Disease area detection method, disease detection method and device for plant leaves

By acquiring multi-spectral leaf images and laser point cloud data, the color, texture and shape characteristics of plant leaves are extracted, and the disease area is identified, which solves the problem of insufficient spectral resolution in the existing technology, and improves the accuracy and detection effect of disease recognition .

CN119064295APending Publication Date: 2024-12-03中交和美环境生态建设有限公司
View PDF 0 Cites 4 Cited by

Patent Information

Application Number
CN202411078394.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-07
Publication Date
2024-12-03

AI Technical Summary

Technical Problem

In the recognition of plant diseases, the plant images captured by the camera cannot capture the spectral range except for the three basic color channels of red, green and blue, which makes it difficult to effectively identify the disease area when the disease features are in an uncaptured or difficult to capture color channels. , affecting the judgment of the type and severity of the disease.

Method used

By obtaining multispectral leaf images of the target plant canopy leaves, including information of green light, red light, near infrared, red edge and short-wave infrared bands, the background image is separated, the disease area image is determined, and the color is extracted by combining laser point cloud data. , texture and shape characteristics, identify the types and levels of diseases.

Benefits of technology

It improves the accuracy of identification of plant leaf disease areas, can reveal nuances on the leaf surface, and facilitates subsequent detection of disease type and severity.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119064295A_ABST
    Figure CN119064295A_ABST
Patent Text Reader

Abstract

The invention discloses a plant leaf disease area detection method, a plant leaf disease detection method and a plant leaf disease detection device. The plant leaf disease area detection method comprises the following steps: acquiring a multispectral leaf image of a target plant canopy leaf; the multispectral leaf image covers various wave band information such as a green light wave band, a red light wave band, a near-infrared wave band, a red edge wave band, an infrared wave band and a short wave infrared wave band; separating a background image in the multispectral leaf image, and extracting a target plant leaf image; and identifying and determining a disease area image in the target plant leaf image according to the various wave band information of the target plant leaf image. The multispectral leaf image can be used for revealing the slight difference of the leaf surface, so that the accuracy of identifying the leaf disease area can be improved, and powerful support is provided for subsequent plant leaf disease type and severity detection of the identified disease area.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of plant disease identification, and specifically to a disease area detection method, a disease detection method and a device for plant leaves. Background Art

[0002] At present, in the field of plant disease identification technology, cameras are often used to capture plant images, and the plant images are analyzed in one step to identify and distinguish the diseased areas of plants. However, when using cameras to capture plant images at this stage, the ordinary cameras used can provide high spatial resolution, but are limited in spectral resolution. They can only capture information from the three basic color channels of red, green and blue (RGB). They are powerless for other important spectral ranges in biological or scientific research (such as ultraviolet or infrared rays), and the dynamic range of camera sensors is limited, which may not be able to capture the details of the brightest and darkest parts of the scene. Therefore, in the current technical solution of capturing plant images with cameras and then identifying them, if the disease characteristics are in a color channel that cannot be captured or is difficult to capture, the areas where these diseases are located will be difficult to be effectively identified, which will affect the judgment of the disease type and severity based on the shape and location of the diseased area in the subsequent stage. Summary of the invention

[0003] The present application provides a plant leaf disease area detection method, a disease detection method and a device, which can solve the problems existing in the above-mentioned prior art.

[0004] In a first aspect, the present application provides a method for detecting diseased areas of plant leaves, using the following technical solutions:

[0005] A method for detecting a diseased area of ​​a plant leaf, the method comprising:

[0006] Acquire a multispectral leaf image of the canopy leaves of the target plant; the multispectral leaf image includes at least a plurality of band information including a green light band, a red light band, a near infrared band, a red edge band, an infrared band and a short-wave infrared band;

[0007] Separating the background image from the multispectral leaf image to obtain a target plant leaf image;

[0008] According to the multiple band information of the target plant leaf image, a diseased area image in the target plant leaf image is determined.

[0009] In combination with the first aspect, in one implementation, determining the diseased area image in the target plant leaf image according to the multiple band information of the target plant leaf image comprises the following steps:

[0010] Separating the target plant leaf image into a discolored leaf image and a normal leaf image according to multiple band information in the target plant leaf image;

[0011] Separating the leaf surface and the leaf tip of the discolored leaf image to obtain a diseased leaf surface image and a discolored leaf tip image;

[0012] Determining a diseased leaf tip image in the discolored leaf tip image according to an edge grayscale value of the discolored leaf tip image;

[0013] The diseased area image in the target plant leaf image is obtained according to the diseased leaf surface image and the diseased leaf tip image.

[0014] In combination with the first aspect, in one embodiment, determining the diseased leaf tip image in the discolored leaf tip image according to the edge grayscale value of the discolored leaf tip image comprises the following steps:

[0015] sharpening the red and green color channels in the color-changing leaf tip image;

[0016] Determine whether there is any pixel point along the edge of the leaf tip region whose grayscale value difference with the adjacent leaf surface pixel point is greater than the set grayscale value range;

[0017] If it exists, it is determined that the image corresponding to the leaf tip area belongs to the diseased leaf tip image.

[0018] In combination with the first aspect, in one embodiment, before separating the background image in the multispectral leaf image to obtain the target plant leaf image, the following steps are included:

[0019] The multispectral leaf image is filtered and denoised.

[0020] In a second aspect, the present application provides a method for detecting plant leaf diseases, which adopts the following technical solution:

[0021] A method for detecting plant leaf diseases, comprising:

[0022] Acquire a diseased area image of a target plant obtained according to the diseased area detection method of plant leaves as described above;

[0023] The leaf disease grade of the target plant is determined according to the pixel ratio between the diseased area image and the canopy leaves of the target plant.

[0024] In a third aspect, the present application provides a method for detecting plant leaf diseases, which adopts the following technical solution:

[0025] A method for detecting plant leaf diseases, characterized in that it comprises:

[0026] Acquire a diseased area image of the target plant obtained according to the plant leaf diseased area detection method as described above and laser point cloud data of the target plant;

[0027] According to the diseased area image, the color characteristics of the target plant leaves and the diseased area shooting data are obtained;

[0028] According to the laser point cloud data, the texture characteristics of the target plant leaves and the diseased area scanning data are obtained;

[0029] According to the photographed data and scanned data of the diseased area, the shape characteristics of the diseased area are obtained;

[0030] The type and grade of the disease on the target plant leaves are determined based on the color characteristics, texture characteristics and shape characteristics of the diseased area.

[0031] In conjunction with the third aspect, in one embodiment, determining the disease type and grade of the target plant leaf according to the color feature, texture feature, and disease area shape feature comprises the following steps:

[0032] Obtaining a feature recognition vector according to the color feature, texture feature and shape feature of the diseased area;

[0033] The leaf disease type and disease grade of the target plant are determined according to the results of the feature recognition vector in a plurality of preset vector recognition models.

[0034] In a fourth aspect, the present application provides a device for detecting diseased areas of plant leaves, which adopts the following technical solutions:

[0035] A device for detecting a diseased area of ​​a plant leaf, the device comprising:

[0036] An acquisition module is configured to acquire a multispectral leaf image of a canopy leaf of a target plant; the multispectral leaf image includes at least a plurality of band information including a green light band, a red light band, a near infrared band, a red edge band, an infrared band, and a short-wave infrared band;

[0037] A separation module, which is configured to separate the background image in the multi-spectral leaf image to obtain a target plant leaf image;

[0038] The recognition module is configured to determine the diseased area image in the target plant leaf image according to the multiple band information of the target plant leaf image. In a fifth aspect, the embodiment of the present application provides a plant leaf disease detection device, which adopts the following technical solution:

[0039] A plant leaf disease detection device, the plant leaf disease detection device comprising:

[0040] An acquisition module, which is configured to acquire a diseased area image of a target plant obtained according to the plant leaf diseased area detection method as described above and laser point cloud data of the target plant;

[0041] The detection module is configured to determine the leaf disease level of the target plant based on the pixel ratio between the diseased area image and the canopy leaves of the target plant.

[0042] In a sixth aspect, the present application provides a plant leaf disease detection device, which adopts the following technical solution:

[0043] A plant leaf disease detection device, the plant leaf disease detection device comprising:

[0044] An acquisition module, which is configured to acquire a diseased area image of a target plant obtained according to the above-mentioned plant leaf diseased area detection method and laser point cloud data of the target plant;

[0045] A fusion detection module is configured to obtain color features of target plant leaves and diseased area shooting data based on the diseased area image; obtain texture features of target plant leaves and diseased area scanning data based on the laser point cloud data; obtain shape features of the diseased area based on the diseased area shooting data and diseased area scanning data; and determine the type and grade of the disease of the target plant leaves based on the color features, texture features and diseased area shape features.

[0046] The beneficial effects brought by the technical solution provided in the embodiments of the present application include:

[0047] By acquiring multispectral leaf images of the canopy leaves of the target plants, we can obtain multiple band information of the canopy leaves, including green light band, red light band, near-infrared band, red edge band, infrared band and short-wave infrared band. These band information cannot be captured by RGB images taken by ordinary cameras. The wavelengths involved include visible light and near-infrared light. Since different substances have different absorption and reflection characteristics for light of different wavelengths, the obtained multispectral leaf images can reveal subtle differences in the leaf surface, which can be used to improve the accuracy of identifying leaf disease areas and facilitate subsequent detection of plant leaf disease types and severity. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 A schematic diagram of a flow chart of an embodiment of a method for detecting diseased areas of plant leaves provided in the present application;

[0049] Figure 2 A schematic diagram of a process flow of an embodiment of a plant leaf disease detection method provided in the present application;

[0050] Figure 3 A schematic diagram of a flow chart of another embodiment of the plant leaf disease detection method provided in the present application;

[0051] Figure 4 This is a schematic diagram of the functional modules of an embodiment of a device for detecting diseased areas on plant leaves of the present application. DETAILED DESCRIPTION

[0052] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0053] First, some technical terms in the present application are explained to facilitate those skilled in the art to understand the present application.

[0054] At present, in the field of plant disease identification technology, a camera is often used to capture plant images, and the plant images are analyzed in one step to identify and distinguish the diseased areas of the plants. However, when using a camera to capture plant images at this stage, the ordinary camera used can provide a high spatial resolution, but it is limited in spectral resolution and can only capture information of the three basic color channels of red, green and blue (RGB). It is powerless for other important spectral ranges in biological or scientific research (such as ultraviolet or infrared), and the dynamic range of the camera sensor is limited, and it may not be able to capture the details of the brightest and darkest parts of the scene. Therefore, the current technical solution of capturing plant images with a camera and then identifying them, if the disease characteristics are in a color channel that cannot be captured or is difficult to capture, it will make it difficult to effectively identify the areas where these diseases are located, which will affect the judgment of the disease type and severity based on the shape and location of the diseased area in the subsequent stage.

[0055] Based on the above problems, the present application provides a method for detecting diseased areas of plant leaves, a disease detection method and a device. The key point of the invention is that by acquiring multi-spectral leaf images of the canopy leaves of the target plant, multiple band information of the canopy leaves including green light band, red light band, near-infrared band, red edge band, infrared band and short-wave infrared band is acquired. These band information is information that cannot be captured by RGB images taken by general cameras. The wavelengths involved include visible light and near-infrared light. Since different substances have different absorption and reflection characteristics for light of different wavelengths, the obtained multi-spectral leaf image can reveal subtle differences on the leaf surface, thereby being able to be used to improve the accuracy of identifying leaf disease areas and facilitate subsequent detection of plant leaf disease types and severity.

[0056] In order to make the objectives, technical solutions and advantages of the present application clearer, the implementation methods of the present application will be further described in detail below with reference to the accompanying drawings.

[0057] First, refer to Figure 1 The present application provides a method for detecting a diseased area of ​​a plant leaf, the method comprising:

[0058] S100, obtaining a multispectral leaf image of a canopy leaf of a target plant; the multispectral leaf image at least includes information of multiple bands including a green light band, a red light band, a near infrared band, a red edge band, an infrared band, and a short-wave infrared band;

[0059] The obtained multispectral leaf image of the canopy leaf of the target plant is obtained by shooting with a multispectral camera in this embodiment. During the acquisition process, the multispectral image of the plant is acquired by using a drone equipped with a multispectral camera.

[0060] S200, separating the background image in the multispectral leaf image to obtain a target plant leaf image;

[0061] Among them, step S200 will be used to separate the background and canopy leaf image in the multispectral leaf image. In step S200, the pixel value of the canopy leaf surface in the image will be calculated by an edge detection algorithm, thereby achieving the separation of the background image and the target plant leaf image.

[0062] S300: Determine a diseased area image in the target plant leaf image according to information of multiple bands of the target plant leaf image.

[0063] Among them, for the target plant leaves screened out in S200, the discolored leaves will first be determined in S300. For the discolored leaves, there may not only be diseased leaves, but also abnormal leaf color caused by factors such as natural lack of nutrition, so for example, the leaves are withered and yellow. Therefore, it is necessary to remove the naturally discolored leaves from the target plant leaves, and then analyze the remaining leaf images, that is, the diseased area images. For the naturally discolored leaves, the yellowing of the leaves caused by lack of nutrition usually occurs from the tip of the leaf, while leaf blight caused by fungal diseases can invade from any part of the leaf surface.

[0064] Based on the above problems and principles, in some embodiments, in the process of determining the diseased area image in step S300, specifically the following steps are included:

[0065] S310, separating the target plant leaf image into a discolored leaf image and a normal leaf image according to multiple band information in the target plant leaf image;

[0066] S320, separating the leaf surface and the leaf tip of the discolored leaf image to obtain a diseased leaf surface image and a discolored leaf tip image;

[0067] S330, determining a diseased leaf tip image in the discolored leaf tip image according to an edge grayscale value of the discolored leaf tip image;

[0068] S340. Obtain a diseased area image in the target plant leaf image according to the diseased leaf surface image and the diseased leaf tip image.

[0069] Specifically, when separating the discolored leaf image from the normal leaf image in S310, the present embodiment will specifically use the K-means clustering (Gaussian kernel) algorithm to mark the discolored leaf surface area. The K-means clustering (Gaussian kernel) algorithm has a higher correct recognition rate than the K-means clustering algorithm. The specific process of the K-means clustering (Gaussian kernel) algorithm is as follows:

[0070] Let x i ∈R n , i = 1, 2, ..., l is the sample point of the original space, and a nonlinear mapping is used to remap the samples of the original space into a high-dimensional kernel space H to obtain Φ(x 1 ),Φ(x 2 ), ..., Φ(x l ), then the dot product of the original space is expressed in the high-dimensional kernel space using the Mercer kernel as

[0071] K(x i , x j )=Φ(x i )Φ(x j ) (1)

[0072] The distance between any sample Φ(x) and the class mean is

[0073]

[0074] where K is the number of clusters

[0075] N k —— the number of samples in the k-th class

[0076] x j 、x p —— the j-th and k-th samples of the k-th class C k

[0077] From equation (7), the K-means clustering objective function J can be expressed in the high-dimensional kernel space as

[0078]

[0079] The essence of the kernel K-means clustering algorithm is to seek the best clustering partition of the samples X in the high-dimensional kernel space based on the sum of squared errors criterion, minimizing the clustering objective function J. According to this clustering criterion and equation (2), the steps of the kernel K-means clustering algorithm are as follows:

[0080] 1. Given the number of clustering classes K (2 < K < l) and the allowable error E max .

[0081] 2. In the kernel space, determine the initial clustering centers according to the algorithm described later

[0082] 3. Use equation (2) to calculate the distance between any sample Φ(x i ) and the clustering centers, and assign the samples to the corresponding classes according to the nearest neighbor principle.

[0083] 4. Recalculate the clustering centers and the J value: Recalculate the clustering centers using equation (3) and calculate J according to equation (4) I+1 value.

[0084] 5. Calculate the error e = |J I - J I+1 |.

[0085] 6. If e ≤ E max , then end; otherwise, go to step 3.

[0086] Determine the initial clustering centers:

[0087] The traditional K-means algorithm is sensitive to the initial clustering centers, and the clustering results fluctuate greatly with different initial inputs. In view of this, initialize the clustering centers based on density estimation ​Define the sample point Φ(x i ) is the density function

[0088]

[0089] in

[0090]

[0091] Where r d A field density radius is 1 / 2 of the root mean square distance of N samples.

[0092] From formula (10), we can know that the sample point Φ(x i ) The denser the field, The larger the value, the greater the The corresponding Φ(x 1 )* is taken as the first initial cluster center. Find the density function adjustment relationship of the subsequent initial cluster centers, that is,

[0093]

[0094] make The corresponding sample point Φ(x k )* is taken as the kth cluster center.

[0095] After the color-changing leaf image is screened, it is necessary to further use the leaf tip judgment algorithm in step S320 to separate the color-changing leaf tip and the color-changing leaf surface in the color-changing leaf image, and to judge whether the color-changing leaf tip is a diseased leaf tip in step S330, so as to screen out the naturally color-changing leaf tips in the color-changing leaf tips.

[0096] Step S330 determines the diseased leaf tip image in the discolored leaf tip image according to the edge grayscale value of the discolored leaf tip image, comprising the following steps:

[0097] S331, sharpening the red and green color channels in the color-changing leaf tip image;

[0098] S332, determining whether there is any pixel point along the edge of the leaf tip region whose grayscale value difference with the adjacent leaf surface pixel point is greater than a set grayscale value range;

[0099] S333: If so, determine that the image corresponding to the leaf tip area is a diseased leaf tip image.

[0100] Specifically, during the process of pathogens invading plants, the pathogens form attachment cells, which produce melanin deposition on the cell wall and eventually form a melanin layer. Through the melanin layer, the pathogens are able to penetrate the surface of the host leaf and infect and parasitize. Melanin is an indispensable substance for rice blast fungi to penetrate the epidermis and invade rice plants. Therefore, in a large area of ​​infected areas, the color difference at the edge of the healthy part of the leaf surface and the infected part can be shown. However, for the yellowing of the leaf surface caused by natural lack of nutrition, its boundary does not have the above characteristics. Therefore, in the separation of natural yellowing leaf tips and yellowing leaf tips caused by pathogen infection, the red and green color channels are first sharpened to make the boundary contours of the diseased and healthy areas clear. In this image, if there is a point in the edge pixel points of the yellowing leaf tip area, the difference between the grayscale value of the adjacent leaf surface pixel point and this point is greater than 2 / 5 of the effective range of grayscale data (0-255), that is, approximately equal to 100, then this yellowing leaf tip area is determined as the diseased area; otherwise, it is determined as a natural yellowing leaf tip.

[0101] Finally, after obtaining the diseased leaf tip image and the diseased leaf surface image, the two are added together to obtain the diseased area image of the target plant.

[0102] Furthermore, in some embodiments, before the step S200 of separating the background image in the multispectral leaf image to obtain the target plant leaf image, the following steps are included:

[0103] S110, filtering and denoising the multispectral leaf image.

[0104] Specifically, in this embodiment, the multispectral leaf image will be filtered and denoised based on the improved cuckoo algorithm. Among them, the Wiener filter is one of the classic linear filters, which can effectively reduce the noise level of the image. In this embodiment, we will use this filter for processing and also optimize it to improve efficiency.

[0105] First, the original multispectral leaf image needs to be preprocessed, including image de-meaning, data standardization, etc. Then, the improved cuckoo algorithm will be used to find the optimal filter parameters. The improved cuckoo algorithm is a heuristic optimization algorithm that can solve the optimal solution by simulating the behavior of cuckoos when looking for food. The advantage of the algorithm is that it can quickly converge to the global optimal solution and also has good anti-noise performance. In this embodiment, the improved cuckoo algorithm will be used to search for the optimal filter parameters.

[0106] In the second aspect, after obtaining the diseased area image of the plant leaf according to the above method, the present application further proposes a plant leaf disease detection method, which detects the disease type or disease level of the plant leaf based on the above diseased area.

[0107] Reference Figure 2In some embodiments, the plant leaf disease detection method comprises the following steps:

[0108] A100, obtaining a diseased area image of a target plant according to the diseased area detection method of plant leaves as described above;

[0109] A200, determining the leaf disease grade of the target plant according to the pixel ratio between the diseased area image and the canopy leaves of the target plant. Specifically, determining the leaf blight evaluation grade corresponding to the ratio according to the leaf area of ​​the diseased part, that is, the ratio of the pixel points of the diseased area image to the total leaf area pixel points of the canopy.

[0110] In other embodiments, referring to Figure 3 , the plant leaf disease detection method comprises the following steps:

[0111] B100, obtaining the diseased area image of the target plant and the laser point cloud data of the target plant according to claim 1;

[0112] Among them, for laser point cloud data, in this embodiment, a drone equipped with a laser scanner is used to collect point cloud data of the target plant.

[0113] B200, obtaining color characteristics of target plant leaves and diseased area shooting data according to the diseased area image;

[0114] Specifically, the color features come from the diseased area image belonging to the multispectral image: the color moment is used to convert the R red light band, G green light band, and B blue light band into the Lab color model. The Lab color mode can make the segmented disease spot features more obvious. Through the Lab color mode image, the histograms of the L, a, and b channels are calculated respectively. By analyzing the histogram in the Lab color space, the key color features can be extracted. Among them, L is brightness; a's positive number represents red, and its negative end represents green; b's positive number represents yellow, and its negative end represents blue. The mean, variance, skewness, peak value, energy, and entropy of L, a, and b are calculated to obtain a total of 6x3=18 features, which together constitute the color vector C.

[0115] The diseased area shooting data is the position and shape of the diseased area displayed under the multispectral image.

[0116] B300, obtaining texture features of target plant leaves and diseased area scanning data according to the laser point cloud data;

[0117] Among them, the laser point cloud data will be processed by the LBP algorithm to obtain a texture weighted feature vector V with a total of 64 features, that is, the texture features of the target plant leaves.

[0118] The scanned data of the diseased area is the position and shape of the diseased area displayed under the laser point cloud data.

[0119] B400, obtaining shape features of the diseased area according to the diseased area shooting data and the diseased area scanning data;

[0120] Specifically, for the shape and position of the diseased area involved in both the diseased area shooting data and the diseased area scanning data, the average of the two will be taken in step B400 as the shape feature of the diseased area involved in the subsequent judgment. The shape feature of the diseased area specifically includes the aspect ratio, rectangularity, sphericity, circularity, eccentricity, perimeter convexity ratio, area convexity ratio and shape parameters of the leaf lesion image, and 7 invariant moment parameters, a total of 15 features that can describe the geometric features of the lesion image, and these features together constitute the shape feature vector H of the diseased area.

[0121] B500. Determine the type and grade of the disease on the target plant leaves according to the color characteristics, texture characteristics and shape characteristics of the diseased area.

[0122] Specifically, the above-mentioned feature vectors will be combined to form a set vector of the sample to be tested, namely:

[0123] x = [V, S, H, C]

[0124] Finally, the support vector machine (SVM) method is used to identify and classify the set vector of the sample to be tested, and the disease type and disease level corresponding to the sample to be tested can be obtained. Among them, the support vector machine method is a prior art in the field, and the specific identification and detection process is not disclosed in detail in the embodiment of this application. The key point of this embodiment is to use the color features with better effects in multispectral images, the texture features with better effects in laser point cloud data, and the shape features of the diseased area obtained by combining the two to perform the final disease fusion detection.

[0125] In a third aspect, an embodiment of the present application also provides a device for detecting diseased areas of plant leaves.

[0126] refer to Figure 4 , the detection device of the diseased area of ​​plant leaves includes:

[0127] An acquisition module is configured to acquire a multispectral leaf image of a canopy leaf of a target plant; the multispectral leaf image includes at least a plurality of band information including a green light band, a red light band, a near infrared band, a red edge band, an infrared band, and a short-wave infrared band;

[0128] A separation module, which is configured to separate the background image in the multi-spectral leaf image to obtain a target plant leaf image;

[0129] The recognition module is configured to determine the diseased area image in the target plant leaf image according to the multiple band information of the target plant leaf image.

[0130] Among them, the functional implementation of each module in the above-mentioned plant leaf diseased area detection device corresponds to the various steps in the above-mentioned plant leaf diseased area detection method embodiment, and its functions and implementation processes are no longer repeated here one by one.

[0131] In a fourth aspect, an embodiment of the present application also provides a plant leaf disease detection device.

[0132] In some embodiments, the plant leaf disease detection device comprises:

[0133] An acquisition module, which is configured to acquire a diseased area image of a target plant obtained according to the above-mentioned plant leaf diseased area detection method and laser point cloud data of the target plant;

[0134] The detection module is configured to determine the leaf disease level of the target plant based on the pixel ratio between the diseased area image and the canopy leaves of the target plant.

[0135] In some other embodiments, the plant leaf disease detection device includes:

[0136] An acquisition module, which is configured to acquire a diseased area image of a target plant obtained according to the above-mentioned plant leaf diseased area detection method and laser point cloud data of the target plant;

[0137] A fusion detection module is configured to obtain color features of target plant leaves and diseased area shooting data based on the diseased area image; obtain texture features of target plant leaves and diseased area scanning data based on the laser point cloud data; obtain shape features of the diseased area based on the diseased area shooting data and diseased area scanning data; and determine the type and grade of the disease of the target plant leaves based on the color features, texture features and diseased area shape features.

[0138] Among them, the functional implementation of each module in the above-mentioned plant leaf disease detection device corresponds to the various steps in the above-mentioned plant leaf disease detection method embodiment, and its functions and implementation processes are no longer repeated here.

[0139] It should be noted that the serial numbers of the above-mentioned embodiments of the present application are only for description and do not represent the advantages or disadvantages of the embodiments.

[0140] The terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally includes other steps or units inherent to these processes, methods, products or devices. The terms "first", "second" and "third" are used to distinguish different objects, etc., and do not represent a sequence, nor do they limit "first", "second" and "third" to different types.

[0141] In the description of the embodiments of the present application, "exemplary", "for example" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary", "for example" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary", "for example" or "for example" is intended to present related concepts in a specific way.

[0142] In the description of the embodiments of the present application, unless otherwise specified, “ / ” means or, for example, A / B can mean A or B; the “and / or” in the text is merely a description of the association relationship of associated objects, indicating that three relationships may exist, for example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, in the description of the embodiments of the present application, “multiple” refers to two or more than two.

[0143] In some processes described in the embodiments of the present application, multiple operations or steps that appear in a specific order are included, but it should be understood that these operations or steps may not be executed in the order in which they appear in the embodiments of the present application or in parallel, and the sequence number of the operation is only used to distinguish the different operations, and the sequence number itself does not represent any execution order. In addition, these processes may include more or fewer operations, and these operations or steps may be executed in sequence or in parallel, and these operations or steps may be combined.

[0144] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, disk, CD) as described above, and includes a number of instructions for a terminal device to execute the methods described in each embodiment of the present application.

[0145] The above are only preferred embodiments of the present application, and are not intended to limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method for detecting diseased areas of plant leaves, characterized in that: The method for detecting diseased areas of plant leaves comprises: Acquire a multispectral leaf image of the canopy leaves of the target plant; the multispectral leaf image includes at least a plurality of band information including a green light band, a red light band, a near infrared band, a red edge band, an infrared band and a short-wave infrared band; Separating the background image from the multispectral leaf image to obtain a target plant leaf image; According to the multiple band information of the target plant leaf image, a diseased area image in the target plant leaf image is determined.

2. The method for detecting diseased areas of plant leaves according to claim 1, characterized in that: Determining the diseased area image in the target plant leaf image according to the multiple band information of the target plant leaf image comprises the following steps: Separating the target plant leaf image into a discolored leaf image and a normal leaf image according to multiple band information in the target plant leaf image; Separating the leaf surface and the leaf tip of the discolored leaf image to obtain a diseased leaf surface image and a discolored leaf tip image; Determining a diseased leaf tip image in the discolored leaf tip image according to an edge grayscale value of the discolored leaf tip image; The diseased area image in the target plant leaf image is obtained according to the diseased leaf surface image and the diseased leaf tip image.

3. The method for detecting diseased areas of plant leaves according to claim 2, characterized in that: Determining the diseased leaf tip image in the discolored leaf tip image according to the edge grayscale value of the discolored leaf tip image comprises the following steps: sharpening the red and green color channels in the color-changing leaf tip image; Determine whether there is any pixel point along the edge of the leaf tip region whose grayscale value difference with the adjacent leaf surface pixel point is greater than the set grayscale value range; If it exists, it is determined that the image corresponding to the leaf tip area belongs to the diseased leaf tip image.

4. The method for detecting diseased areas of plant leaves according to claim 1, characterized in that: Before separating the background image in the multispectral leaf image to obtain the target plant leaf image, the following steps are included: The multispectral leaf image is filtered and denoised.

5. A method for detecting plant leaf diseases, characterized in that: The plant leaf disease detection method comprises the following steps: Acquire a diseased area image of a target plant obtained by the diseased area detection method for plant leaves according to claim 1; The leaf disease grade of the target plant is determined according to the pixel ratio between the diseased area image and the canopy leaves of the target plant.

6. A method for detecting plant leaf diseases, characterized in that: The plant leaf disease detection method comprises the following steps: Acquire the diseased area image of the target plant obtained according to claim 1 and the laser point cloud data of the target plant; According to the diseased area image, the color characteristics of the target plant leaves and the diseased area shooting data are obtained; According to the laser point cloud data, the texture characteristics of the target plant leaves and the diseased area scanning data are obtained; According to the photographed data and scanned data of the diseased area, the shape characteristics of the diseased area are obtained; The type and grade of the disease on the target plant leaves are determined based on the color characteristics, texture characteristics and shape characteristics of the diseased area.

7. The plant leaf disease detection method according to claim 6, characterized in that: Determining the disease type and grade of the target plant leaf according to the color feature, texture feature and disease area shape feature comprises the following steps: Obtaining a feature recognition vector according to the color feature, texture feature and shape feature of the diseased area; The leaf disease type and disease grade of the target plant are determined according to the results of the feature recognition vector in a plurality of preset vector recognition models.

8. A device for detecting diseased areas of plant leaves, characterized in that: The device for detecting the diseased area of ​​plant leaves comprises: An acquisition module is configured to acquire a multispectral leaf image of a canopy leaf of a target plant; the multispectral leaf image includes at least a plurality of band information including a green light band, a red light band, a near infrared band, a red edge band, an infrared band, and a short-wave infrared band; A separation module, which is configured to separate the background image in the multi-spectral leaf image to obtain a target plant leaf image; The recognition module is configured to determine the diseased area image in the target plant leaf image according to the multiple band information of the target plant leaf image.

9. A plant leaf disease detection device, characterized in that: The plant leaf disease detection device comprises: An acquisition module, which is configured to acquire a diseased area image of a target plant obtained by the diseased area detection method of a plant leaf according to claim 1 and laser point cloud data of the target plant; The detection module is configured to determine the leaf disease level of the target plant based on the pixel ratio between the diseased area image and the canopy leaves of the target plant.

10. A plant leaf disease detection device, characterized in that: The plant leaf disease detection device comprises: An acquisition module, which is configured to acquire a diseased area image of a target plant obtained by the diseased area detection method of a plant leaf according to claim 1 and laser point cloud data of the target plant; A fusion detection module is configured to obtain color features of target plant leaves and diseased area shooting data based on the diseased area image; obtain texture features of target plant leaves and diseased area scanning data based on the laser point cloud data; obtain shape features of the diseased area based on the diseased area shooting data and diseased area scanning data; and determine the type and grade of the disease of the target plant leaves based on the color features, texture features and diseased area shape features.

Citation Information

Cited By

  • Forest pest control unmanned aerial vehicle spraying area spraying method and system

    CN119672543A

  • Plant leaf scab measurement system and method based on infrared light anchoring and image segmentation

    CN120125608A

  • Garden plant health intelligent monitoring system based on image recognition

    CN121392579A

  • Intelligent monitoring system for health of garden plants based on image recognition

    CN121392579B