CN118115881B - Garden plant diseases and insect pests detection method and device - Google Patents
Garden plant diseases and insect pests detection method and device Download PDFInfo
- Publication number
- CN118115881B CN118115881B CN202410490773.6A CN202410490773A CN118115881B CN 118115881 B CN118115881 B CN 118115881B CN 202410490773 A CN202410490773 A CN 202410490773A CN 118115881 B CN118115881 B CN 118115881B
- Authority
- CN
- China
- Prior art keywords
- tree
- path
- insect pests
- detection
- curve
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 99
- 241000607479 Yersinia pestis Species 0.000 title claims abstract description 93
- 201000010099 disease Diseases 0.000 title claims abstract description 83
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 title claims abstract description 83
- 241000238631 Hexapoda Species 0.000 title claims abstract description 68
- 238000005070 sampling Methods 0.000 claims abstract description 81
- 238000000034 method Methods 0.000 claims abstract description 35
- 238000013527 convolutional neural network Methods 0.000 claims description 25
- 238000004140 cleaning Methods 0.000 claims description 10
- 230000001133 acceleration Effects 0.000 claims description 9
- 238000012360 testing method Methods 0.000 claims description 8
- 238000012549 training Methods 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 7
- 230000004913 activation Effects 0.000 claims description 5
- 238000009826 distribution Methods 0.000 claims description 5
- 238000010606 normalization Methods 0.000 claims description 5
- 238000004422 calculation algorithm Methods 0.000 claims description 4
- 238000005516 engineering process Methods 0.000 claims description 4
- 238000004088 simulation Methods 0.000 claims description 3
- 238000012876 topography Methods 0.000 abstract description 4
- 230000006870 function Effects 0.000 description 11
- 238000010586 diagram Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000005520 cutting process Methods 0.000 description 3
- 230000007613 environmental effect Effects 0.000 description 3
- 238000007726 management method Methods 0.000 description 3
- 238000003860 storage Methods 0.000 description 3
- 230000005484 gravity Effects 0.000 description 2
- 230000036541 health Effects 0.000 description 2
- 238000005304 joining Methods 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 230000001953 sensory effect Effects 0.000 description 2
- 101100083446 Danio rerio plekhh1 gene Proteins 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000012888 cubic function Methods 0.000 description 1
- 238000013434 data augmentation Methods 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 230000003631 expected effect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012886 linear function Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000000575 pesticide Substances 0.000 description 1
- 230000035943 smell Effects 0.000 description 1
- 238000011895 specific detection Methods 0.000 description 1
- 239000007921 spray Substances 0.000 description 1
- 230000007480 spreading Effects 0.000 description 1
- 238000003892 spreading Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Abstract
The invention relates to the technical field of pest and disease detection, in particular to a method and a device for detecting garden pest and disease. The detection method comprises the following steps: s1: planning a detection path: s11: constructing a three-dimensional map based on the terrain and the tree position; s12: setting three sampling points for each tree, and constructing an initial moving path passing through each sampling point in sequence; s13: updating the initial moving path according to the characteristic data of the detection equipment, so that the time for the detection equipment to pass through all sampling points is shortest, and the optimal moving path is obtained; s2: and the detection equipment runs along the moving path, judges whether the tree has diseases and insect pests when each sampling point is reached, and records and samples if the tree has the diseases and insect pests. According to the invention, the self performance of the detection equipment is fully considered, and the moving path capable of acquiring the image of each tree is generated by combining with the actual topography, so that the efficiency and the precision of pest and disease detection are improved, and the running stability and the running safety of the detection equipment are also improved.
Description
Technical Field
The invention relates to the technical field of plant diseases and insect pests, in particular to a detection method and a detection device for garden plant diseases and insect pests.
Background
The detection method of garden plant diseases and insect pests mainly comprises sensory detection and instrument detection. Sensory testing is a traditional method of detecting pests by human vision, hearing, smell and touch. The instrument detection utilizes the principles of optics, electricity, chemistry and the like to detect the plant diseases and insect pests through modern instruments and equipment. The specific detection method will vary depending on the type of pest and environmental conditions. The significance of the detection of the plant diseases and insect pests is that the plant diseases and insect pests are found in time so as to take corresponding control measures, thus further spreading of the plant diseases and insect pests can be avoided, loss is reduced, meanwhile, the ecological environment is protected, the use of chemical pesticides is reduced, and ecological balance is realized. Through regular detection, the health of garden plants can be maintained, the ornamental value is improved, and a healthy ecological environment is provided for citizens.
However, when detection is performed manually or by an instrument, each tree in gardens needs to be observed or analyzed manually, so that the amount of manual labor is excessive, the detection efficiency is low, and misjudgment is easy to occur. Some existing garden management adopts an image recognition method to detect garden trees, however, due to lack of comprehensive photographing of the garden trees during detection, missed detection is easy to occur, or due to excessively fine photographing, the image processing amount is excessively large, and therefore the final detection precision and efficiency are difficult to achieve the expected effect. Moreover, although the existing image recognition technology is mature, when the plant diseases and insect pests are detected aiming at trees of different types, the problems of low recognition precision and low efficiency still exist, and the image recognition still has errors and cannot be completely used as the unique standard of the plant diseases and insect pests.
Disclosure of Invention
Based on the above, it is necessary to provide a method and a device for detecting a garden pest, which aims at the problems of low precision and low efficiency of pest detection caused by low precision and low efficiency of image recognition of the pest and the problems of large workload of manually detecting the pest and easy misjudgment in the existing garden management.
The invention is realized by the following technical scheme: the detection method of the garden plant diseases and insect pests comprises the following steps:
s1: and planning a detection path. The specific method comprises the following steps:
S11: a three-dimensional map based on terrain and tree position is constructed.
S12: three sampling points are set for each tree, and an initial moving path passing through each sampling point in turn is constructed.
S13: and updating the initial moving path according to the characteristic data of the detection equipment, so that the time for the detection equipment to pass through all the sampling points is shortest, and the optimal moving path is obtained.
S2: and the detection equipment runs along the moving path, judges whether the tree has diseases and insect pests when each sampling point is reached, and records and samples if the tree has the diseases and insect pests. The method for judging whether the tree has diseases and insect pests concretely comprises the following steps:
s21: the detection equipment reaches the sampling point, and the tree is photographed to obtain a tree image.
S22: and (3) primarily identifying the tree image, judging whether the tree has diseases and insect pests, if so, identifying the positions of the diseases and insect pests, and if not, moving to the next sampling point.
S23: and (3) primarily identifying the trees with the diseases and insect pests, cleaning the positions of the diseases and insect pests, and photographing to obtain the images of the diseases and insect pests.
S24: and identifying and verifying the disease and pest image, judging whether the tree has disease and pest, if so, recording and sampling, otherwise, moving to the next sampling point.
S25: and judging whether all sampling points are identified, if yes, ending and transporting the collected samples to a manual detection point, otherwise, returning to the step S21.
According to the detection method, the self performance of the detection equipment is fully considered, the actual topography is combined, the moving path capable of acquiring each tree image is generated, the highest speed is set at each sampling point, the detection equipment passes through the sampling points at low speed, so that the performance requirement on photographing equipment is reduced, clear and stable tree images are photographed, the efficiency and the precision of pest and disease detection are improved, meanwhile, the detection equipment always needs to turn through the sampling points, and the running stability and the running safety of the detection equipment can be improved through the sampling points at low speed.
Further, in step S12, assuming that the coordinates of a tree are X i（xi,yi,zi), and the detection device moves from the third sampling point of the last tree to the three sampling points of the tree, the first sampling point of the tree is set as: x i1（xi+rcosθ1,yi+rcosθ2,zi+rcosθ3). The other two sampling points are set to ：Xi2=（xi+rcos（θ1+2π/3）,yi+rcos（θ2+2π/3）,zi+rcos（θ3+2π/3））、Xi3=（xi+rcos（θ1-2π/3）,yi+rcos（θ2-2π/3）,zi+rcos（θ3-2π/3）） or Xi2=（xi+rcos（θ1-2π/3）,yi+rcos（θ2-2π/3）,zi+rcos（θ3-2π/3））、Xi3=（xi+rcos（θ1+2π/3）,yi+rcos（θ2+2π/3）,zi+rcos（θ3+2π/3））.
Wherein r is a preset photographing distance, namely the distance between each tree and the corresponding sampling point, and theta 1、θ2、θ3 is the included angle between the line where the third sampling point of the last tree and the current tree are located and the x axis, the y axis and the z axis respectively.
Further, in step S13, the method for generating the optimal movement path is as follows:
s131: and updating the initial moving path by adopting a pure curve path or a straight curve combined path.
The calculation formula of the total length of the pure curve path is as follows:
；
Where N is the number of sampling points, i=1, 2, …, N, y '=f' (x) is the derivative of the corresponding curve, and p i、pi+1 is the two endpoints of the i-th curve.
The time for the detection device to traverse all pure curve paths is:
；
Where v maxi is the maximum speed of the detection device in the ith curve, v maxi≤vmax,di is the path length of the detection device traveling at a constant speed in the ith curve, s i is the length of the ith curve, v 0 is the initial speed, a C is the acceleration, and a d is the deceleration acceleration.
The calculation formula of the total length of the straight-curved combined path is as follows:
；
Where d li is the length of the line segment portion in the straight-curved joining path, and p i1、pi2 is the two ends of the curve portion in the straight-curved joining path.
The time for the detection device to pass through the straight and curved combination path is as follows:
；
where a l is the linear acceleration and v i+1 is the end point velocity of the i-th curve.
S132: selecting an optimal route in a pure curve path and a straight curve combined path, namely solving the shortest passing time T, wherein the formula is as follows:
；
where min () is the minimum, T Cmin is the shortest transit time of the pure curve path, and T Lmin is the shortest transit time of the straight curve combined path.
Further, in step S22, the method of preliminary identification includes the steps of:
S221: and constructing a basic AlexNet convolutional neural network model.
S222: by adding a normalization algorithm, replacing an activation function and adopting a random discarding technology, the AlexNet convolutional neural network is improved.
S223: and collecting a disease and insect pest image set, dividing the image set into a training set and a testing set, training the AlexNet convolutional neural network model, and reserving model parameters meeting the testing precision.
S224: dividing the shot tree image into M x N small images, inputting the small images into AlexNet convolutional neural network models respectively, and outputting detection results.
The invention also provides a detection device for the garden plant diseases and insect pests, which comprises a moving module, a camera module, a path planning module, an identification module, a cleaning module and a sampling module.
The moving module is used for driving the detecting device to integrally move. The camera module is used for collecting tree images. The path planning module is used for generating a simulation map of the garden to be detected and generating a moving path according to tree distribution in the garden. The identification module adopts AlexNet convolutional neural network model to judge whether each tree has plant diseases and insect pests. The cleaning module is used for cleaning the plant diseases and insect pests of the trees. The sampling module is used for sampling trees with diseases and insect pests.
Compared with the prior art, the invention has the following beneficial effects:
According to the invention, the self performance of the detection equipment is fully considered, the actual topography is combined, a moving path capable of acquiring each tree image is generated, and the highest speed is set at each sampling point, so that the detection equipment passes through the sampling points at a low speed, the performance requirement on photographing equipment is reduced, clear and stable tree images are photographed, the efficiency and the precision of pest and disease detection are improved, meanwhile, the detection equipment always needs to turn through the sampling points, and the running stability and the running safety of the detection equipment can be improved through the sampling points at a low speed.
The improved convolutional neural network model is adopted to identify the plant diseases and insect pests, and compared with the traditional convolutional neural network model, the improved convolutional neural network model has higher identification accuracy and higher reliability. In the detection process, the tree image is primarily identified, and after the positions of the diseases and insect pests are cleaned, the identification is identified and verified again, so that the identification error caused by the environmental problem is effectively eliminated, and the identification precision is simply and rapidly improved.
Drawings
FIG. 1 is a step diagram of a method for detecting garden plant diseases and insect pests in embodiment 1 of the present invention;
FIG. 2 is a schematic diagram of a pure curve path;
FIG. 3 is a schematic diagram of a straight-curved joint path;
fig. 4 is a step diagram of the method of initially identifying pests of fig. 1.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
It is noted that when an element is referred to as being "mounted to" another element, it can be directly on the other element or intervening elements may also be present. When an element is referred to as being "disposed on" another element, it can be directly on the other element or intervening elements may also be present. When an element is referred to as being "secured to" another element, it can be directly secured to the other element or intervening elements may also be present.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "or/and" as used herein includes any and all combinations of one or more of the associated listed items.
Example 1: referring to fig. 1, the embodiment provides a method for detecting garden plant diseases and insect pests, the method comprises the following steps:
s1: and (3) detecting path planning: generating a garden map according to the distribution of garden trees, and generating a basic moving path in the map by adopting a manual selection or intelligent generation mode. The detection equipment automatically updates the moving path according to road conditions and detection requirements in the moving process.
S11: and constructing a three-dimensional map. And acquiring the position of each tree and the ground height in the garden according to the garden construction data, and constructing a three-dimensional map based on the terrain and the tree position. In the three-dimensional map, the position, growth information, and the like of each tree can also be marked. The growth information may include tree type, age, height, health information, etc.
S12: three sampling points are set for each tree, and an initial moving path passing through each sampling point in turn is constructed. Wherein, three sampling points form an equilateral triangle taking the tree as the center.
Assuming that a tree has a coordinate of X i（xi,yi,zi) and the center of gravity of the detection device is X 0（xi-1,yi-1,zi-1), and the detection device is moved toward the tree from the third sampling point of the last tree, the first sampling point of the tree is set to: x i1（xi+rcosθ1,yi+rcosθ2,zi+rcosθ3), wherein r is a preset photographing distance, that is, the distance between each tree and the corresponding sampling point, θ 1、θ2、θ3 is the included angle between the line where the third sampling point of the last tree and the current tree are located and the X axis, the y axis and the z axis, and the calculation formula is as follows:
；
The specific coordinates of the other two sampling points set to ：Xi2=（xi+rcos（θ1+2π/3）,yi+rcos（θ2+2π/3）,zi+rcos（θ3+2π/3））、Xi3=（xi+rcos（θ1-2π/3）,yi+rcos（θ2-2π/3）,zi+rcos（θ3-2π/3）） or Xi2=（xi+rcos（θ1-2π/3）,yi+rcos（θ2-2π/3）,zi+rcos（θ3-2π/3））、Xi3=（xi+rcos（θ1+2π/3）,yi+rcos（θ2+2π/3）,zi+rcos（θ3+2π/3））,Xi2、Xi3 may be determined based on the position of the next tree, where X i3 is located closer to the next tree.
S13: and updating the initial moving path according to the characteristic data of the detection equipment, so that the time for the detection equipment to pass through each sampling point is shortest, and the optimal moving path is obtained. Wherein the moving speed of the detection device is not higher than a preset moving speed when passing each sampling point. The specific method comprises the following steps:
s131: and constructing a space coordinate system by taking the gravity center of the detection equipment as an origin and the moving direction as an X axis, and calculating the optimal path of the detection equipment passing through the current tree according to the positions of the front tree and the rear tree.
Referring to fig. 2 and 3, the maximum steering angle is set to phi, the maximum speed v max, and the maximum speed passing through the detection point is set to v max1. The travel path of the detection device may be a pure curved path or a straight curved combined path.
The pure curve path is suitable for path planning with smaller tree spacing, and the corresponding path can calculate the corresponding path function by adopting a cubic spline difference method, and is expressed as:
y=ax3+bx2+cx+d，m＜x≤n；
Wherein a, b, c, d is the cubic function coefficient of the curve, and m and n are the two endpoints of the curve.
In this embodiment, assuming that a curve path is constructed for the sampling points X i-1、Xi、Xi+1, a point is arbitrarily selected between X i-2、Xi-1、Xi、Xi+1、Xi+2 and denoted as p 1、p2、p3、p4, p 1、Xi-1、p2 is substituted into the pure curve path function to obtain a first section of curve path, p 2、Xi、p3 is substituted into the pure curve path function to obtain a second section of curve path, p 3、Xi+1、p4 is substituted into the pure curve path function to obtain a third section of curve path, the curvature at p 2、p3 is the same, the corresponding curvature k is less than or equal to tan phi, and so on, to obtain a curve function passing through all the sampling points.
In a pure curve path, the calculation formula of the total path length is as follows:
；
where N is the number of sampling points, i=1, 2, …, N, y '=f' (x) is the derivative of the corresponding curve.
Assuming that the initial speed of the detection device from the last sampling point is v 0, the acceleration is a C, the deceleration acceleration is a d, and the time of the corresponding detection device passing through all curve paths is recorded as follows:
；
；
Wherein v maxi is the maximum speed of the detection device in the ith section of curve, v maxi≤vmax,di is the path length of the detection device in the ith section of curve, p i、pi+1 is two end points of the ith section of curve respectively, and s i is the length of the ith section of curve.
The straight-curved combined path is suitable for planning paths with larger tree spacing, and the whole path comprises a line segment part and a curve part, wherein the line segment part can be expressed as:
；
In the formula, e and f are linear function coefficients where line segments are located.
The curve part is also calculated by adopting a cubic spline interpolation method, the corresponding curve end points are the end points and the start points of the front and rear line segments, the curve is tangent to the front and rear line segments, and the middle of the curve passes through the corresponding sampling points.
In the straight-curved combined path, the straight line segment and the curve segment corresponding to each sampling point xi share three endpoints, which are sequentially marked as p i0、pi1、pi2, and the calculation formula of the total path length is as follows:
；
where d li is the linear part length, i.e. the distance between two points of p i0、pi1.
The time for the corresponding detection device to pass through all paths is recorded as:
；
Where a l is the linear acceleration, v i+1 is the final speed of the i-th curve, and is also the initial speed of the i+1-th curve.
S132: the optimal route is selected, namely the shortest passing time T is solved, and the formula is as follows:
；
where min () is the minimum, T Cmin is the shortest transit time of the pure curve path, and T Lmin is the shortest transit time of the straight curve combined path.
The curve path corresponding to the shortest passing time is the optimal moving path.
S2: and the detection equipment runs along the moving path, judges whether the tree has diseases and insect pests when each sampling point is reached, and records and samples if the tree has the diseases and insect pests.
The method for judging whether the tree has diseases and insect pests concretely comprises the following steps:
s21: the detection equipment reaches the sampling point, and the tree is photographed to obtain a tree image.
S22: and (3) primarily identifying the tree image, judging whether the tree has diseases and insect pests, if so, identifying the positions of the diseases and insect pests, and if not, moving to the next sampling point.
Please refer to fig. 4, wherein the method of preliminary identification includes the following steps:
S221: a base AlexNet convolutional neural network model is constructed, which includes an input layer, an implied layer, and an output layer, wherein the implied layer includes a fully connected layer, a pooled layer, a convolutional layer, an activation function layer, and a Softmax layer.
The Softmax layer determines the optimal classification method by calculating the probability, and the formula is:
；
where Zi is the i-th element in the original vector Z, i=1, 2, …, n.
S222: the AlexNet convolutional neural network is improved. In this embodiment, the improvement of AlexNet convolutional neural networks is achieved by adding a normalization algorithm, replacing the activation function, and employing a random discard technique.
Adding a batch normalization algorithm in the first five layers, firstly, calculating the average sum variance of the whole data set, and processing the average sum variance by normalization data, wherein the formula is expressed as follows:
；
In the method, in the process of the invention, And δ are the mean and variance, respectively, of the whole dataset and ε is a coefficient for increasing the stability of the values.
Introducing the stretchable parameter and the offset parameter for correction, the normalized output y j can be expressed as:
；
Wherein, gamma and beta are the stretchable parameter and the offset parameter respectively, which are updated along with the updating of the network weight.
ReLU6 is used as the activation function. The formula for the ReLU6 function is:
；
In the formula, min () is the minimum value, and max () is the maximum value.
A random discard technique is added at the sixth and seventh layers. The data operation amount can be reduced by random discarding, and the operation efficiency is improved.
The phenomenon of under fitting of the model can be avoided by adopting a random discarding technology. This embodiment is implemented by changing the structure of the convolutional neural network, i.e., setting the neuron number value of a certain layer to 0, so that forward and backward propagation is not performed as if it were deleted in the network, thus effectively reducing single neurons in training, and avoiding overfitting by data augmentation.
S223: and collecting a disease and insect pest image set, dividing the image set into a training set and a testing set, training the AlexNet convolutional neural network model, and reserving model parameters meeting the testing precision.
The plant disease and insect pest image set can be obtained from garden management data and network big data, and can also be obtained by direct shooting. In this embodiment, 200 images of each plant disease and insect pest are obtained by combining network big data with direct shooting, and the number of images is increased by image enhancement, so that AlexNet convolutional neural network models are trained, and model parameters meeting the test precision are reserved.
S224: dividing the shot tree image into M x N small images, inputting the small images into AlexNet convolutional neural network models respectively, and outputting detection results.
The pest is not evident in the early manifestation, especially in the tree image, the exposed pest area is smaller compared to the whole tree image. If the tree image is detected as a whole, the accuracy of identifying the plant diseases and insect pests is not high, and the plant diseases and insect pests occupy a larger proportion in a smaller image by dividing the area of the tree, so that the detection accuracy is improved.
S23: and (3) primarily identifying the trees with the diseases and insect pests, cleaning the positions of the diseases and insect pests, and photographing to obtain the images of the diseases and insect pests.
S24: and identifying and verifying the disease and pest image, judging whether the tree has disease and pest, if so, recording and sampling, otherwise, moving to the next sampling point. Inputting the shot plant disease and insect pest images into a AlexNet convolutional neural network model after training, and outputting a detection result.
S25: and judging whether all sampling points are identified, if yes, ending and transporting the collected samples to a manual detection point, otherwise, returning to the step S21.
According to the embodiment, three sampling points are arranged for each tree, so that an appearance image based on the whole tree can be acquired as much as possible, and then a moving path is constructed according to the position of the tree, so that the whole moving path is shorter, the speed is higher, and the efficiency of detecting the diseases and insect pests of the garden tree is effectively improved. In general, the embodiment fully considers the self performance of the detection equipment, combines the actual topography, generates a moving path capable of acquiring each tree image, and sets the highest speed at each sampling point, so that the detection equipment passes through the sampling points at a low speed, thereby reducing the performance requirement on photographing equipment, photographing clear and stable tree images, improving the efficiency and precision of pest and disease detection, simultaneously, the detection equipment always needs to turn through the sampling points, and the running stability and safety of the detection equipment can be improved through the sampling points at a low speed.
In the embodiment, the improved convolutional neural network model is adopted to identify the plant diseases and insect pests, and compared with the traditional convolutional neural network model, the method has the advantages of higher identification accuracy and higher reliability. In the detection process, the tree image is primarily identified, and after the positions of the diseases and insect pests are cleaned, the identification is identified and verified again, so that the identification error caused by the environmental problem is effectively eliminated, and the identification precision is simply and rapidly improved.
Example 2: the embodiment provides a detection device for garden plant diseases and insect pests, which can adopt the detection method for garden plant diseases and insect pests of embodiment 1 to control, and the detection device comprises a mobile module, a camera module, a path planning module, an identification module and a sampling module.
The moving module is used for driving the detecting device to integrally move. In this embodiment, the mobile module may be a mobile robot or an unmanned aerial vehicle. The mobile robot or the unmanned aerial vehicle has the function of autonomously planning paths, and in actual operation, the path control of the mobile module can be realized only by inputting simple paths for tree distribution positions in gardens. The mobile robot or unmanned aerial vehicle is also provided with a detection radar which is used for detecting obstacle information on one hand and detecting the distance between the tree and the mobile robot or unmanned aerial vehicle on the other hand so as to determine the photographing or sampling position.
The camera module is used for collecting tree images. The general mobile robot or unmanned aerial vehicle is from taking the module of making a video recording, can directly realize the collection to trees image and plant diseases and insect pests image. When the image is acquired, the position and the posture of the tree can be adjusted by combining radar detection, so that the front photographing of the tree is realized, and the identifiable degree of the image is improved.
The path planning module is used for generating a simulation map of the garden to be detected and generating a moving path according to tree distribution in the garden. The path planning method is as described in embodiment 1.
The identification module adopts AlexNet convolutional neural network model to judge whether each tree has plant diseases and insect pests. The method of constructing AlexNet convolutional neural network model is described in example 1.
The cleaning module is used for cleaning the plant diseases and insect pests of the trees, can be directly installed on the mobile module, and generally consists of a water storage container, a pipeline assembly and a spray head. The pipeline component comprises an electronic valve, the valve can be controlled to be opened according to the identification result, and the positions of the tree diseases and insect pests are cleaned.
The sampling module is used for sampling trees with diseases and insect pests. The sampling module comprises a cutting module and a storage module. The cutting module is used for cutting off local samples of pest and disease damage positions on the trees, and the storage module is used for storing the samples and marking.
The technical features of the above-described embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above-described embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The above examples illustrate only a few embodiments of the invention, which are described in detail and are not to be construed as limiting the scope of the invention. It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the invention, which are all within the scope of the invention. Accordingly, the scope of protection of the present invention is to be determined by the appended claims.
Claims (4)
1. The detection method for garden diseases and insect pests is characterized by comprising the following steps:
s1: planning a detection path; the specific method comprises the following steps:
s11: constructing a three-dimensional map based on the terrain and the tree position;
S12: setting three sampling points for each tree, and constructing an initial moving path passing through each sampling point in sequence;
S13: updating the initial moving path according to the characteristic data of the detection equipment, so that the time for the detection equipment to pass through all sampling points is shortest, and the optimal moving path is obtained;
the generation method of the optimal moving path is as follows:
s131: updating the initial moving path by adopting a pure curve path or a straight curve combined path;
The calculation formula of the total length of the pure curve path is as follows:
；
Wherein N is the number of sampling points, i=1, 2, …, N, y '=f' (x) is the derivative of the corresponding curve, and p i、pi+1 is the two endpoints of the i-th curve respectively;
the time for the detection device to traverse all pure curve paths is:
；
Wherein v maxi is the maximum speed of the detection device in the ith section of curve, v maxi≤vmax,di is the path length of the detection device in the ith section of curve, s i is the length of the ith section of curve, v 0 is the initial speed, a C is the acceleration, and a d is the deceleration acceleration;
The calculation formula of the total length of the straight-curved combined path is as follows:
；
Wherein d li is the length of the line segment part in the straight-curved combined path, and p i1、pi2 is the two end points of the curve part in the straight-curved combined path;
The time for the detection device to pass through the straight and curved combination path is as follows:
；
Wherein a l is the linear section acceleration, v i+1 is the end point speed of the ith section curve;
s132: selecting an optimal route in a pure curve path and a straight curve combined path, namely solving the shortest passing time T, wherein the formula is as follows:
；
In the formula, min () is the minimum value, T Cmin is the shortest passing time of a pure curve path, and T Lmin is the shortest passing time of a straight curve combined path;
S2: the detection equipment runs according to the moving path, judges whether the tree has diseases and insect pests when reaching each sampling point, and records and samples if the tree has the diseases and insect pests; the method for judging whether the tree has diseases and insect pests concretely comprises the following steps:
s21: the detection equipment reaches a sampling point, and photographs the tree to obtain a tree image;
S22: the tree image is primarily identified, whether the tree has diseases and insect pests is judged, if yes, the positions of the diseases and insect pests are identified, and if not, the tree is moved to the next sampling point;
S23: for the trees with the diseases and insect pests in the primary identification, cleaning and photographing positions of the diseases and insect pests to obtain images of the diseases and insect pests;
s24: identifying and verifying the disease and pest image, judging whether the tree has disease and pest, if so, recording and sampling, otherwise, moving to the next sampling point;
S25: and judging whether all sampling points are identified, if yes, ending and transporting the collected samples to a manual detection point, otherwise, returning to the step S21.
2. A method for detecting a garden pest and disease damage according to claim 1, wherein in step S12, assuming that a tree has a coordinate X i（xi,yi,zi, and the detecting device moves from the third sampling point of the previous tree to the three sampling points of the tree, the first sampling point of the tree is set as follows: x i1（xi+rcosθ1,yi+rcosθ2,zi+rcosθ3); the other two sampling points are set to ：Xi2=（xi+rcos（θ1+2π/3）,yi+rcos（θ2+2π/3）,zi+rcos（θ3+2π/3））、Xi3=（xi+rcos（θ1-2π/3）,yi+rcos（θ2-2π/3）,zi+rcos（θ3-2π/3）） or Xi2=（xi+rcos（θ1-2π/3）,yi+rcos（θ2-2π/3）,zi+rcos（θ3-2π/3））、Xi3=（xi+rcos（θ1+2π/3）,yi+rcos（θ2+2π/3）,zi+rcos（θ3+2π/3））;
Wherein r is a preset photographing distance, namely the distance between each tree and the corresponding sampling point, and theta 1、θ2、θ3 is the included angle between the line where the third sampling point of the last tree and the current tree are located and the x axis, the y axis and the z axis respectively.
3. A method for detecting a garden pest and disease damage according to claim 1, wherein in step S22, the preliminary identifying method includes the steps of:
s221: constructing a basic AlexNet convolutional neural network model;
S222: by adding a normalization algorithm, replacing an activation function and adopting a random discarding technology, improving AlexNet convolutional neural networks;
S223: collecting a disease and insect pest image set, dividing the image set into a training set and a testing set, training a AlexNet convolutional neural network model, and reserving model parameters meeting the testing precision;
s224: dividing the shot tree image into M x N small images, inputting the small images into AlexNet convolutional neural network models respectively, and outputting detection results.
4. A detection apparatus for a garden pest as claimed in any one of claims 1 to 3, wherein the detection apparatus comprises:
the moving module is used for driving the detection device to integrally move;
The camera module is used for collecting tree images;
the path planning module is used for generating a simulation map of the garden to be detected and generating a moving path according to the tree distribution in the garden;
The identification module is used for judging whether each tree has diseases and insect pests or not by adopting AlexNet convolutional neural network models;
the cleaning module is used for cleaning the positions of the diseases and insect pests of the trees;
And the sampling module is used for sampling the trees with diseases and insect pests.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410490773.6A CN118115881B (en) | 2024-04-23 | Garden plant diseases and insect pests detection method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410490773.6A CN118115881B (en) | 2024-04-23 | Garden plant diseases and insect pests detection method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN118115881A CN118115881A (en) | 2024-05-31 |
| CN118115881B true CN118115881B (en) | 2024-06-28 |
Family
ID=
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103869820A (en) * | 2014-03-18 | 2014-06-18 | 北京控制工程研究所 | Ground navigation planning control method of rover |
| AU2020103613A4 (en) * | 2020-11-23 | 2021-02-04 | Agricultural Information and Rural Economic Research Institute of Sichuan Academy of Agricultural Sciences | Cnn and transfer learning based disease intelligent identification method and system |
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103869820A (en) * | 2014-03-18 | 2014-06-18 | 北京控制工程研究所 | Ground navigation planning control method of rover |
| AU2020103613A4 (en) * | 2020-11-23 | 2021-02-04 | Agricultural Information and Rural Economic Research Institute of Sichuan Academy of Agricultural Sciences | Cnn and transfer learning based disease intelligent identification method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108496127B (en) | Efficient three-dimensional reconstruction focused on an object | |
| Subramanian et al. | Sensor fusion using fuzzy logic enhanced kalman filter for autonomous vehicle guidance in citrus groves | |
| CN109755995B (en) | Robot automatic charging docking method based on ROS robot operating system | |
| Silwal et al. | Bumblebee: A Path Towards Fully Autonomous Robotic Vine Pruning. | |
| CN113701780B (en) | Real-time obstacle avoidance planning method based on A star algorithm | |
| CN114460941B (en) | Robot path planning method and system based on improved sparrow search algorithm | |
| CN110181508A (en) | Underwater robot three-dimensional Route planner and system | |
| CN114237235B (en) | Mobile robot obstacle avoidance method based on deep reinforcement learning | |
| CN113110513A (en) | ROS-based household arrangement mobile robot | |
| CN112344945A (en) | Indoor distribution robot path planning method and system and indoor distribution robot | |
| Lin et al. | Wall-following and navigation control of mobile robot using reinforcement learning based on dynamic group artificial bee colony | |
| Li et al. | Learning view and target invariant visual servoing for navigation | |
| CN110610130A (en) | Multi-sensor information fusion power transmission line robot navigation method and system | |
| Menon et al. | NBV-SC: Next best view planning based on shape completion for fruit mapping and reconstruction | |
| CN116540731A (en) | Path planning method and system integrating LSTM and SAC algorithms | |
| Nie et al. | A forest 3-D LiDAR SLAM system for rubber-tapping robot based on trunk center atlas | |
| Short et al. | Abio-inspiredalgorithminimage-based pathplanning and localization using visual features and maps | |
| CN118115881B (en) | Garden plant diseases and insect pests detection method and device | |
| CN114667852A (en) | Hedge trimming robot intelligent cooperative control method based on deep reinforcement learning | |
| CN111310919B (en) | Driving control strategy training method based on scene segmentation and local path planning | |
| CN111950524B (en) | Orchard local sparse mapping method and system based on binocular vision and RTK | |
| CN118115881A (en) | Garden plant diseases and insect pests detection method and device | |
| CN116540709A (en) | Obstacle avoidance path planning method based on robot formation | |
| CN116563341A (en) | Visual positioning and mapping method for processing dynamic object in complex environment | |
| CN115619953A (en) | Rugged terrain-oriented mobile robot terrain mapping method and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant |