A method and device for evaluating corn sowing quality

Abstract

The present application discloses a method and device for evaluating corn sowing quality. The method for evaluating corn sowing quality includes: driving a drone equipped with an image acquisition device to cruise along a target route to collect images of a corn sowing area to obtain original image data; preprocessing the original image data to obtain binary image data; and evaluating the corn sowing quality of the corn sowing area based on the binary image data. The present application can provide a real-time, efficient, refined, and objective evaluation of the quality of third-party sowing operations, realize effective supervision of third-party sowing operations, and can also analyze and determine the basic seedling conditions of a plot, providing a basis for subsequent field management inputs.

Description

Technical Field

[0001] The present application belongs to the technical field of computer data processing, and specifically relates to a method and device for evaluating corn sowing quality. Background Art

[0002] Corn is the largest crop in my country, ranking first in both planting area and total output, and plays an irreplaceable role in national food security. Domestic and foreign studies have shown that increasing planting density is the main technical approach to increase corn production, and precise application of water and fertilizer according to actual planting density is also the key to achieving efficient production of corn resources. Therefore, sowing density is a widely concerned issue in current corn production. Factors affecting sowing density involve many aspects, including climatic conditions, seed quality, soil moisture, sowing quality, and pests and diseases. The emergence density of corn often fails to reach the expected sowing density.

[0003] The climate and production methods in different regions determine the suitable window period for local corn sowing. Under the conditions of large-scale production, completing corn sowing within a limited window period places extremely high demands on the scale of agricultural machinery, the team of operators and the management level of growers. In recent years, with the gradual scale of corn planting, hiring a large number of third-party operators and machines during the peak sowing period has become a common method for production. However, the adjustment status of the machine tools and the operator's operation level are often uneven, which has a great impact on the sowing quality of corn; at the same time, in order to catch up with the farming season and strive for the greatest economic benefits, the hired operators will increase the speed of the machine tools, resulting in seed bouncing, uneven seed placement, inconsistent sowing depth, and even missed sowing and re-sowing, which leads to seedling shortages and broken ridges, reducing the sowing quality. However, due to the lack of effective supervision methods, the sowing quality of third-party operators and machines is difficult to guarantee, which has become the main factor affecting the corn planting density in large-scale production fields. How to efficiently and accurately evaluate the sowing quality of third-party operators, set remuneration according to seedlings, and avoid affecting the sowing density due to problems with machines or operators has become an important production issue to improve sowing quality and lay a key foundation for high grain yields throughout the year.

[0004] Farmers usually obtain the number of corn plants by manual counting during the corn seedling stage, and use this as the basis for evaluating the quality of the work of hired machines and operators, and decide whether to pay the full operating fees of the machines and operators. However, for large-scale field planting, the workload of corn plant counting is large and inefficient, which is not conducive to timely evaluation and will also cause disputes due to human judgment. In production, there is an urgent need for a real-time, efficient, accurate and objective sowing quality evaluation method to supervise the work of hired operators and improve sowing quality.

[0005] With the rapid development of drone and sensor technology, drone platforms equipped with various sensors can obtain high-resolution images of corn populations in a timely and effective manner, and use image feature extraction methods and machine learning algorithms to analyze the number and distribution of corn plants. Through the sowing quality evaluation algorithm, key data such as corn seedling plant density, emergence rate, plant spacing, group uniformity and sowing quality can be obtained. Compared with manual surveys, drone images can quickly and objectively evaluate corn seedling plant density. Based on high-resolution images obtained by drones, large-scale, rapid and accurate evaluation of corn sowing quality can be achieved. Summary of the invention

[0006] The present application aims to provide a method for evaluating corn sowing quality, the evaluation method comprising:

[0007] A drone equipped with an image acquisition device is driven to cruise along a target route to collect images of a corn planting area to obtain original image data;

[0008] Preprocessing the original image data to obtain binary image data;

[0009] The corn sowing quality of the corn sowing area is evaluated according to the binary image data.

[0010] Furthermore, the method of driving a drone equipped with an image acquisition device to cruise along a target route to acquire images of a corn planting area to obtain raw image data includes:

[0011] Driving a drone equipped with an image acquisition device to cruise and acquire images of the corn planting area at an angle perpendicular to the horizontal direction and along a target route parallel to the planting rows of the corn planting area;

[0012] During the cruise to collect images of the corn planting area, the cruise altitude of the drone is limited according to the following formula:

[0013]

[0014] In the process of cruising and collecting images of the corn planting area, the flight speed of the UAV and the exposure speed of the image acquisition device are limited according to the following formula:

[0015]

[0016] Among them, h is the cruising altitude, f is the focal length of the lens of the image acquisition device, GSD is the ground resolution, and the ground resolution is limited to be less than or equal to 0.005, α is the pixel size of the image acquisition device; δ is the phase point displacement of the UAV during the cruising process, and the phase point displacement is limited to be less than or equal to 0.2, V is the flight speed, and t is the exposure speed.

[0017] Furthermore, the target route is planned in the following manner:

[0018] Obtaining the boundary and sowing row direction of the corn sowing area;

[0019] Within the boundary, along the direction of the sowing row, planning the route of the drone so that the drone randomly samples 5 to 20 monitoring areas within the corn sowing area, and patrols each of the monitoring areas one by one by patrolling once in each of the monitoring areas;

[0020] The route of the UAV is planned to ensure that the cruising length is greater than or equal to 30 meters, the row overlap rate in the original image data is less than or equal to 50%, and the inter-row overlap rate in the original image data is 0%.

[0021] Further, the obtaining of the boundary and the direction of the corn sowing area includes:

[0022] receiving a first input from an operator;

[0023] In response to the first input, obtaining the boundary and sowing row direction of the corn sowing area;

[0024] or,

[0025] Reading the track file of the agricultural machinery sowing in the corn sowing area;

[0026] According to the track file, the boundary and the direction of the corn planting area are obtained.

[0027] Furthermore, the preprocessing of the original image data to obtain the binary image data includes:

[0028] Using the EGI indicator, performing histogram equalization on the original image data to determine the segmentation threshold;

[0029] Using a threshold segmentation and binarization method, the original image data is subjected to a first preprocessing according to a relationship between the RGB color space of the original image data and the segmentation threshold, to obtain first binarized image data;

[0030] In order to distinguish the boundary between the corn seedlings and the land in the first binarized image data, the binarization follows the following formula:

[0031]

[0032] Among them, Green threshold is the segmentation threshold.

[0033] Further, after obtaining the first binarized image data, the preprocessing of the original image data to obtain the binarized image data further includes:

[0034] Applying morphological opening operation to smooth the outline of the corn seedling in the binary image data, cut off narrow discontinuous points, and eliminate small protrusions;

[0035] Applying a closing operation to fill narrow gaps within the binary image data;

[0036] The corn seedlings are subjected to noise removal by combining the same number of opening operations and closing operations to obtain second binary image data.

[0037] Furthermore, the step of evaluating the corn sowing quality of the corn sowing area according to the binary image data includes:

[0038] Obtaining the sowing row direction in the binary image data through Hough transform;

[0039] Applying the nearest neighbor interpolation method, rotating the binary image data so that the sowing row direction is parallel to the y-axis of the two-dimensional plane rectangular coordinate system;

[0040] Calculating the sum of pixels of each corn seedling in the binary image data, and generating a pixel statistical map for all corn seedlings, wherein a local peak in the pixel statistical map corresponds to a corn seedling, and identifying each sowing row and / or each corn seedling in the binary image data by identifying each local peak in the pixel statistical map;

[0041] The corn sowing quality of the corn sowing area is evaluated according to each sowing row and / or each corn seedling in the binary image data.

[0042] Further, in the process of identifying each sowing row and / or each corn seedling in the binary image data by identifying each local peak in the pixel statistical map, the peaks whose peaks are lower than the target threshold value in the pixel statistical map are ignored, and when there are multiple peaks in the pixel statistical map and the number of the peaks is greater than the number of planting rows, the row peaks are determined by the following two formulas:

[0043]

[0044] Among them, Min pd is the distance between two adjacent peaks, and Min pd The number of pixels greater than 50% of the nominal crop row spacing, Row s is the theoretical row spacing during planting, Resolution is the calculated image resolution result;

[0045] Min ph =MaxSum g *Row ps ;

[0046] Among them, Min ph is the peak value, MaxSum g is the maximum value of the sum of pixels with a pixel value of 1 in the binary image data, Row ps is the maximum peak height fraction of the corresponding row.

[0047] Furthermore, the corn sowing quality in the corn sowing area is evaluated, including:

[0048] The uniformity of the row spacing or plant spacing in the corn sowing area is calculated by the following formula:

[0049]

[0050] in, is the average value of row spacing or plant spacing, sd is the standard deviation of row spacing or plant spacing, UI is the uniformity of the row spacing or plant spacing, and when UI≥5, it is judged that the uniformity of the corn sowing area is qualified;

[0051] The sowing density of the corn sowing area is calculated by the following formula:

[0052]

[0053] Where A is the area of ​​the corn planting area, D o For the sowing density, the median of the peak spacing was taken as the actual crop spacing (D plant ); the pixel column where the peak is located is used as the center line of the crop row; the median of the difference between the center line positions of every two rows between the peak values ​​is regarded as the actual crop row spacing (D row );

[0054] The emergence rate of the corn sowing area is calculated by the following formula:

[0055]

[0056] Among them, D ais the number of seedlings per unit area in the corn planting area, D o is the sowing density, ER is the emergence rate;

[0057] The actual germination rate of the corn sowing area is calculated by the following formula:

[0058]

[0059] Among them, N s is the number of germinated grains in the plot of corn sowing area, N p is the actual number of sown kernels in the corn sowing area, and GRα is the actual germination rate;

[0060] The germination rate deviation of the corn sowing area is calculated by the following formula:

[0061]

[0062] Wherein, GR is the germination rate of the field seeds in the corn sowing area obtained by measurement, and D gr If the soil moisture or ground temperature in the corn planting area is less than or equal to 5%, it is determined that the soil moisture or ground temperature in the corn planting area is suitable; otherwise, it is determined that the soil moisture or ground temperature in the corn planting area is unsuitable;

[0063] The theoretical emergence rate of the corn sowing area is calculated by the following formula:

[0064]

[0065] Among them, N e is the number of unearthed seedlings in the corn planting area, N s is the actual number of germinated kernels in the corn sowing area, ER t is the theoretical emergence rate;

[0066] The actual emergence rate of the corn sowing area is calculated by the following formula:

[0067]

[0068] Among them, N e is the number of seedlings unearthed in the survey, N p is the actual number of seeds sown on the plot, ER a is the actual emergence rate;

[0069] The emergence rate deviation of the corn sowing area is calculated by the following formula:

[0070]

[0071] Among them, ER t is the theoretical emergence rate, ERa is the actual emergence rate, D er is the germination rate deviation, in D er If the soil tillage quality of the corn sowing area is less than or equal to 5%, the soil tillage quality of the corn sowing area is determined to be qualified; otherwise, the soil tillage quality of the corn sowing area is determined to be unqualified;

[0072] The sowing quality index of the corn sowing area is calculated by the following formula:

[0073]

[0074] Among them, ER a is the actual emergence rate, ER is the emergence rate, GR is the germination rate of the seeds used in the field, and F sq is the sowing quality index, in F sq ≥0, the sowing quality index of the corn sowing area is determined to be unqualified, otherwise the sowing quality index of the corn sowing area is determined to be qualified.

[0075] The present application also provides a device for evaluating corn sowing quality, the device comprising a processor, the processor being used to implement the steps of the evaluation method described in any technical solution of the present application when executing a computer program stored in a memory.

[0076] The beneficial effects of the present application are as follows: the present application uses the high-definition images acquired in real time by the visible light imaging device carried by the drone to analyze the number and distribution of corn seedlings in the corn field sown by the agricultural machinery, compares the difference between the actual number of corn seedlings and the number of corn seedlings required for sowing, and evaluates the corn emergence rate of the plot, so as to achieve the purpose of real-time, efficient, accurate and objective evaluation of the sowing quality of the machine and the operator. Compared with the method used in production, the present application uses the visible light imaging device carried by the drone, combined with the key technologies such as the optimized drone flight route planning method, the corn seedling identification and counting method, the comprehensive evaluation method of sowing quality and the optimal evaluation window period of sowing quality, to achieve real-time, efficient, accurate and objective sowing quality evaluation. Not only can it realize the effective supervision of the sowing operation of the third party, but also can realize the analysis and judgment of the basic seedling situation of the plot, and provide a basis for the subsequent field management investment. The present application can meet the real-time, efficient, refined and objective evaluation requirements for the quality of the sowing operation of the third party under the conditions of large-scale production, realize the effective supervision of the sowing operation of the third party, and can also analyze and judge the basic seedling situation of the plot, provide a basis for the subsequent field management investment, evaluate the operation quality of the sowing machine in real time, and achieve the effect of determining the payment by seedling.

[0077] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] Figure 1 The original image data of the corn planting area;

[0079] Figure 2 The first binary image data of the corn planting area;

[0080] Figure 3 The second binary image data is the corn planting area;

[0081] Figure 4 This is a schematic diagram of Hough transform and peak detection results for corn planting area.

[0082] Figure 5 This is the second schematic diagram of Hough transform and peak detection results for corn planting area;

[0083] Figure 6 One of the efficient data collection and analysis patterns for randomly sampled emergence analysis areas in corn planting areas;

[0084] Figure 7 Figure 2 of the efficient data collection and analysis model for randomly sampled emergence analysis areas in corn planting areas;

[0085] Figure 8 The original image data of the corn planting area in Example 1;

[0086] Fig. 9 The first binary image data of the corn planting area in Example 1;

[0087] Fig.10 The second binary image data of the corn planting area of ​​Example 1;

[0088] Fig.11 It is one of the analysis precisions of the number of seedlings in the sampling area at different leaf ages;

[0089] Fig.12 The second is the analysis accuracy of the number of seedlings in the sampling area at different leaf ages;

[0090] Fig.13 The third precision of the analysis of the number of seedlings in the sampling area at different leaf ages;

[0091] Fig.14 This is the fourth precision of the analysis of the number of seedlings in the sampling areas at different leaf ages. DETAILED DESCRIPTION

[0092] The embodiments of the present application will be described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and should not be construed as limitations on the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present application.

[0093] The present application embodiment provides a method for evaluating corn sowing quality, the evaluation method comprising S100 to S300:

[0094] S100, driving a drone equipped with an image acquisition device to cruise along a target route to collect images of a corn planting area to obtain original image data.

[0095] in, Figure 1 The original image data of the corn planting area. The drone monitoring platform is built by using a multi-rotor drone equipped with a stabilization gimbal and a high-definition camera; the stabilization gimbal is used to ensure that the high-definition camera always collects images perpendicular to the horizontal direction during the flight of the drone, thereby improving the accuracy of the size measurement of ground objects. The planning strategy for the drone monitoring route is that the drone flight route should be parallel to the sowing row, simplifying the process of "seedling row processing" of the image and improving the accuracy of row spacing detection.

[0096] The method of driving a drone equipped with an image acquisition device to cruise along a target route to acquire images of a corn planting area to obtain raw image data includes:

[0097] S110, driving a drone equipped with an image acquisition device to cruise and acquire images of the corn planting area at an angle perpendicular to the horizontal direction and along a target route parallel to the planting rows of the corn planting area;

[0098] S120. During the process of cruising and collecting images of the corn planting area, the cruising altitude of the drone is limited according to the following formula:

[0099]

[0100] S130. During the cruise to collect images of the corn planting area, the flight speed of the drone and the exposure speed of the image acquisition device are limited according to the following formula:

[0101]

[0102] Among them, h is the cruising altitude, f is the focal length of the lens of the image acquisition device, GSD is the ground resolution, and the ground resolution is limited to be less than or equal to 0.005, α is the pixel size of the image acquisition device; δ is the phase point displacement of the UAV during the cruising process, and the phase point displacement is limited to be less than or equal to 0.2, V is the flight speed, and t is the exposure speed.

[0103] Specifically, the flight altitude (h) of the UAV route should be between 20m and 50m. The pixel size (α) of the camera, the route altitude (h) and the focal length (f) of the lens can be used to calculate the ground resolution based on the sensor size and image resolution of the high-definition camera. The ground resolution (GSD) should satisfy GSD≤0.005m. The flight speed (V) of the UAV and the exposure speed (t) of the high-definition camera are set with the requirement that the phase point displacement δ≤0.2 pixels during image acquisition.

[0104] The target route is planned in the following way:

[0105] Obtaining the boundary and sowing row direction of the corn sowing area;

[0106] Within the boundary, along the direction of the sowing row, planning the route of the drone so that the drone randomly samples 5 to 20 monitoring areas within the corn sowing area, and patrols each of the monitoring areas one by one by patrolling once in each of the monitoring areas;

[0107] The route of the UAV is planned to ensure that the cruising length is greater than or equal to 30 meters, the row overlap rate in the original image data is less than or equal to 50%, and the inter-row overlap rate in the original image data is 0%.

[0108] The method of obtaining the boundary and sowing row direction of the corn sowing area includes: receiving a first input from an operator; obtaining the boundary and sowing row direction of the corn sowing area in response to the first input; or reading a track file of a sowing agricultural machinery in the corn sowing area; and obtaining the boundary and sowing row direction of the corn sowing area according to the track file.

[0109] Specifically, the drone monitoring route is planned according to the operation plot. Different from the route planning strategy of drone covering the entire plot, sowing quality monitoring requires efficient and simple operation procedures, and does not require full plot image acquisition and splicing into an overall image. Therefore, there are mainly two strategies for drone route planning for corn sowing quality evaluation:

[0110] Figure 6 One of the diagrams for efficient data collection and analysis patterns for randomly sampled emergence analysis areas in corn planting areas. Figure 7Figure 2 of the efficient data collection and analysis model for the randomly sampled emergence analysis area of ​​the corn planting area.

[0111] (1) Plot sampling strategy: After obtaining the plot boundary, the row angle of sowing is manually input to make the drone route parallel to the sowing row direction. Randomly sample 5 to 20 monitoring areas in the plot to be evaluated, and fly one flight in each monitoring area, with a continuous collection length of ≥30m; since there is no need to stitch images, the row overlap rate can be set below 50% when setting the drone route, and the row overlap rate can be set to 0%; after the flight of one monitoring area is completed, the drone navigation system guides the flight to the next sampling monitoring area, and guides the image collection of all sampling areas one by one.

[0112] (2) Strategy for importing machine operation routes. In order to improve operation efficiency, large agricultural machinery is often integrated with Beidou navigation-based automatic driving equipment, and its operation track can be exported by the machine navigation equipment. Read the exported track file for drone route planning. Since the track is the sowing row direction, there is no need to manually input the sowing row angle. The track coverage area is the plot to be evaluated, so there is no need to manually input the plot boundary. According to the operation track and the monitoring width of the drone image, 5 to 20 non-overlapping tracks are randomly sampled as the monitoring area, and the continuous collection length is ≥30m; the row overlap rate can be set below 50%, and the row overlap rate can be set to 0%; after an operation track is completed, the drone navigation system guides the flight to the next sampled operation track, and guides all sampled tracks one by one to complete the collection.

[0113] S200: Preprocess the original image data to obtain binary image data.

[0114] The preprocessing of the original image data to obtain binary image data includes:

[0115] Using the EGI indicator, performing histogram equalization on the original image data to determine the segmentation threshold;

[0116] Using a threshold segmentation and binarization method, the original image data is subjected to a first preprocessing according to a relationship between the RGB color space of the original image data and the segmentation threshold, to obtain first binarized image data; Figure 2 This is the first binary image data of the corn planting area.

[0117] In order to distinguish the boundary between the corn seedlings and the land in the first binarized image data, the binarization follows the following formula:

[0118]

[0119] Among them, Green threshold is the segmentation threshold.

[0120] After obtaining the first binarized image data, preprocessing the original image data to obtain the binarized image data further includes:

[0121] Applying morphological opening operation to smooth the outline of the corn seedling in the binary image data, cut off narrow discontinuous points, and eliminate small protrusions;

[0122] Applying a closing operation to fill narrow gaps within the binary image data;

[0123] The corn seedlings are subjected to noise removal by combining the same number of opening operations and closing operations to obtain second binary image data. Figure 3 The second binarized image data of the corn planting area.

[0124] Specifically, the image of the area to be tested is transmitted back to the control tablet computer at the remote control end by the drone image transmission system. Since the tablet computer has weak computing and processing capabilities, it cannot handle complex image operations. Therefore, the present invention uses the threshold segmentation binarization method of RGB color space, uses the EGI index for histogram equalization, and determines the segmentation threshold Green threshold , the binarization method is shown in Formula 1.

[0125]

[0126] The morphological opening operation is applied to smooth the outline of the corn seedling object, cut off narrow discontinuities and eliminate small protrusions. The closing operation is applied to fill the narrow gaps inside the image. It is usually necessary to combine the same number of opening and closing operations to remove noise from the corn seedling object. The effect of the algorithm is as follows Figure 2 and 3 shown.

[0127] S300: Evaluate corn sowing quality in the corn sowing area according to the binary image data.

[0128] Specifically, the step of evaluating the corn sowing quality of the corn sowing area according to the binary image data includes:

[0129] Obtaining the sowing row direction in the binary image data through Hough transform;

[0130] Applying the nearest neighbor interpolation method, rotating the binary image data so that the sowing row direction is parallel to the y-axis of the two-dimensional plane rectangular coordinate system;

[0131] Calculating the sum of pixels of each corn seedling in the binary image data, and generating a pixel statistical map for all corn seedlings, wherein a local peak in the pixel statistical map corresponds to a corn seedling, and identifying each sowing row and / or each corn seedling in the binary image data by identifying each local peak in the pixel statistical map;

[0132] The corn sowing quality of the corn sowing area is evaluated according to each sowing row and / or each corn seedling in the binary image data.

[0133] The binary corn seedling image can be used for simple counting analysis, but in order to more accurately evaluate the sowing quality, it is necessary to obtain the row spacing and plant spacing information of the corn seedlings, and then analyze the uniformity of their emergence. The corn seedling image has discontinuous lines and nonlinear coexistence structures. The Hough method (Hough transform is used to detect curves in the image that can be described by a certain functional relationship, such as straight lines, circles, parabolas, ellipses, etc. It has been successfully applied in many fields such as image analysis and pattern recognition. The basic principle of Hough transform is to transform the curve (including straight lines) in the image space into the parameter space, and determine the description parameters of the curve by detecting the extreme points in the parameter space, so as to extract the regular curve in the image.) is used to obtain the direction of the corn sowing row, and the nearest neighbor interpolation method is used to rotate the image so that the corn row is parallel to the y-axis of the image coordinate system. Since the drone route planning is required to be parallel to the direction of the corn row, the row detection calculation amount in this step is relatively small.

[0134] After the corn rows are identified, the pixel sums for each crop row in the binary image are calculated and a statistical map is generated. Individual crop rows are identified by searching for each local peak. At the same time, peaks that are very close to each other or have lower peak values ​​are ignored to improve the identification of crop rows.

[0135] In the process of identifying each sowing row and / or each corn seedling in the binary image data by identifying each local peak in the pixel statistical graph, the peaks in the pixel statistical graph whose peaks are lower than the target threshold are ignored, and when there are multiple peaks in the pixel statistical graph and the number of the peaks is greater than the number of planting rows, the row peaks are determined by the following two formulas:

[0136]

[0137] Among them, Min pd is the distance between two adjacent peaks, and Min pd The number of pixels greater than 50% of the nominal crop row spacing, Row s is the theoretical row spacing during planting, Resolution is the calculated image resolution result;

[0138] Minph =MaxSum g *Row ps ;

[0139] Among them, Min ph is the peak value, MaxSum g is the maximum value of the sum of pixels with a pixel value of 1 in the binary image data, Row ps is the maximum peak height fraction of the corresponding row, generally a value in the range of 10 to 40%.

[0140] In other words, for images with multiple peaks and the number of peaks is greater than the number of planting rows, two restrictions are set to determine the row peaks: the distance between two adjacent peaks, Min pd Greater than 50% of the number of pixels of the nominal crop row spacing (Formula 2); Min ph The maximum fraction of peak heights greater than the sum of green pixels in the entire image Row ps (Equation 3); the median of the peak spacing is regarded as the actual crop spacing (D plant ); the pixel column where the peak is located is used as the center line of the crop row; the median of the difference between the center line positions of every two rows between the peak values ​​is regarded as the actual crop row spacing (D row ).

[0141]

[0142] Row in formula 2 s is the theoretical row spacing during planting, and Resolution is the calculated image resolution result.

[0143] Min ph =MaxSum g *Row ps (Formula 3)

[0144] In formula 3, MaxSum g The maximum value of the sum of pixels with a pixel value of 1 in the binary image. ps is the maximum peak height fraction of the corresponding row.

[0145] Figure 4 This is a schematic diagram of Hough transform and peak detection results for corn planting areas. Figure 5 This is the second schematic diagram of the Hough transform and peak detection results of the corn planting area.

[0146] The step of evaluating the corn sowing quality in the corn sowing area comprises:

[0147] The uniformity of the row spacing or plant spacing in the corn sowing area is calculated by the following formula:

[0148]

[0149] in, is the average value of row spacing or plant spacing, sd is the standard deviation of row spacing or plant spacing, UI is the uniformity index (UI) of the row spacing or plant spacing, and when UI≥5, the uniformity of the corn planting area is judged to be qualified.

[0150] The sowing density of the corn sowing area is calculated by the following formula:

[0151]

[0152] Where A is the area of ​​the corn planting area, D o For the sowing density, the median of the peak spacing was taken as the actual crop spacing (D plant ); the pixel column where the peak is located is used as the center line of the crop row; the median of the difference between the center line positions of every two rows between the peak values ​​is regarded as the actual crop row spacing (D row ).

[0153] The emergence rate of the corn sowing area is calculated by the following formula:

[0154]

[0155] Among them, D a is the number of seedlings per unit area in the corn planting area, D o is the sowing density, and ER is the emergence rate (ER).

[0156] The actual germination rate of the corn sowing area is calculated by the following formula:

[0157]

[0158] Among them, N s is the number of germinated grains in the plot of corn sowing area, N p is the actual number of sown grains in the corn sowing area, and GRα is the actual germination rate. A sample area of ​​a certain length was selected to measure the sowing depth of the grains. The average value was 3 to 6 cm, which was considered to be appropriate. When the uniformity of the sowing depth was ≤ 2.5, it was considered that the sowing depth was uneven. The grains were picked out one by one along the sowing row, and the number of sprouted grains (N) in the monitoring plot was investigated. s ) and the actual number of sown grains (Planting Grain Number, N p ),

[0159] The germination rate deviation (D) of the corn sowing area was calculated by the following formula: gr ):

[0160]

[0161] Wherein, GR is the germination rate (GR) of the field seeds in the corn sowing area obtained by measurement, and D gr If the soil moisture or ground temperature in the corn planting area is less than or equal to 5%, it is determined that the soil moisture or ground temperature in the corn planting area is suitable; otherwise, it is determined that the soil moisture or ground temperature in the corn planting area is unsuitable.

[0162] The theoretical emergence rate of the corn sowing area is calculated by the following formula:

[0163]

[0164] Among them, N e is the number of emerging grains in the corn planting area (N e ), N s is the actual number of germinated kernels in the corn sowing area, ER t is the theoretical emergence rate (ER t ).

[0165] The actual emergence rate of the corn sowing area is calculated by the following formula:

[0166]

[0167] Among them, N e is the number of seedlings unearthed in the survey, N p is the actual number of seeds sown on the plot, ER a is the actual emergence rate.

[0168] The emergence rate deviation of the corn sowing area is calculated by the following formula:

[0169]

[0170] Among them, ER t is the theoretical emergence rate, ER a is the actual emergence rate, D er Deviation of Emergence rate, D er ), in D erIf the soil tillage quality of the corn planting area is less than or equal to 5%, it is determined that the soil tillage quality of the corn planting area is qualified; otherwise, it is determined that the soil tillage quality of the corn planting area is unqualified.

[0171] Among them, ER≤ER a ≤ER t ≤GR a ≤GR.

[0172] The sowing quality index of the corn sowing area is calculated by the following formula:

[0173]

[0174] Among them, ER a is the actual emergence rate, ER is the emergence rate, GR is the germination rate of the seeds used in the field, and F sq is the sowing quality index (Sowing Quality Index), in F sq ≥0, the sowing quality index of the corn sowing area is determined to be unqualified, otherwise the sowing quality index of the corn sowing area is determined to be qualified.

[0175] Example 1

[0176] Method and device for evaluating corn sowing quality The key to achieving efficient and accurate evaluation of corn seedling numbers is to optimize and plan the UAV flight route based on the equipment and accuracy requirements. A drone monitoring platform was constructed using the DJI M600 pro six-rotor drone equipped with a Ruying gimbal and a Sony α7 high-definition camera (fixed-focus lens 35mm). The drone flight altitude was set to 30m and the ground resolution was greater than 0.005m. Strategy 1 was used to plan the flight route. The boundaries of the plot were manually input, and several monitoring areas were randomly selected from the plot. Combined with the flight altitude, route overlap rate and other settings, the respective flight routes of all sampled monitoring areas were regressed, as well as the shortest flight route between each monitoring area. The collected images were transmitted back to the control-end tablet computer, and the information on plant spacing and seedling number was obtained through binarization, morphological operations, Hough transformation and other processing processes. The above process is as follows: Figures 8 to 10 shown.

[0177] According to the monitoring image analysis, the average row spacing is 53.8 cm, the average plant spacing is 23.6 cm, the sowing density is 4265 plants / mu, the average germination rate is 85.3%, and the sowing quality index F sq =0.425, the seedling shortage caused by sowing factors accounts for 42.5% of the total seedling shortage, and the sowing quality of this plot is good.

[0178] Table 1 Examples of evaluation of plot sowing quality

[0179]

[0180] Example 2

[0181] During the early development of corn seedlings, their leaves are upright and difficult to identify and count in the soil. In the later stages of development, as the number of corn leaves increases and the length of the leaves grows, the leaves between plants in the drone orthophoto images overlap, which also has an adverse effect on the identification and counting of corn plants. Therefore, as the morphological characteristics of corn change with growth and development, there is a suitable window period for monitoring the number of corn seedlings. The analysis results of monitoring images dynamically acquired after the emergence of corn seedlings show that the monitoring accuracy of the number of corn seedlings at different leaf ages is different, and the accuracy in the early and late stages of seedlings is low and there are systematic deviations. However, the recognition accuracy of the number of corn plants in certain leaf ages is high, and there is no systematic deviation between the monitoring results and the manual counting results, such as Figures 11 to 14 shown.

[0182] The term "first" or "second" in the specification and claims of this application may include one or more of the features explicitly or implicitly. In the description of this application, unless otherwise specified, "plurality" means two or more. In addition, "and / or" in the specification and claims means at least one of the connected objects, and the character " / " generally means that the objects connected before and after are in an "or" relationship.

[0183] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "illustrative embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0184] Although the embodiments of the present application have been shown and described, those skilled in the art will appreciate that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present application, and that the scope of the present application is defined by the claims and their equivalents.

Claims

1. A method for evaluating corn sowing quality, characterized in that: The evaluation method includes: A drone equipped with an image acquisition device is driven to cruise along a target route to collect images of a corn planting area to obtain original image data; Preprocessing the original image data to obtain binary image data; According to the binary image data, evaluating the corn sowing quality of the corn sowing area; The step of evaluating the corn sowing quality in the corn sowing area comprises: The uniformity of the row spacing or plant spacing in the corn sowing area is calculated by the following formula: in, is the average value of row spacing or plant spacing, sd is the standard deviation of row spacing or plant spacing, UI is the uniformity of the row spacing or plant spacing, and when UI≥5, it is judged that the uniformity of the corn sowing area is qualified; The sowing density of the corn sowing area is calculated by the following formula: Where A is the area of ​​the corn planting area, D o For the sowing density, the median of the peak spacing is regarded as the actual crop spacing D plant The pixel column where the peak is located is used as the center line of the crop row; the median of the difference between the center line positions of every two rows between the peak values ​​is regarded as the actual crop row spacing D row ; The emergence rate of the corn sowing area is calculated by the following formula: Among them, D a is the number of seedlings per unit area in the corn planting area, D o is the sowing density, ER is the emergence rate; The actual germination rate of the corn sowing area is calculated by the following formula: Among them, N s is the number of germinated grains in the plot of corn sowing area, N p is the actual number of sown kernels in the corn sowing area, and GRα is the actual germination rate; The germination rate deviation of the corn sowing area is calculated by the following formula: Wherein, GR is the germination rate of the field seeds in the corn sowing area obtained by measurement, and D gr If the soil moisture or ground temperature in the corn planting area is less than or equal to 5%, it is determined that the soil moisture or ground temperature in the corn planting area is suitable; otherwise, it is determined that the soil moisture or ground temperature in the corn planting area is unsuitable; The theoretical emergence rate of the corn sowing area is calculated by the following formula: Among them, N e is the number of unearthed seedlings in the corn planting area, N s is the actual number of germinated kernels in the corn sowing area, ER t is the theoretical emergence rate; The actual emergence rate of the corn sowing area is calculated by the following formula: Among them, N e is the number of seedlings unearthed in the survey, N p is the actual number of seeds sown in the plot, ER a is the actual emergence rate; The emergence rate deviation of the corn sowing area is calculated by the following formula: Among them, ER t is the theoretical emergence rate, ER a is the actual emergence rate, D er is the germination rate deviation, in D er If the soil tillage quality of the corn sowing area is less than or equal to 5%, the soil tillage quality of the corn sowing area is determined to be qualified; otherwise, the soil tillage quality of the corn sowing area is determined to be unqualified; The sowing quality index of the corn sowing area is calculated by the following formula: Among them, ER a is the actual emergence rate, ER is the emergence rate, GR is the germination rate of the seeds used in the field, and F sq is the sowing quality index, in F sq ≥0, the sowing quality index of the corn sowing area is determined to be unqualified, otherwise the sowing quality index of the corn sowing area is determined to be qualified.

2. The evaluation method according to claim 1, characterized in that: The method of driving a drone equipped with an image acquisition device to cruise along a target route to acquire images of a corn planting area to obtain raw image data includes: Driving a drone equipped with an image acquisition device to cruise and acquire images of the corn planting area at an angle perpendicular to the horizontal direction and along a target route parallel to the planting rows of the corn planting area; During the cruise to collect images of the corn planting area, the cruise altitude of the drone is limited according to the following formula: In the process of cruising and collecting images of the corn planting area, the flight speed of the UAV and the exposure speed of the image acquisition device are limited according to the following formula: Among them, h is the cruising altitude, f is the focal length of the lens of the image acquisition device, GSD is the ground resolution, and the ground resolution is limited to be less than or equal to 0.005, α is the pixel size of the image acquisition device; δ is the phase point displacement of the UAV during the cruising process, and the phase point displacement is limited to be less than or equal to 0.2, V is the flight speed, and t is the exposure speed.

3. The evaluation method according to claim 1, characterized in that: The target route is planned in the following way: Obtaining the boundary and sowing row direction of the corn sowing area; Within the boundary, along the direction of the sowing row, planning the route of the drone so that the drone randomly samples 5 to 20 monitoring areas within the corn sowing area, and patrols each of the monitoring areas one by one by patrolling once in each of the monitoring areas; The route of the UAV is planned to ensure that the cruising length is greater than or equal to 30 meters, the row overlap rate in the original image data is less than or equal to 50%, and the inter-row overlap rate in the original image data is 0%.

4. The evaluation method according to claim 3, characterized in that: The step of obtaining the boundary of the corn planting area and the planting row direction comprises: receiving a first input from an operator; In response to the first input, obtaining the boundary and sowing row direction of the corn sowing area; or, Reading the track file of the agricultural machinery sowing in the corn sowing area; According to the track file, the boundary and the direction of the corn planting area are obtained.

5. The evaluation method according to any one of claims 1 to 4, characterized in that: The preprocessing of the original image data to obtain binary image data includes: Using the EGI indicator, performing histogram equalization on the original image data to determine the segmentation threshold; Using a threshold segmentation and binarization method, the original image data is subjected to a first preprocessing according to a relationship between the RGB color space of the original image data and the segmentation threshold, to obtain first binarized image data; In order to distinguish the boundary between the corn seedlings and the land in the first binarized image data, the binarization follows the following formula: Among them, Green threshold is the segmentation threshold.

6. The evaluation method according to claim 5, characterized in that: After obtaining the first binarized image data, preprocessing the original image data to obtain the binarized image data further includes: Applying morphological opening operation to smooth the outline of the corn seedling in the binary image data, cut off narrow discontinuous points, and eliminate small protrusions; Applying a closing operation to fill narrow gaps within the binary image data; The corn seedlings are subjected to noise removal by combining the same number of opening operations and closing operations to obtain second binary image data.

7. The evaluation method according to any one of claims 1 to 4, characterized in that: The step of evaluating the corn sowing quality of the corn sowing area according to the binary image data comprises: Obtaining the sowing row direction in the binary image data through Hough transform; Applying the nearest neighbor interpolation method, rotating the binary image data so that the sowing row direction is parallel to the y-axis of the two-dimensional plane rectangular coordinate system; Calculating the sum of pixels of each corn seedling in the binary image data, and generating a pixel statistical map for all corn seedlings, wherein a local peak in the pixel statistical map corresponds to a corn seedling, and identifying each sowing row and / or each corn seedling in the binary image data by identifying each local peak in the pixel statistical map; The corn sowing quality of the corn sowing area is evaluated according to each sowing row and / or each corn seedling in the binary image data.

8. The evaluation method according to claim 7, characterized in that: In the process of identifying each sowing row and / or each corn seedling in the binary image data by identifying each local peak in the pixel statistical graph, the peaks in the pixel statistical graph whose peaks are lower than the target threshold are ignored, and when there are multiple peaks in the pixel statistical graph and the number of the peaks is greater than the number of planting rows, the row peaks are determined by the following two formulas: Among them, Min pd is the distance between two adjacent peaks, and Min pd The number of pixels greater than 50% of the nominal crop row spacing, Row s is the theoretical row spacing during planting, Resolution is the calculated image resolution result; Who ph =MaxSum g *Row ps ; Among them, Min ph is the peak value, MaxSum g is the maximum value of the sum of pixels with a pixel value of 1 in the binary image data, Row ps is the maximum peak height fraction of the corresponding row.

9. A device for evaluating corn sowing quality, characterized in that: The evaluation device comprises a processor, and the processor is used to implement the steps of the evaluation method according to any one of claims 1 to 8 when executing a computer program stored in a memory.

Images