Grain monitoring method for harvester, grain monitoring equipment and harvester

Abstract

The invention discloses a grain monitoring method and grain monitoring equipment for a harvester and the harvester, and the grain monitoring method for the harvester comprises the steps: collecting a real-time image of harvested grains on a grain elevator, and determining a first pixel number occupied by impurities in the harvested grains and a second pixel number occupied by grains according to the real-time image, according to the method, the first pixel number and the second pixel number are obtained, the first mass and the second mass of impurities and grains are determined according to the first pixel number and the second pixel number respectively, the impurity rate of harvested grains is determined according to the first mass and the second mass, and working parameters of the harvester are adjusted under the condition that the impurity rate exceeds a preset impurity value. The working parameters of the harvester can be dynamically adjusted, and the grain quality of grains harvested by the harvester is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of agricultural production equipment, and in particular relates to a grain monitoring method for a harvester, grain monitoring equipment and a harvester. Background Art

[0002] The impurity rate and breakage rate are important indicators for measuring the working quality of agricultural machinery, and are also important bases for adjusting the operating parameters of the working parts of agricultural machinery. In the existing technology, there is a lack of dynamic adjustment mechanism for the working parameters of agricultural machinery, and the working parameters cannot be adjusted in time according to the quality of the grain harvested by the agricultural machinery, resulting in low grain quality. Summary of the invention

[0003] In view of the above-mentioned defects or shortcomings, the present invention provides a grain monitoring method for a harvester, a grain monitoring device and a harvester, aiming to solve the technical problem that the existing agricultural machinery is unable to dynamically adjust the operating parameters, resulting in low grain quality.

[0004] To achieve the above-mentioned object, the present invention provides a grain monitoring method for a harvester, a grain monitoring device and a harvester, wherein the grain monitoring method for a harvester comprises:

[0005] Capture real-time images of harvested grain on the grain elevator;

[0006] Determining a first number of pixels occupied by impurities and a second number of pixels occupied by kernels in the harvested grains based on the real-time image;

[0007] Determine a first mass and a second mass of the impurities and the seeds, respectively, based on the first number of pixels and the second number of pixels;

[0008] The impurity content of the harvested grains is determined based on the first mass and the second mass, and the working parameters of the harvester are adjusted when the impurity content exceeds a preset impurity value.

[0009] In an embodiment of the present invention, determining a first number of pixels occupied by impurities and a second number of pixels occupied by grains in the harvested grains according to the real-time image includes:

[0010] The real-time image is input into a preset impurity and fragmentation rate recognition model to obtain a first pixel number and a second pixel number through the output of the impurity and fragmentation rate recognition model.

[0011] In an embodiment of the present invention, a preset construction method of a mixed fragmentation rate identification model includes:

[0012] Pre-collecting a first sample image of a plurality of grain samples, wherein the grain samples include intact grains, broken grains and impurities;

[0013] Calibrate the intact kernels, broken kernels and impurities in each first sample image, and divide the multiple first sample images into a type training set and a type verification set;

[0014] The type training set and the type verification set are input into the image segmentation recognition model to construct a generation impurity fragmentation rate recognition model.

[0015] In an embodiment of the present invention, inputting the type training set and the type verification set into the image segmentation recognition model to construct and generate the impurity fragmentation rate recognition model includes:

[0016] Inputting the type training set and the type verification set into the image segmentation recognition model, wherein the image segmentation recognition model is set to recognize images at different pixel ratios;

[0017] Substitute the feature values ​​output by the image segmentation recognition model at different pixel ratios into a preset weight calculation formula to obtain a weighted feature value;

[0018] Determine the output result of the type of the impurity and broken rate identification model according to the weighted eigenvalue;

[0019] The type output results corresponding to the type verification set are used to adjust the weight coefficient in the weight calculation formula to construct a model for identifying the impurity and fragmentation rate.

[0020] In an embodiment of the present invention, the image segmentation recognition model is set to recognize images at different pixel ratios and further includes:

[0021] A plurality of output branch groups for parallel calculation are set in the image segmentation and recognition model, wherein each output branch group is used to recognize the image at a corresponding pixel ratio.

[0022] In an embodiment of the present invention, the preset construction method of the impurity breakage rate identification model further includes:

[0023] Pre-collecting a second sample image of a plurality of grain samples, wherein the grain samples include intact grains, broken grains, and impurities;

[0024] Calibrate the complete kernels, broken kernels and impurities in each second sample image, and divide the multiple second sample images into type test sets;

[0025] Input the type test set into the impurity fragmentation rate recognition model to obtain the test impurity rate of the type test set;

[0026] The preset impurity value is adjusted according to the deviation value between the test impurity rate of the type test set and the actual impurity rate.

[0027] In an embodiment of the present invention, determining the first mass and the second mass of the impurities and the seeds respectively according to the first number of pixels and the second number of pixels includes:

[0028] The real-time image, the first pixel number and the second pixel number are input into a preset quality estimation model to obtain the first quality and the second quality through the quality estimation model output.

[0029] In an embodiment of the present invention, the preset construction method of the quality estimation model includes:

[0030] respectively collecting third sample images and weights of a plurality of intact kernel samples, fourth sample images and weights of a plurality of broken kernel samples, and fifth sample images and weights of a plurality of impurity samples;

[0031] labeling weights for each third sample image, each fourth sample image, and each fifth sample image, respectively, and dividing the images into a quality training set and a quality verification set;

[0032] The quality training set and the quality verification set are input into the image segmentation and recognition model to construct a generation quality estimation model.

[0033] In an embodiment of the present invention, the harvester grain monitoring method further includes:

[0034] determining a third quality of broken kernels in the harvested grains according to the number of kernel pixels occupied by broken kernels in the kernels, wherein the kernels include broken kernels and intact kernels;

[0035] The crushing rate of the harvested grains is determined according to the second mass and the third mass, and the working parameters of the harvester are adjusted when the crushing rate exceeds a preset crushing value.

[0036] In order to achieve the above-mentioned objectives, the present invention also provides a grain monitoring device, wherein the grain monitoring device includes an image analysis device and an image acquisition device, the image analysis device is configured with a processor, the processor is communicatively connected to the image acquisition device, and is used to execute the grain monitoring method for a harvester according to any one of claims 1 to 7, and the image acquisition device is used to acquire real-time images.

[0037] In an embodiment of the present invention, the image acquisition device includes:

[0038] An installation shell is mounted on the grain elevator of the harvester and forms an installation space with an image acquisition window, and the image acquisition window is located on one side of the installation space close to the transmission surface of the grain elevator;

[0039] The image acquisition component is arranged in the installation space and includes a light source and a high-speed camera. The light source can emit visible light to illuminate the image acquisition window. The high-speed camera is communicatively connected to the processor. The lens of the high-speed camera is set toward the image acquisition window, and the high-speed camera can move toward or away from the image acquisition window.

[0040] In order to achieve the above objectives, the present invention also provides a harvester, wherein the harvester includes the above-mentioned grain monitoring device.

[0041] Through the above technical solution, the grain monitoring method for a harvester, the grain monitoring device and the harvester provided by the embodiment of the present invention have the following beneficial effects:

[0042] The above-mentioned grain monitoring method for harvesters first collects real-time images of harvested grains on a grain elevator. In the real-time images, impurities, intact grains and broken grains are in different colors and forms. By selecting the impurities, intact grains and broken grains in the real-time images, the first pixel number occupied by impurities and the second pixel number occupied by grains in the real-time images of harvested grains can be statistically obtained. Considering the difference in density between impurities and grains, combined with the density parameters of impurities and grains, the first pixel number is converted into the first mass of impurities, and the second pixel number is converted into the second mass of grains. The first mass and the second mass are substituted into the impurity content calculation formula to obtain the impurity content of the harvested grains, thereby judging the grain quality of the harvested grains. When the impurity content exceeds the preset impurity content value, the working parameters of the harvester are adjusted to reduce the impurity content of the harvested grains. Through the above-mentioned method, dynamic adjustment of the working parameters of the harvester can be achieved, and the grain quality of the grain harvested by the harvester can be improved.

[0043] Other features and advantages of the present invention will be described in detail in the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The accompanying drawings are used to provide an understanding of the present invention and constitute a part of the specification. Together with the following specific embodiments, they are used to explain the present invention, but do not constitute a limitation of the present invention. In the accompanying drawings:

[0045] Figure 1 is a flow chart of a method for monitoring grains for a harvester according to an embodiment of the present invention;

[0046] Figure 2 is a flow chart of a preset construction method of a mixed and broken rate identification model according to an embodiment of the present invention;

[0047] Figure 3 is a flow chart of a preset construction method of a quality estimation model according to an embodiment of the present invention;

[0048] Figure 4 is an overall structural diagram of a seed monitoring device according to an embodiment of the present invention;

[0049] Figure 5 is a side view of a seed monitoring device according to an embodiment of the present invention;

[0050] Figure 6 is a bottom structural diagram of a seed monitoring device according to an embodiment of the present invention;

[0051] Figure 7 1 is a structural diagram of a mounting stand according to an embodiment of the present invention.

[0052] Description of Reference Numerals

[0053] 1 Mounting housing 10 Image acquisition window

[0054] 11 Bottom plate 12 Side panel

[0055] 120 lateral notch 121 first enclosure body

[0056] 122 second enclosure body 123 first connection portion

[0057] 124 second connection portion 13 cover plate

[0058] 131 top plate 132 side plate

[0059] 2 Image acquisition component 21 Light source component

[0060] 211 Rotating shaft 212 Adjusting shaft

[0061] 22 High-speed camera 23 Mounting stand

[0062] 231 Fixed column 232 Lifting block

[0063] 233 Camera mounting plate 24 Light source mounting seat

[0064] 241 Install vertical board 242 Shading horizontal board

[0065] 241a Angle adjustment hole 31 Dust cover

[0066] 32 pressure plate 33 notch DETAILED DESCRIPTION

[0067] The specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. It should be understood that the specific embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0068] The following describes the grain monitoring method for a harvester, the grain monitoring device, and the harvester of the present invention with reference to the accompanying drawings.

[0069] like Figure 1 As shown, the present invention provides a method for monitoring grains for a harvester, wherein the method for monitoring grains for a harvester comprises:

[0070] Step S110: collecting real-time images of harvested grains on the grain elevator;

[0071] Specifically, a high-speed camera 22 is arranged above the grain elevator to continuously collect real-time images of harvested grains in the transport state to ensure the timeliness of the images.

[0072] Step S120: determining the first number of pixels occupied by impurities and the second number of pixels occupied by grains in the harvested grains according to the real-time image;

[0073] Step S130: determining the first mass and the second mass of the impurities and the seeds respectively according to the first pixel number and the second pixel number;

[0074] Step S140: Determine the impurity content of the harvested grains according to the first mass and the second mass, and adjust the working parameters of the harvester when the impurity content exceeds a preset impurity value.

[0075] The above-mentioned grain monitoring method for harvesters first collects real-time images of harvested grains on a grain elevator. In the real-time images, impurities and grains are in different colors and forms. By selecting impurities, complete grains and broken grains in the real-time images, the first pixel number occupied by impurities and the second pixel number occupied by grains in the real-time images of harvested grains can be statistically obtained. Considering the difference in density between impurities and grains, combined with the density parameters of impurities and grains, the first pixel number is converted into the first mass of impurities, and the second pixel number is converted into the second mass of grains. The first mass and the second mass are substituted into the impurity content calculation formula to obtain the impurity content of the harvested grains, thereby judging the grain quality of the harvested grains. When the impurity content exceeds the preset impurity content value, the working parameters of the harvester are adjusted to reduce the impurity content of the harvested grains. Through the above-mentioned method, dynamic adjustment of the working parameters of the harvester can be achieved, and the grain quality of the grain harvested by the harvester can be improved.

[0076] In the embodiment of the present invention, step S120, determining the first number of pixels occupied by impurities and the second number of pixels occupied by grains in the harvested grains according to the real-time image, comprises:

[0077] The real-time image is input into a preset impurity and fragmentation rate recognition model to obtain a first pixel number and a second pixel number through the output of the impurity and fragmentation rate recognition model.

[0078] The impurity and broken rate recognition model is an image recognition model that can identify and select complete kernels, broken kernels, and impurities in real-time images, respectively, and then count the number of pixels of the selected part. The recognition process is automatically implemented by the impurity and broken rate recognition model, and the detection efficiency is high. Specifically, the impurity and broken rate recognition model will calculate multiple results based on an input real-time image, and each result includes the position parameters and category parameters of the image (indicating that the image is a complete kernel, a broken kernel, an impurity or other category). Combining the position parameters and category parameters of multiple results, impurities or complete kernels or broken kernels can be selected, and the first number of pixels occupied by impurities or the second number of pixels occupied by kernels (including complete kernels and broken kernels) can be determined.

[0079] like Figure 2 As shown, in an embodiment of the present invention, the preset construction method of the impurity breakage rate identification model includes:

[0080] Step S210: pre-collecting a first sample image of a plurality of grain samples, wherein the grain samples include intact grains, broken grains, and impurities;

[0081] Step S220: calibrating the impurities, intact kernels, and broken kernels in each first sample image, and dividing the multiple first sample images into a type training set and a type verification set;

[0082] Step S230: input the type training set and the type verification set into the image segmentation recognition model to construct a generated impurity fragmentation rate recognition model.

[0083] Preferably, the image segmentation recognition model adopts a YOLO model, specifically a YOLOv8 model. The YOLOv8 model is a target detection algorithm model that can be used for tasks such as image classification, object detection, and instance segmentation. The impurity and breakage rate recognition model of the present invention is based on the YOLOv8 model, and the above-mentioned type training set and type verification set are input and iteratively trained to obtain a recognition model suitable for grains. Specifically, in some embodiments of the present invention, different impurity and breakage rate recognition models are trained based on the YOLOv8 model for different types of grains, for example, a wheat recognition algorithm model is trained for wheat, and a corn recognition algorithm model is trained for corn.

[0084] In the embodiment of the present invention, step S220: inputting the type training set and the type verification set into the image segmentation recognition model to construct and generate the impurity fragmentation rate recognition model includes:

[0085] Inputting the type training set and the type verification set into the image segmentation recognition model, wherein the image segmentation recognition model is set to recognize images at different pixel ratios;

[0086] Substitute the feature values ​​output by the image segmentation recognition model at different pixel ratios into a preset weight calculation formula to obtain a weighted feature value;

[0087] Determine the output result of the type of the impurity and broken rate identification model according to the weighted eigenvalue;

[0088] The type output results corresponding to the type verification set are used to adjust the weight coefficient in the weight calculation formula to construct a model for identifying the impurity and fragmentation rate.

[0089] Specifically, the training set is input into the YOLOv8 model, and the image is identified with pixel ratios of 80×80, 40×40, and 20×20 respectively. A eigenvalue is obtained for each pixel ratio. In the linear weight calculation formula, the eigenvalue of each pixel ratio is multiplied by a weight coefficient. Finally, the weighted eigenvalue is obtained by the sum of three eigenvalues ​​and three weight coefficients. The type output result can be obtained according to the size of the weighted eigenvalue and the preset judgment interval, thereby realizing the mapping of image information (such as color, position) to type output result. The initial model for grain type recognition can be obtained through multiple iterative training, and the validation set is used to verify the accuracy of the type output result. The weight coefficient corresponding to each pixel ratio in the weight calculation formula is adjusted by comparing the output result of the initial model with the calibration data of the validation set, and an accurate recognition model for impurity and broken rate is constructed.

[0090] Of course, this is only one setting scheme of the present invention. Depending on the specific type of grain to be identified, pixel ratios of other sizes or quantities may also be set.

[0091] In an embodiment of the present invention, the image segmentation recognition model is set to recognize images at different pixel ratios and further includes:

[0092] A plurality of output branch groups for parallel calculation are set in the image segmentation and recognition model, wherein each output branch group is used to recognize the image at a corresponding pixel ratio.

[0093] The above method can effectively utilize the multi-threaded computing capability of the processor by splitting the image segmentation recognition model into at least multiple parallel output branches, thereby improving the efficiency of image recognition. Specifically, in the YOLOv8 model, one image corresponds to 8400 output results. The traditional computing solution is only provided with one output branch, and the processor can only calculate one output branch at a time. In the present invention, three output branch groups are provided, and the image is recognized and calculated at pixel ratios of 80×80, 40×40, and 20×20, respectively, and the multi-threaded computing capability of the processor is utilized to improve the efficiency of image recognition.

[0094] Furthermore, each output branch group includes a position parameter branch and a category parameter branch, the position parameter branch is used to calculate the position parameter in each output result, and the category parameter branch is used to calculate the category parameter in each output result. The present invention further provides a 6-branch calculation scheme, each branch is processed by a thread of the processor, and the efficiency of image recognition is further improved. Of course, the present invention is not limited to this, and each output branch group can be further split to obtain 3N parallel branches, where N is a positive integer.

[0095] In an embodiment of the present invention, the preset construction method of the impurity breakage rate identification model further includes:

[0096] Pre-collecting a second sample image of a plurality of grain samples, wherein the grain samples include intact grains, broken grains, and impurities;

[0097] Calibrate the complete kernels, broken kernels and impurities in each second sample image, and divide the multiple second sample images into type test sets;

[0098] Input the type test set into the impurity fragmentation rate recognition model to obtain the test impurity rate of the type test set;

[0099] The preset impurity value is adjusted according to the deviation value between the test impurity rate of the type test set and the actual impurity rate.

[0100] The type test set can be used to further verify the accuracy of the impurity content and fragmentation rate recognition model. If the test impurity content is greater than the actual impurity content, it means that the impurity content and fragmentation rate recognition model is more stringent and the preset impurity value can be appropriately increased. If the test impurity content is less than the actual impurity content, it means that the impurity content and fragmentation rate recognition model is more loose and the preset impurity value can be appropriately reduced. On the basis of model recognition, manual recognition can be combined to reduce the judgment error of the impurity content.

[0101] In an embodiment of the present invention, determining the first mass and the second mass of the impurities and the seeds respectively according to the first number of pixels and the second number of pixels includes:

[0102] The real-time image, the first pixel number and the second pixel number are input into a preset quality estimation model to obtain the first quality and the second quality through the quality estimation model output.

[0103] like Figure 3 As shown, in an embodiment of the present invention, the preset construction method of the quality estimation model includes:

[0104] Step S310: respectively collecting a third sample image and weight of a plurality of intact kernel samples, a fourth sample image and weight of a broken kernel sample, and a fifth sample image and weight of a plurality of impurity samples;

[0105] Specifically, the sampling process of the complete grain sample, the broken grain sample and the impurity sample is as follows: 5 samples are randomly selected from the total sample (including the complete grain sample, the broken grain sample and the impurity sample), each of which is not less than 2000g, and 5 samples of 500g each are obtained by quartering, so 25 samples are obtained, and 25 weighing data are obtained, and the image of each sample is collected by the camera. The image resolution of the third sample image, the fourth sample image and the fifth sample image are all 2592×2048, and the number of pixels occupied by each complete grain sample, each broken grain sample and each impurity sample in the third sample image, the fourth sample image and the fifth sample image is different to simulate different distribution density states, so the trained quality estimation model can obtain the accurate value of the mass corresponding to each impurity pixel, each complete grain pixel and each broken grain pixel according to the distribution density of the harvested grain.

[0106] Step S320: labeling weights for each of the third sample image, the fourth sample image, and the fifth sample image, respectively, and dividing the images into a quality training set and a quality verification set;

[0107] Step S330: input the quality training set and the quality verification set into the image segmentation and recognition model to construct a generation quality estimation model.

[0108] The specific setting method of the quality estimation model is the same as the setting method of the above-mentioned impurity and breakage rate identification model, which will not be repeated here.

[0109] In an embodiment of the present invention, the harvester grain monitoring method further includes:

[0110] determining a third quality of broken kernels in the harvested grains according to the number of kernel pixels occupied by broken kernels in the kernels, wherein the kernels include broken kernels and intact kernels;

[0111] The crushing rate of the harvested grains is determined according to the second mass and the third mass, and the working parameters of the harvester are adjusted when the crushing rate exceeds a preset crushing value.

[0112] Specifically, the impurity rate calculation formula is:

[0113] Lz=(Sz×Mz) / (Sz×Mz+Sw×Mw+Ss×Ms);

[0114] The formula for calculating the crushing rate is:

[0115] Ls=(Ss×Ms) / (Sw×Mw+Ss×Ms);

[0116] In the above formula, Lz is the impurity rate, Ls is the breakage rate, Sz is the number of pixels occupied by impurities in the first pixel number, Mz is the single-pixel mass of impurities, Sw is the number of pixels of intact kernels in the second pixel number, Mw is the single-pixel mass of intact kernels, Ss is the number of pixels of broken kernels in the second pixel number, and Ms is the single-pixel mass of broken kernels, where the single-pixel mass of intact kernels, impurities, and broken kernels is calculated by averaging the mass of every 1000 pixels obtained by the above-mentioned mass estimation model.

[0117] In order to achieve the above-mentioned objectives, the present invention also provides a grain monitoring device, wherein the grain monitoring device includes an image analysis device and an image acquisition device, the image analysis device is configured with a processor, the processor is communicatively connected to the image acquisition device, and is used to execute the grain monitoring method for a harvester according to any one of claims 1 to 7, and the image acquisition device is used to acquire real-time images.

[0118] Specifically, the image analysis device includes a self-developed 10-inch integrated display and control screen based on the domestic computing chip RV1126, with the processing power of a 4-core Cortex A7, 1GB DDR3 memory, 8GB eMMC storage, NPU2.0T computing power, 8-channel AHD camera interface, and a QT interface program developed in C++ language. The integrated display and control screen displays the real-time detection of intact grains, broken grains, and impurities on its display interface by calling the above model, and dynamically displays the impurity rate and breakage rate of harvested grains.

[0119] The image analysis device identifies intact grains, broken grains, and non-grain impurities through the impurity and breakage rate recognition model, calculates their respective masses through the mass estimation model, and infers their corresponding total masses. According to the calculation formulas for the impurity and breakage rates, the value of each mass as a percentage of the total mass is calculated, and the impurity and breakage rates of the grains are output. When the impurity or breakage rate exceeds the preset value, the harvester speed, drum speed, etc. are controlled to provide feedback to reduce the impurity and breakage rates, and further optimize the quality of the harvested grain.

[0120] The image analysis device is equipped with the following interfaces:

[0121] Grain model initialization interface, to realize model loading and configuration initialization;

[0122] Grain model uninstallation interface to realize model uninstallation and memory cleanup;

[0123] Picture saving interface, which can save the collected data as pictures;

[0124] The kernel image inference interface can identify and output the coordinates of complete kernels, broken kernels, and impurity object frames, and calculate and output the kernel impurity rate and breakage rate;

[0125] Grain control interface, to set and control parameters such as harvester speed and drum speed.

[0126] In an embodiment of the present invention, the image acquisition device includes:

[0127] An installation shell 1 is mounted on a grain elevator of the harvester and forms an installation space having an image acquisition window 10, and the image acquisition window 10 is located on one side of the installation space close to the transmission surface of the grain elevator;

[0128] The image acquisition component 2 is arranged in the installation space and includes a light source 21 and a high-speed camera 22. The light source 21 can emit visible light to illuminate the image acquisition window 10. The high-speed camera 22 is communicatively connected to the processor. The lens of the high-speed camera 22 is set toward the image acquisition window 10, and the high-speed camera 22 can move toward or away from the image acquisition window 10.

[0129] When the harvester uses the above-mentioned grain monitoring device, the mounting shell 1 is fixed on the grain elevator of the harvester through a connecting piece, and an image acquisition component 2 is arranged in the mounting shell 1. The grains harvested by the harvester are transmitted to the grain box through the grain elevator. The light source 21 emits light to illuminate the grains on the transmission surface through the image acquisition window 10. The high-speed camera 22 is used to continuously capture images of the grains on the transmission surface, and the operator can adjust the position of the high-speed camera 22 so that the high-speed camera 22 is in the best imaging position, thereby capturing a clear image of the grains in the transportation state. Compared with the technical solution of sampling and shooting, the image acquisition cycle is shortened and the timeliness of image analysis is improved. Based on the image information collected by the grain monitoring device, the performance of the harvester can be judged more accurately and intuitively.

[0130] like Figure 4 As shown, in the embodiment of the present invention, the image acquisition component 2 further includes a light source mounting seat 24 located in the installation space, the light source mounting seat 24 is mounted on the periphery of one side of the image acquisition window 10, the light source component 21 is mounted on the light source mounting seat 24, and the light emitting surface of the light source component 21 is arranged toward the image acquisition window 10, and the high-speed camera 22 is arranged directly opposite the image acquisition window 10. Then the light source mounting seat 24 and the light source component 21 are arranged outside the image acquisition window 10, and the light emitting surface of the light source component 21 is directly toward the transmission surface of the grain elevator, which can ensure the brightness of the image collected by the high-speed camera 22 and improve the accuracy of subsequent image analysis.

[0131] Of course, the present invention is not limited thereto. In some embodiments, the light source 21 may also be arranged in a ring shape and disposed directly below the high-speed camera 22 .

[0132] like Figure 5As shown, in the embodiment of the present invention, the light source mounting seat 24 includes two mounting uprights 241 relatively spaced apart along the periphery of the image acquisition window 10, and angle adjustment holes 241a are provided on the two mounting uprights 241. The opposite ends of the light source component 21 are respectively and one-to-one correspondingly penetrated into the angle adjustment holes 241a of the two mounting uprights 241, so that the angle of the light source component 21 can be adjusted. Then, the angle of the light source component 21 can be adjusted through the light source mounting seat 24 to make it in the best angle position. It can be understood that the installation tolerance of the grain monitoring device on the grain elevator, the type of grain harvested by the harvester and other factors will affect the lighting conditions required for the high-speed camera 22 to shoot. A single-angle light source is difficult to ensure that the image collected by the high-speed camera 22 has sufficient clarity. The grain monitoring device provided by the present invention can adjust the light source component 21 according to the actual working conditions to ensure the clarity of the image, thereby improving the accuracy of the judgment of the image analysis system.

[0133] like Figure 5 As shown, in the embodiment of the present invention, the light source component 21 is provided with a rotating shaft 211 and an adjusting shaft 212 at opposite ends, and the adjusting shaft 212 is provided on the upper side of the rotating shaft 211, the light source component 21 is rotatably provided on the mounting vertical plate 241 through the rotating shaft 211, and the adjusting shaft 212 is passed through the angle adjustment hole 241a, and the angle adjustment hole 241a is provided to extend in a circle with the center of the rotating shaft 211 as the center. Specifically, the adjusting shaft 212 is provided as a bolt, and by screwing the bolt, the light source component 21 can be loosened or fixed, and the inclination angle of the light source component 21 can be adjusted steplessly.

[0134] Furthermore, the top ends of the two mounting vertical plates 241 are connected with a shading horizontal plate 242, which can form a shielding above the light source 21 to prevent the light emitted by the light source 21 from directly shining on the lens of the high-speed camera 22 and causing light spots to appear in the image, thereby ensuring the accuracy of image recognition.

[0135] like Figure 4 and Figure 7 As shown, in the embodiment of the present invention, the image acquisition component 2 further includes a mounting stand 23 located in the installation space, the mounting stand 23 and the light source mounting seat 24 are respectively arranged on opposite sides of the image acquisition window 10, and the mounting stand 23 includes a fixed column 231 and a lifting block 232 which is height-adjustable and arranged on the fixed column 231, and the high-speed camera 22 is arranged on the lifting block 232. The mounting stand 23 and the light source mounting seat 24 are arranged opposite to each other with the image acquisition window 10 separated, and there is no interference between the two. Figure 1In the figure, the high-speed camera 22 is located above and to the side of the light source mounting seat 24. Obviously, the high-speed camera 22 can also be lowered to be flush with the light source mounting seat 24 by adjusting the lifting block 232. A certain space is still reserved between the two to avoid bumping the high-speed camera 22 during the adjustment process.

[0136] A bent camera mounting plate 233 is provided on the lifting block 232, and a lateral section of the camera mounting plate 233 extends from the lifting block 232 toward a side where a light source mounting seat 24 is provided, and a vertical section of the camera mounting plate 233 extends upward from an end of the lateral section away from the lifting block 232 to form an installation facade for installing the high-speed camera 22.

[0137] Since the light source 21 only needs to be adjusted to the optimal imaging position during installation, the mounting bracket 23 is preferably a manually adjustable structure, such as a bolt adjustment structure, which has a simple structure and low cost.

[0138] like Figure 4 and Figure 6 As shown, in the embodiment of the present invention, the seed monitoring device further includes a dustproof plate 31 detachably placed in the image acquisition window 10 to prevent dust from entering the installation space, and to ensure image acquisition by the high-speed camera 22, the dustproof plate 31 is set to a transparent material. Preferably, the dustproof plate 31 is set to an acrylic plate, which is light in weight and easy to disassemble and wipe; or the dustproof plate 31 can also be set to a glass plate.

[0139] In the embodiment of the present invention, a step portion for overlapping the dustproof plate 31 is formed on the side wall of the image acquisition window 10, and the seed monitoring device further includes a pressing plate 32, one end of the pressing plate 32 is hinged to a peripheral edge of one side of the image acquisition window 10, and the other end is used to press on the dustproof plate 31. The dustproof plate 31 can be quickly pressed or loosened by rotating the pressing plate 32, so that the dustproof plate 31 can be easily disassembled and assembled.

[0140] It is understandable that the harvester is a grain harvesting equipment. The uneven road surface and the operation of the engine will cause the vibration of the harvester shell. The grain monitoring device is directly installed on the harvester. If the dustproof plate 31 is not fixed, the dustproof plate 31 will shake violently, affecting the image acquisition quality, and even causing the dustproof plate 31 to fall out of the image acquisition window 10. By setting a pressure plate 32, the dustproof plate 31 can be pressed against the step portion to reduce or avoid the shaking of the dustproof plate 31.

[0141] Specifically, one end of the pressing plate 32 is hinged to the bottom plate 11 by a connecting bolt, and the pressing plate 32 can rotate horizontally around the bolt, and by screwing the connecting bolt, the pressing force of the pressing plate 32 on the dustproof plate 31 can also be adjusted. Of course, the present invention is not limited to this, and the dustproof plate 31 can also be fixedly connected to the bottom plate 11 by bolt connection.

[0142] Furthermore, a notch 33 is provided on the side of the image acquisition window 10 to facilitate the operator to take out the dustproof plate 31 from the image acquisition window 10 .

[0143] like Figures 4 to 6 As shown, in the embodiment of the present invention, the mounting housing 1 includes a bottom plate 11, a side panel 12 and a cover plate 13 enclosing a mounting space, an image acquisition window 10 is provided on the bottom plate 11, the side panels 12 are enclosed on at least two sides of the bottom plate 11, and a lateral notch 120 is formed on at least one side, and the cover plate 13 is detachably connected to the side panels 12 and can cover the top and lateral notches 120 of the mounting space. The mounting housing 1 is designed as a split structure, which is convenient for assembly and disassembly on the harvester.

[0144] Specifically, when installing the grain monitoring device, first install the mounting stand 23, the light source mounting seat 24 and the light source component 21 on the base plate 11, then fix the base plate 11 and the side panel 12 together on the grain elevator, and then install the high-speed camera 22, and adjust the installation height of the high-speed camera 22 and the pitch angle of the light source component 21 to obtain the best imaging clarity, and then install the cover plate 13.

[0145] Specifically, the side panels 12 are provided with heat dissipation holes and wiring holes for the cables of the high-speed camera and the light source 21 to pass through, and a protective cover is also provided on the outside of the side panels 12, which covers the heat dissipation holes and the wiring holes to prevent rain and dust from entering the installation space. Furthermore, the protective cover adopts a detachable structure, which can be a connection hole provided on the side of the protective cover, or a buckle can be provided on the protective cover, and correspondingly, the side panels 12 also need to be provided with a connection hole or a card slot.

[0146] like Figure 5 As shown, in the embodiment of the present invention, the bottom end of the side panel 12 is bent inward to form a first connection portion 123, and the bottom plate 11 is placed on the first connection portion 123 and is detachably connected to the first connection portion 123 through fasteners. The side panel 12 and the bottom plate 11 can be fixed to the grain elevator at the same time through a set of fasteners, and there is no need to fix the side panel 12 and the bottom plate 11 separately, thereby reducing the use of fasteners.

[0147] like Figure 6As shown, in the embodiment of the present invention, the side panel 12 includes a first panel body 121 and a second panel body 122, the first ends of the first panel body 121 and the second panel body 122 are spliced ​​along the first side of the bottom plate 11, the second ends of the first panel body 121 and the second panel body are bent and extended along the adjacent two sides of the first side, and a lateral notch 120 is formed on the opposite side of the first side, and the cover plate 13 includes a top plate portion 131 and a side plate portion 132, the top plate portion 131 is used to cover the top of the installation space, and the side plate portion 132 is used to cover the lateral notch 120. The side panel 12 is configured as two independent panel bodies, which is convenient for storage after disassembly.

[0148] Specifically, the second ends of the first enclosure body 121 and the second enclosure body 122 are both formed with a second connecting portion 124 for splicing with the side plate portion 132, and the second connecting portion 124 and the side plate portion 132 are correspondingly formed with connecting holes for bolts to pass through.

[0149] Furthermore, the second connecting portion 124 of the second enclosure body 122 and the side plate portion 132 are formed with a lock hole corresponding to the side plate portion 132, and the lock hole is used for the operator to set a padlock, see Figure 1 The connecting bolts pass through the bottom plate 11 and the first connecting portion 123 of the side panel 12 from top to bottom and fix the two to the grain elevator. The nuts of the connecting bolts are located in the installation space. Without opening the cover plate 13, the connecting bolts cannot be screwed and disassembled. After the operator covers the cover plate 13 and sets a padlock accordingly, the cover plate 13 and the second panel body 122 cannot be detached, thereby improving the anti-theft performance of the grain monitoring device and preventing the high-speed camera 22 from being stolen.

[0150] Specifically, the first enclosure body 121 and the second enclosure body 122 are spliced ​​with a mortise and tenon structure, and a mortise groove is formed at the first end of one of the two enclosure bodies, and a tenon is formed at the first end of the other enclosure body. This mortise and tenon structure can make the connection between the first enclosure body 121 and the second enclosure body 122 more stable. During the operation of the harvester, the vibrations of the first enclosure body 121 and the second enclosure body 122 are synchronized, which can reduce or even avoid the noise generated by the collision between the two.

[0151] In order to achieve the above-mentioned object, the present invention further provides a harvester, wherein the harvester includes the above-mentioned grain monitoring device. The harvester also includes a vehicle body and a harvesting drum, the vehicle speed of the vehicle body and the rotation speed of the harvesting drum are adjustable in 1-12 gears, the processor is respectively connected to the drivers of the two and the output gear of the drivers of the two is adjusted according to the calculated impurity content and crushing rate, and the beneficial effects are the same as those of the above-mentioned embodiment, which will not be described in detail here.

[0152] In the description of the present invention, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0153] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral one; it can be a mechanical connection, an electrical connection, or communication with each other; it can be a direct connection, or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements, unless otherwise clearly defined. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0154] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. 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 invention. In this specification, the schematic representations of the above terms do 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. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0155] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A method for monitoring grains for a harvester, characterized in that: The harvester grain monitoring method comprises: Capture real-time images of harvested grain on the grain elevator; Determining a first number of pixels occupied by impurities and a second number of pixels occupied by kernels in the harvested grains based on the real-time image; Determine a first mass and a second mass of impurities and seeds, respectively, according to the first number of pixels and the second number of pixels; The impurity content of the harvested grain is determined based on the first mass and the second mass, and the working parameters of the harvester are adjusted when the impurity content exceeds a preset impurity value.

2. The method for monitoring grains for a harvester according to claim 1, characterized in that: Determining the first number of pixels occupied by impurities and the second number of pixels occupied by grains in the harvested grains according to the real-time image includes: The real-time image is input into a preset impurity and fragmentation rate recognition model, so as to obtain the first number of pixels and the second number of pixels through the output of the impurity and fragmentation rate recognition model.

3. The method for monitoring grains for a harvester according to claim 2, characterized in that: The preset construction method of the impurity and breakage rate identification model includes: Pre-collecting a first sample image of a plurality of grain samples, wherein the grain samples include intact grains, broken grains and impurities; Calibrate the intact kernels, broken kernels and impurities in each of the first sample images, and divide the plurality of the first sample images into a type training set and a type verification set; The type training set and the type verification set are input into the image segmentation recognition model to construct and generate the impurity fragmentation rate recognition model.

4. The method for monitoring grains for a harvester according to claim 3, characterized in that: The step of inputting the type training set and the type verification set into an image segmentation recognition model to construct and generate the impurity fragmentation rate recognition model comprises: Inputting the type training set and the type verification set into the image segmentation recognition model, wherein the image segmentation recognition model is configured to recognize images at different pixel ratios; Substituting the feature values ​​output by the image segmentation recognition model for recognizing the image at different pixel ratios into a preset weight calculation formula to obtain a weighted feature value; Determining the type output result of the impurity and breakage rate identification model according to the weighted eigenvalue; The weight coefficient in the weight calculation formula is adjusted using the type output result corresponding to the type verification set to construct and generate the impurity and fragmentation rate recognition model.

5. The method for monitoring grains for a harvester according to claim 4, characterized in that: The image segmentation recognition model is set to recognize images at different pixel ratios and further includes: A plurality of output branch groups for parallel calculation are arranged in the image segmentation and recognition model, wherein each of the output branch groups is used for recognizing an image at a corresponding pixel ratio.

6. The method for monitoring grains for a harvester according to claim 2, characterized in that: The preset construction method of the impurity and breakage rate identification model also includes: Pre-collecting a second sample image of a plurality of grain samples, wherein the grain samples include intact grains, broken grains and impurities; Calibrate the intact kernels, broken kernels and impurities in each of the second sample images, and divide the plurality of the second sample images into a type test set; Inputting the type test set into the impurity and fragmentation rate recognition model to obtain the test impurity rate of the type test set; The preset impurity value is adjusted according to a deviation value between a test impurity rate of the type test set and an actual impurity rate.

7. The method for monitoring grains for a harvester according to claim 1, characterized in that: The determining the first mass and the second mass of the impurities and the seeds respectively according to the first number of pixels and the second number of pixels comprises: The real-time image, the first number of pixels, and the second number of pixels are input into a preset quality estimation model, so as to obtain the first quality and the second quality through the output of the quality estimation model.

8. The method for monitoring grains for a harvester according to claim 7, characterized in that: The preset construction method of the quality estimation model includes: respectively collecting third sample images and weights of a plurality of intact kernel samples, fourth sample images and weights of a plurality of broken kernel samples, and fifth sample images and weights of a plurality of impurity samples; labeling weights for each of the third sample images, each of the fourth sample images, and each of the fifth sample images, respectively, and dividing them into a quality training set and a quality verification set; The quality training set and the quality verification set are input into an image segmentation and recognition model to construct and generate the quality estimation model.

9. The method for monitoring grains for a harvester according to any one of claims 1 to 8, characterized in that: The harvester grain monitoring method further comprises: determining a third quality of broken kernels in the harvested grains according to the number of kernel pixels occupied by broken kernels in the kernels, wherein the kernels include broken kernels and intact kernels; The crushing rate of the harvested grains is determined according to the second mass and the third mass, and the working parameters of the harvester are adjusted when the crushing rate exceeds a preset crushing value.

10. A seed monitoring device, characterized in that: The grain monitoring equipment includes an image analysis device and an image acquisition device. The image analysis device is configured with a processor. The processor is communicatively connected to the image acquisition device and is used to execute the grain monitoring method for a harvester as described in any one of claims 1 to 9. The image acquisition device is used to acquire real-time images.

11. The seed monitoring device according to claim 10, characterized in that: The image acquisition device comprises: An installation shell (1), the installation shell being mounted on a grain elevator of a harvester and forming an installation space having an image acquisition window (10), and the image acquisition window (10) being located on one side of the installation space close to a transmission surface of the grain elevator; An image acquisition component (2), the image acquisition component (2) being arranged in the installation space and comprising a light source (21) and a high-speed camera (22), the light source (21) being capable of emitting visible light for illuminating the image acquisition window (10), the high-speed camera (22) being communicatively connected to the processor, the lens of the high-speed camera (22) being arranged toward the image acquisition window (10), and the high-speed camera (22) being capable of moving toward or away from the image acquisition window (10).

12. A harvester, characterized in that: The harvester comprises the grain monitoring device according to claim 10 or 11.

Images