronously. The collection direction is a horizontal direction, and the height is a preset value, which is generally set at a height that can obtain the overall appearance of the crops. The wide angle of the collection device at the visual monitoring point is also fixed, while the horizontal direction is a very wide range, which can be set within a 360-degree range. The specific angle of the technical scheme of the present application is determined as follows: a direction with the most crops in a preset collection height and collection wide angle is used.

[0085] After the visual monitoring points are selected and the collection direction is determined, the crops that can be collected by the visual monitoring points are queried, and the areas corresponding to the crops that have been collected in the crop area are removed (generally, an equidistant entity is created based on the contour of the crops according to a preset length, and of course, the circumscribed circle of the contour of the crops can also be obtained, which is determined by the staff independently). An updated crop area is obtained. For the updated crop area, the determination process of the visual monitoring points and the collection direction is performed again based on it, and the crop area is updated again. When each crop is collected by at least one visual monitoring point, all selected visual monitoring points are counted as a selected scheme.

[0086] The above content is regarded as a large cycle, and a preset number of selection schemes are obtained by performing the cycle a preset number of times. In these selection schemes, the selection scheme with the least visual monitoring point is selected as the final scheme. This is the optimization process, and the obtained scheme is a better scheme, but not necessarily the best scheme. In fact, although it is not necessarily the best scheme, it is good enough relative to the cost input, and the cost utilization rate is very high.

[0087] In step S200, the real-time recording of the fertilization parameter, the generation of the acquisition instruction pointing to the visual monitoring point based on the fertilization parameter, and the acquisition of the crop image of the crop corresponding to the fertilization parameter include the following steps:

[0088] When the fertilization behavior is monitored, the fertilization point, the fertilization type, and the fertilization amount are recorded;

[0089] The influence area is determined according to the fertilization point, the fertilization type, and the fertilization amount;

[0090] The crops in the influence area are queried as the acquisition crops;

[0091] The visual monitoring point corresponding to the acquisition crops is queried, the acquisition instruction pointing to the visual monitoring point is generated, and the crop image of the crop corresponding to the fertilization parameter is acquired.

[0092] The fertilization behavior is monitored in real time, and when the fertilization behavior is monitored, the fertilization point, the fertilization type, and the fertilization amount are recorded. In fact, in the existing fertilization management architecture, the fertilization behavior is directly uploaded by the staff. When the fertilization behavior uploaded by the staff is received, it is considered that the fertilization behavior is monitored. The influence area is determined according to the fertilization point, the fertilization type, and the fertilization amount. The specific process is to determine a radius according to the fertilization type and the fertilization amount, and to create a circular area with the fertilization point as the center, which is called the influence area. The crops in the influence area are queried as the acquisition crops, the visual monitoring point corresponding to the acquisition crops is queried, the acquisition instruction pointing to the visual monitoring point is generated, and the crop image of the crop corresponding to the fertilization parameter is acquired.

[0093] Regarding step S300, the crop image is identified, the crop pulse is determined, the initial detection point is determined in real time based on the crop pulse, and the detection area is determined based on the region extension of the initial detection point.

[0094] The planting duration of the crop is acquired, and the standard image is acquired according to the planting duration;

[0095] The markers in the standard image and the crop image are located, and the standard image and the crop image are calibrated according to the markers; the markers are preset image features;

[0096] The difference matrices of the standard image and the crop image are calculated based on the same difference step, and the two difference matrices are traversed to determine and fit the crop pulse;

[0097] Based on the crop pulse real-time determination of the initial detection point, based on the initial detection point, regional extension is carried out, and the detection area is determined.

[0098] In an example of the technical scheme of the present application, the planting time of the crop is obtained, and the standard image is obtained according to the planting time. The standard image belongs to known data. The standard image and the crop image are positioned. The standard image and the crop image are calibrated according to the mark. The mark is generally an object related to the crop, such as soil or a limiting frame (even soil). The calibration process is only to calibrate the color value, which is used to convert the crop image into a crop image under standard lighting conditions (the lighting conditions of the standard image are the default standard lighting conditions).

[0099] The same calculation process is used for the standard image and the crop image. The calculation process is to calculate the difference between the pixels of the adjacent step. The difference reflects the difference degree of each pixel point and the surrounding pixel points. The pixel point with larger difference degree is selected, that is, the crop pulse can be obtained. The description of the larger difference degree needs to be determined by all difference values (difference matrix), such as the mean value of all difference values. The standard image and the crop image are analyzed at the same time, two results can be obtained, which can be verified with each other. The two results are fitted, and the final crop pulse is obtained.

[0100] Finally, the crop pulse is analyzed, and the detection area is determined.

[0101] The calculation process of the difference matrix is as follows:

[0102] ;

[0103] In the formula, the difference between the square root of the sum of squares of n and m and b is less than a predetermined threshold.

[0104] In the above content, And Actually contained in They are calculated separately as boundary values. In the difference matrix obtained from the above process, each value represents the maximum difference between the corresponding pixel point and the surrounding pixel points, which can be compared with the concept of gradient.

[0105] Further, the step of determining the initial detection point based on the crop pulse in real time, and the step of regional extension based on the initial detection point to determine the detection area include:

[0106] According to the preset direction, the pixel points in the crop pulse are traversed. When there is no next pixel point in the preset angle range, the current pixel point is marked as the initial detection point.

[0107] According to a preset direction, pixel points in the crop vein are read in sequence, and the reading process is as follows: taking the preset direction as a center line, an angle range is determined, when there is a pixel point in the angle range, the next pixel point is read, when there is no pixel point in the angle range, the end point is taken as a primary detection point;

[0108] Taking the primary detection point as a center, color value differences of adjacent pixel points are sequentially calculated in a preset direction;

[0109] This direction is different from the traversal direction, and the traversal direction is generally a large direction, such as from the bottom to the top; and the direction in the color value difference calculation process is a direction group, for example, a direction group is determined with an angle interval of 30 degrees, and color value differences of adjacent pixel points need to be calculated in all directions in the direction group.

[0110] When the color value difference is less than a preset color value threshold, the corresponding pixel point is marked;

[0111] The pixel points are counted to determine an extension line;

[0112] When the color value difference is small, the corresponding pixel point is marked; according to the direction, the pixel points marked can be obtained. The extension line can be obtained.

[0113] A center point is determined according to the extension line, and a detection area is determined according to the center point.

[0114] The extension line is analyzed to determine the center point, and the detection area can be obtained by taking the center point as a starting point and performing an extension process.

[0115] Specifically, the step of determining the center point according to the extension line and determining the detection area according to the center point comprises:

[0116] The pixel length of each extension line is obtained, and a direction weight is determined according to the pixel weight;

[0117] A center point direction is determined according to the direction weight, and a mean length of the pixel length of all extension lines is intercepted in the center point direction to determine the center point;

[0118] A detection area is determined according to the center point, and a color value mean and a standard deviation in the detection area are calculated in real time; wherein, the area size of the detection area is an incremental value, and the incremental direction is determined by the difference value of adjacent pixel points;

[0119] The above provides a center point determination process, which includes two stages, i.e., determining a direction and determining a position; specifically, the pixel lengths of the various extension lines are calculated first, and by comparing the pixel lengths, the influence amplitude of each extension line on the final center point can be determined, which is represented by a direction weight; according to the direction weight, the direction can be determined on one hand, and a mean length can be determined on the other hand; the mean length is intercepted in the determined direction, and the center point can be obtained.

[0120] According to the center point, the color value mean and standard deviation in the detection area are calculated in real time; wherein the area size of the detection area is an incremental value, and the incremental direction is determined by the difference value of adjacent pixel points.

[0121] Taking the center point as the starting point, the area is continuously expanded around, and when the difference value of adjacent pixel points is less than a preset threshold, the extension is performed in the direction.

[0122] The color value mean and standard deviation are compared with a preset numerical condition to determine the detection area.

[0123] The purpose of determining the detection area is to make the detection area correspond to the fruit (including flowers), so that the color values of the pixel points in the most appropriate detection area should be similar, i.e., the standard deviation is small; based on this, the color value mean and standard deviation are continuously calculated in the extension process, and according to the comparison between the standard deviation and the preset numerical condition, a final detection area can be determined.

[0124] Regarding step S400, the detection area determined by screening is obtained, the region connection graph is input into the trained neural network recognition model, the growth potential evaluation report is output, the mapping relationship between the fertilization parameters and the growth potential evaluation report is constructed, and the step of evaluating the fertilization parameters includes:

[0125] The detection area is input into a preset contour recognition model to determine the effectiveness of the detection area; the effectiveness includes invalid and non-invalid;

[0126] The generation process of the detection area is an extension process, and the shape is not limited, therefore, the generated detection area is likely to have invalid detection areas, especially in the extension process starting from the initial detection point, which is likely to detect the entire vein drop against the vein drop direction, at this time, the detection area is invalid; the shape of the effective detection area is limited to several types, such as flowers and fruits.

[0127] The non-invalid detection area is selected and input into a preset statistical layer to obtain a region connection graph.

[0128] The non-invalid detection area is selected and connected, and input into a preset statistical layer to obtain a connection graph in the form of a layer.

[0129] Input the region connection diagram into the trained neural network recognition model, and output a growth vigor evaluation report;

[0130] With the existing neural network recognition model, a mapping relationship from the region connection diagram to the growth vigor evaluation report can be established, and when the region connection diagram is obtained, the growth vigor evaluation report can be directly output.

[0131] The growth vigor evaluation reports before and after the fertilization behavior are read, the growth vigor difference is calculated as the label of the corresponding water and fertilizer parameters of the crop, a sample set is constructed, and a water and fertilizer efficiency evaluation model is trained based on the sample set, which is used to obtain the predicted growth vigor difference of the water and fertilizer parameters.

[0132] Since the ultimate goal of the technical scheme of the present application is to obtain the relationship between the fertilization process and the growth vigor, the water and fertilizer parameters are used as features, the growth vigor evaluation report is used as a label, a sample set is constructed, and a water and fertilizer efficiency evaluation model is trained based on the sample set, which is used to obtain the predicted growth vigor difference of the water and fertilizer parameters.

[0133] It should be noted that the fertilization behavior occurs between time A and time B, and it corresponds to the difference between time A and time B. Because the samples in the sample set are the water and fertilizer parameters of the fertilization behavior and the growth vigor difference before and after the fertilization behavior, in addition, the growth vigor difference actually includes the growth vigor difference of different crop stages, and the water and fertilizer efficiency evaluation model is essentially a clustering model that clusters all growth vigor differences corresponding to each water and fertilizer parameter into a class.

[0134] Figure 2 The composition structure block diagram of the intelligent evaluation operation system of water and fertilizer utilization efficiency, in the embodiment of the present application, an intelligent evaluation operation system of water and fertilizer utilization efficiency, the system 10 comprises:

[0135] A monitoring point selection module 11 is used to obtain crop distribution information and select visual monitoring points according to the crop distribution information.

[0136] A crop image acquisition module 12 is used to record fertilization parameters in real time, generate collection instructions pointing to visual monitoring points according to the fertilization parameters, and obtain crop images of crops corresponding to the fertilization parameters; the fertilization parameters include fertilization points, fertilization types and fertilization amounts.

[0137] A detection area determination module 13 is used to identify the crop image, determine crop veins, determine initial detection points in real time based on the crop veins, and determine a detection area based on the initial detection points.

[0138] The fertilization parameter evaluation module 14 is configured to screen the determined detection area, obtain a region connection graph, input the region connection graph into a trained neural network identification model, output a growth vigor evaluation report, and construct a mapping relationship of the fertilization parameter to the growth vigor evaluation report, so as to evaluate the fertilization parameter.

[0139] Further, the monitoring point selection module 11 comprises:

[0140] The crop positioning unit is configured to obtain a remote sensing image of the crop area, identify the remote sensing image, and locate the crops.

[0141] The direction determination unit is configured to randomly select a visual monitoring point in the crop area and synchronously determine the collection direction of the visual monitoring point; the collection direction is a horizontal direction, which adopts a preset collection height and a direction with the most crops in a preset collection wide angle.

[0142] The region updating unit is configured to remove the region corresponding to the collected crops in the crop area, and obtain an updated crop area.

[0143] The selection scheme generation unit is configured to randomly select a visual monitoring point in the updated crop area, synchronously determine the collection direction of the visual monitoring point, and cyclically execute until each crop is collected by at least one visual monitoring point, to obtain a selection scheme; in the random selection process, the selection probability of each position in the crop area is determined by the distance between the position and the center of the region, and the selection probability is proportional to the distance.

[0144] The selection scheme screening unit is configured to cyclically execute until the number of selection schemes reaches a preset number threshold, select a selection scheme with the least visual monitoring points from all selection schemes, and take the selection scheme as a final scheme.

[0145] Specifically, the crop image acquisition module 12 comprises:

[0146] The behavior monitoring unit is configured to record the fertilization point, the fertilization type and the fertilization amount when the fertilization behavior is monitored.

[0147] The influence region determination unit is configured to determine the influence region according to the fertilization point, the fertilization type and the fertilization amount.

[0148] The crop query unit is configured to query the crops in the influence region as the collected crops.

[0149] The instruction sending unit is configured to query the visual monitoring point corresponding to the collected crops, generate a collection instruction pointing to the visual monitoring point, and obtain the crop image of the crop corresponding to the fertilization parameter.

[0150] The above merely describes preferred embodiments of the present application, and is not used to limit the present application, any modification, equivalent replacement and improvement within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. An intelligent evaluation method for water and fertilizer utilization efficiency, characterized in that, The method comprises: acquiring crop distribution information, and selecting a visual monitoring point according to the crop distribution information; real-time recording of fertilization parameters, generation of a collection instruction pointing to the visual monitoring point according to the fertilization parameters, and acquisition of a crop image of the crop corresponding to the fertilization parameters; the fertilization parameters include a fertilization point, a fertilization type, and a fertilization amount; identifying the crop image, determining a crop ramification, determining an initial detection point in real time based on the crop ramification, and performing regional extension based on the initial detection point to determine a detection area; wherein the crop ramification is a standard tree structure, the tail of each branch is used as the initial detection point, and then the initial detection point is used as the center to perform regional extension to obtain the detection area; screening the determined detection area to obtain a regional connection graph, inputting the regional connection graph into a trained neural network recognition model to output a growth vigor evaluation report, and constructing a mapping relationship from the fertilization parameters to the growth vigor evaluation report for evaluating the fertilization parameters; The step of identifying the crop image, determining a crop ramification, determining an initial detection point in real time based on the crop ramification, and performing regional extension based on the initial detection point to determine a detection area comprises: acquiring a planting duration of the crop, and acquiring a standard image according to the planting duration; positioning an identifier in the standard image and the crop image, and calibrating the standard image and the crop image according to the identifier; the identifier is a preset image feature; calculating difference matrices of the standard image and the crop image based on the same difference step, traversing the two difference matrices to determine and fit the crop ramification; determining an initial detection point in real time based on the crop ramification, and performing regional extension based on the initial detection point to determine a detection area; wherein the calculation process of the difference matrix is: ； wherein the difference between the square root of the sum of squares of n and m and b is less than a preset threshold value; The step of screening the determined detection area to obtain a regional connection graph, inputting the regional connection graph into a trained neural network recognition model to output a growth vigor evaluation report, and constructing a mapping relationship from the fertilization parameters to the growth vigor evaluation report for evaluating the fertilization parameters comprises: inputting the detection area into a preset contour recognition model to determine the effectiveness of the detection area; the effectiveness includes invalid and non-invalid; selecting a non-invalid detection area and inputting it into a preset statistical layer to obtain a regional connection graph; inputting the regional connection graph into a trained neural network recognition model to output a growth vigor evaluation report; using the growth vigor evaluation report as a label of water and fertilizer parameters, constructing a sample set, training a water and fertilizer efficiency evaluation model based on the sample set, and using the model to obtain a predicted growth vigor evaluation report of the water and fertilizer parameters.

2. The intelligent evaluation method of water and fertilizer utilization efficiency according to claim 1, characterized in that, The step of acquiring crop distribution information and selecting a visual monitoring point according to the crop distribution information comprises: acquiring a remote sensing image of a crop area, identifying the remote sensing image, and positioning the crop; randomly selecting a visual monitoring point in the crop area and synchronously determining a collection direction of the visual monitoring point; the collection direction is a horizontal direction, which adopts a preset collection height and a direction with the most crops in a preset collection angle; removing the area corresponding to the collected crops in the crop area to obtain an updated crop area; Randomly selecting visual monitoring points in the updated crop area, synchronously determining the collection direction of the visual monitoring points, and cyclically executing until each crop is collected by at least one visual monitoring point, to obtain a selection scheme; wherein in the random selection process, the selection probability of each position in the crop area is determined by the distance between the position and the center of the area, and the selection probability is directly proportional to the distance; Cyclically executing until the number of selection schemes reaches a preset number threshold, and selecting the selection scheme with the least visual monitoring points from all the selection schemes as the final scheme.

3. The intelligent evaluation method of water and fertilizer utilization efficiency according to claim 1, characterized in that, The step of recording the fertilization parameters in real time, generating the collection instruction pointing to the visual monitoring point according to the fertilization parameters, and obtaining the crop image of the crop corresponding to the fertilization parameters comprises: When the fertilization behavior is monitored, recording the fertilization point, the fertilization type and the fertilization amount; Determining the influence area according to the fertilization point, the fertilization type and the fertilization amount; Querying the crop in the influence area as the collection crop; Querying the visual monitoring point corresponding to the collection crop, generating the collection instruction pointing to the visual monitoring point, and obtaining the crop image of the crop corresponding to the fertilization parameters.

4. The intelligent evaluation method of water and fertilizer utilization efficiency according to claim 1, characterized in that, The step of determining the initial detection point based on the crop vein in real time, and performing regional extension based on the initial detection point to determine the detection area comprises: Traversing the pixel points in the crop vein according to a preset traversal direction, and marking the current pixel point as the initial detection point when there is no next pixel point within a preset angle range; Calculating the color value difference of the adjacent pixel points in a preset calculation direction with the initial detection point as the center; When the color value difference is less than a preset color value threshold, marking the corresponding pixel point; Counting the pixel points to determine the extension line; Determining the center point according to the extension line, and determining the detection area according to the center point.

5. The intelligent evaluation method of water and fertilizer utilization efficiency according to claim 4, characterized in that, The step of determining the center point according to the extension line, and determining the detection area according to the center point comprises: Obtaining the pixel length of each extension line, and determining the direction weight according to the pixel length; Determining the center point direction according to the direction weight, and determining the center point by intercepting the average length of the pixel length of all extension lines in the center point direction; Determining the detection area according to the center point, and calculating the color value mean and standard deviation in the detection area in real time; wherein the area size of the detection area is an incremental value, and the incremental direction is determined by the difference value of the adjacent pixel points; Comparing the color value mean and standard deviation with a preset numerical condition to determine the detection area.

6. An intelligent evaluation operation system for water and fertilizer utilization efficiency, characterized in that, The system comprises: A monitoring point selection module for obtaining crop distribution information and selecting visual monitoring points according to the crop distribution information; A crop image acquisition module for recording fertilization parameters in real time, generating a collection instruction pointing to a visual monitoring point according to the fertilization parameters, and obtaining a crop image of a crop corresponding to the fertilization parameters; the fertilization parameters include a fertilization point, a fertilization type and a fertilization amount; A detection area determination module for identifying the crop image, determining a crop vein, determining an initial detection point based on the crop vein in real time, performing regional extension based on the initial detection point, and determining a detection area; wherein the crop vein is a standard tree structure, the tail of each branch is taken as the initial detection point, and then the initial detection point is taken as the center to perform regional extension to obtain the detection area. The fertilization parameter evaluation module is configured to screen the determined detection area, obtain a region connection graph, input the region connection graph into a trained neural network recognition model, output a growth vigor evaluation report, and construct a mapping relationship between the fertilization parameter and the growth vigor evaluation report, so as to evaluate the fertilization parameter. The step of identifying the crop image, determining a crop vein drop, determining an initial detection point in real time based on the crop vein drop, and determining a detection area based on the initial detection point comprises: acquiring a planting time length of the crop, and acquiring a standard image according to the planting time length; positioning an identifier in the standard image and the crop image, and calibrating the standard image and the crop image according to the identifier; the identifier is a preset image feature; calculating a difference matrix of the standard image and the crop image based on a same difference step, traversing the two difference matrices, and determining and fitting the crop vein drop; determining an initial detection point in real time based on the crop vein drop, and determining a detection area based on the initial detection point; wherein, the calculation process of the difference matrix is: ； wherein, the difference between the square root of the sum of squares of n and m and b is less than a preset threshold value; The step of screening the determined detection area, obtaining a region connection graph, inputting the region connection graph into a trained neural network recognition model, outputting a growth vigor evaluation report, and constructing a mapping relationship between the fertilization parameter and the growth vigor evaluation report, so as to evaluate the fertilization parameter comprises: inputting the detection area into a preset contour recognition model to determine the effectiveness of the detection area; the effectiveness includes invalid and non-invalid; selecting the non-invalid detection area and inputting it into a preset statistical layer to obtain a region connection graph; inputting the region connection graph into a trained neural network recognition model to output a growth vigor evaluation report; using the growth vigor evaluation report as a label of the water and fertilizer parameter, constructing a sample set, training a water and fertilizer efficiency evaluation model based on the sample set, and obtaining a predicted growth vigor evaluation report of the water and fertilizer parameter. 7.The intelligent evaluation system for water and fertilizer utilization efficiency according to claim 6, characterized in that, The monitoring point selection module comprises: a crop positioning unit configured to acquire a remote sensing image of a crop area, identify the remote sensing image, and position the crop; a direction determination unit configured to randomly select a visual monitoring point in the crop area and synchronously determine a collection direction of the visual monitoring point; the collection direction is a horizontal direction, which adopts a preset collection height and a direction in which the number of crops in a preset collection angle is the largest; a region updating unit configured to remove a region corresponding to a crop that has been collected in the crop area to obtain an updated crop area; a selection scheme generation unit configured to randomly select a visual monitoring point in the updated crop area and synchronously determine a collection direction of the visual monitoring point, and repeatedly execute until each crop is collected by at least one visual monitoring point to obtain a selection scheme; wherein, in the random selection process, the selection probability of each position in the crop area is determined by the distance between the position and the center of the region, and the selection probability is proportional to the distance; a selection scheme screening unit configured to repeatedly execute until the number of selection schemes reaches a preset number threshold, select a selection scheme in which the number of visual monitoring points is the least from all the selection schemes, and use the selection scheme as a final scheme. 8.The intelligent evaluation system for water and fertilizer utilization efficiency according to claim 6, characterized in that, The crop image acquisition module comprises: a behavior monitoring unit configured to record a fertilization point, a fertilization type, and a fertilization amount when a fertilization behavior is monitored. An influence area determining unit is configured to determine an influence area according to a fertilization point, a fertilization type, and a fertilization amount; A crop querying unit is configured to query crops in the influence area as the crops to be collected; An instruction sending unit is configured to query a visual monitoring point corresponding to the crops to be collected, generate a collection instruction pointing to the visual monitoring point, and acquire a crop image of the crops corresponding to the fertilization parameters.

Images