Potato leaf disease area positioning method, system and device

Abstract

The invention relates to a potato leaf disease area positioning method, system and device.The method is applied to the potato leaf disease area positioning device.The method comprises the steps that a target infrared image of a potato leaf is segmented into a normal area and an abnormal area according to the environment temperature through a disease segmentation model obtained in advance; determining the water content of the target infrared image according to the leaf spectrum information through a pre-obtained water content prediction model; the leaf emissivity characteristic is calculated according to the water content through a pre-obtained temperature correction model, the temperature of the abnormal area is corrected based on the emissivity characteristic and the water content characteristic, and the temperature correction model is obtained based on potato leaf characteristic training; and through a pre-trained disease identification model, identifying a disease area from the abnormal area after temperature correction.

Description

Technical Field

[0001] The present application relates to the field of crop disease monitoring, and in particular to a method, system and device for locating potato leaf disease areas. Background Art

[0002] With the development of smart agriculture, infrared thermal imaging technology, due to its non-contact, non-destructive, and visual advantages, has been gradually introduced into the agricultural field for plant health monitoring. As potatoes are an important food and cash crop, leaf disease detection is crucial for increasing yields and reducing pesticide use.

[0003] Disease identification methods in existing industrial general-purpose thermal imaging equipment usually use universal imaging parameters, which ignore the characteristics of potato leaves and make it impossible to accurately identify the diseased areas on potato leaves.

[0004] Therefore, the accuracy of existing potato leaf disease area positioning methods is not high and there is room for optimization. Summary of the Invention

[0005] The embodiments of the present application provide a method, system, and device for locating a potato leaf disease area, to at least solve the problem of low accuracy of potato leaf disease area locating methods in related technologies.

[0006] In a first aspect, the present invention provides a method for locating a diseased area on potato leaves.

[0007] The method is applied to a device for locating a diseased area on potato leaves, and the method comprises:

[0008] Using the pre-acquired disease segmentation model, the target infrared image of potato leaves is segmented into normal areas and abnormal areas according to the ambient temperature;

[0009] Determining the moisture content of the target infrared image according to the leaf spectrum information using a pre-acquired moisture content prediction model;

[0010] Calculating leaf emissivity characteristics according to the moisture content using a pre-acquired temperature correction model, and correcting the temperature of the abnormal area based on the emissivity characteristics and the moisture content characteristics, wherein the temperature correction model is trained based on potato leaf characteristics;

[0011] The pre-trained disease recognition model is used to identify the diseased area from the temperature-corrected abnormal area.

[0012] In one embodiment, the temperature correction model is configured to be obtained by:

[0013] Acquiring training data, the training data including an infrared image of a potato leaf with a pre-marked diseased area, a temperature value of each pixel in the image, a moisture content of the leaf, a position of the pixel in the diseased area on the leaf, and a true temperature corresponding to each pixel in the corrected image;

[0014] Obtaining the water content feature of the leaf in each image and the position feature of each pixel in the lesion area, determining the emissivity feature based on the water content feature, and using the water content feature and the emissivity feature as global features;

[0015] Predicting a corrected predicted temperature of a corresponding pixel based on the moisture content characteristic, the emissivity characteristic, and the position characteristic of each pixel in the lesion area;

[0016] An error value between the predicted temperature and the actual temperature is calculated, and in response to the error value being less than a preset threshold, iteration is stopped to obtain the temperature correction model.

[0017] In one embodiment, correcting the temperature of the abnormal area includes:

[0018] Identifying the relative position of each pixel in the abnormal area in the leaf using a pre-trained position recognition model;

[0019] In response to the relative position being within a preset target position range, the temperature of the abnormal area is corrected according to the water content and the relative position through the temperature correction model.

[0020] In one embodiment, obtaining the disease identification area further includes:

[0021] In response to the relative position being outside a preset target position range, taking the corresponding pixel as a first abnormal area;

[0022] In response to the relative position being within the preset target position range, taking the corresponding pixel as the middle area, and identifying a second abnormal area from the temperature-corrected middle area using the disease recognition model;

[0023] The first abnormal area and the second abnormal area are regarded as diseased areas.

[0024] In one embodiment, acquiring an infrared image of a target includes:

[0025] Obtaining an infrared image captured by an infrared thermal imager, and improving the resolution of the infrared image using a pre-trained super-resolution model to obtain a super-resolution image;

[0026] The leaf region in the super-resolution image is identified by using a pre-trained leaf recognition model to obtain a target infrared image.

[0027] In one embodiment, the disease segmentation model is obtained in the following manner:

[0028] The disease segmentation model is obtained by training infrared images of potato leaves with pre-labeled diseased areas, the ambient temperature corresponding to the images, and the temperature of each pixel in the images.

[0029] In a second aspect, an embodiment of the present application provides a potato leaf disease area positioning system, which is deployed in a potato leaf disease area positioning device, and includes:

[0030] Segmentation module: Using the pre-acquired disease segmentation model, the target infrared image of potato leaves is segmented into normal areas and abnormal areas according to the ambient temperature;

[0031] Moisture content prediction module: determines the moisture content of the target infrared image according to the leaf spectrum information using a pre-acquired moisture content prediction model;

[0032] Temperature correction module: using a pre-acquired temperature correction model, calculating the leaf emissivity characteristics according to the moisture content, and correcting the temperature of the abnormal area based on the emissivity characteristics and the moisture content characteristics, wherein the temperature correction model is trained based on potato leaf characteristics;

[0033] Disease area positioning module: used to identify the diseased area from the temperature-corrected abnormal area through a pre-trained disease recognition model.

[0034] In a third aspect, an embodiment of the present application provides a device for locating a diseased area on potato leaves, the device comprising an infrared thermal image acquisition module, a spectrum detection module, and an image processing module.

[0035] The acquisition module is used to acquire infrared images of the blades. The resolution of the acquisition module is greater than 640×480, and the thermal sensitivity of the acquisition module is ≤50mk;

[0036] The spectrum detection module is used to obtain the spectrum information of the blade;

[0037] The image processing module is used to process the infrared image and the leaf spectral image to implement the potato leaf disease area positioning method described in the first aspect.

[0038] In one embodiment, the device further includes a data storage module and a power supply module.

[0039] The data storage module is used to store the thermal map data of potato leaves for users to view or synchronize to other devices;

[0040] The power supply module is used to supply power to the device, and the power supply time is at least 10 hours.

[0041] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for locating a potato leaf disease area as described in the first aspect.

[0042] The embodiments of the present application provide a method, system, and device for locating a diseased area on potato leaves, which have at least the following technical effects.

[0043] This application uses a disease segmentation model to initially segment potato leaves into normal and abnormal regions. During the segmentation process, the segmentation threshold is corrected using ambient temperature to mitigate the impact of ambient temperature on disease segmentation and improve the accuracy of initial disease segmentation. Furthermore, considering the moisture content of potato leaves, this feature is introduced to correct the pixel temperature of abnormal regions obtained after the initial segmentation, thereby achieving accurate identification of potato leaf temperature and improving the accuracy of subsequent disease region location. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0045] Figure 1 This is a flow chart of a method for locating a potato leaf disease area according to an embodiment of the present application;

[0046] Figure 2 This is a structural block diagram of a potato leaf disease area positioning system according to an embodiment of the present application;

[0047] Figure 3 This is a structural block diagram of a potato leaf disease area positioning device provided in accordance with an embodiment of the present application;

[0048] Figure 4 is a schematic diagram showing a device structure according to an exemplary embodiment;

[0049] Figure 5 The figure is a schematic diagram of the size and structure of an infrared thermal imaging acquisition module according to an exemplary embodiment. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solutions and advantages of this application more clearly understood, the present application is described and illustrated below in conjunction with the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely used to explain this application and are not intended to limit this application. Based on the embodiments provided in this application, all other embodiments obtained by those of ordinary skill in the art without making any creative efforts are within the scope of protection of this application.

[0051] Obviously, the drawings described below are merely examples or embodiments of the present application. Those skilled in the art can, without inventive effort, apply the present application to other similar scenarios based on these drawings. Furthermore, it is also understood that, although the effort involved in such a development process may be complex and lengthy, for those skilled in the art related to the content disclosed in this application, changes in design, manufacturing, or production based on the technical content disclosed in this application are merely conventional technical means and should not be construed as an insufficiency of the content disclosed in this application.

[0052] References to "embodiments" in this application mean that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it refer to independent or alternative embodiments that are mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described in this application may be combined with other embodiments unless there is a conflict.

[0053] Unless otherwise defined, the technical or scientific terms used in this application should have the ordinary meaning understood by a person of ordinary skill in the technical field to which this application belongs. The words "one", "a", "the" and the like used in this application do not indicate a limit on quantity and may indicate the singular or plural. The terms "include", "comprise", "have" and any variations thereof used in this application are intended to cover non-exclusive inclusions; for example, a process, method, system, product or device that includes a series of steps or modules (units) is not limited to the listed steps or units, but may also include steps or units that are not listed, or may also include other steps or units that are inherent to these processes, methods, products or devices. The words "connect", "connected", "coupled" and the like used in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The word "multiple" used in this application refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships can exist. For example, "A and / or B" can mean: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the objects before and after are in an "or" relationship. The terms "first", "second", "third", etc. involved in this application are only used to distinguish similar objects and do not represent a specific order for the objects.

[0054] First, embodiments of the present application provide a method for locating diseased areas on potato leaves, which is used in conjunction with an infrared thermal imager. Optionally, the infrared thermal imager has a thermal sensitivity of ≤50mK and a resolution >640×480. The infrared thermal imager can use an uncooled microbolometer, and the thermal sensitive layer material includes vanadium oxide (VOx), amorphous silicon (a-Si), and polycrystalline silicon (Poly-Si).

[0055] Figure 1 FIG. 1 is a flow chart of a method for locating a potato leaf disease area according to an embodiment of the present application. Figure 1 As shown, the method includes:

[0056] Step S101 : using a pre-acquired disease segmentation model, the target infrared image of potato leaves is segmented into a normal area and an abnormal area according to the ambient temperature.

[0057] Optionally, the pre-acquired disease segmentation model includes pre-trained parameters, including a segmentation threshold, which is the ratio of the segmentation temperature to the ambient temperature. The disease segmentation model is used to segment potato leaves into normal and abnormal regions, and the segmentation threshold is corrected by the ambient temperature during the segmentation process. This mitigates the impact of ambient temperature on disease segmentation and improves the accuracy of initial disease segmentation. Abnormal regions are areas on potato leaves where disease has occurred or is at risk of developing disease.

[0058] In one example, acquiring an infrared image of a target includes:

[0059] Step S1011: Acquire an infrared image captured by an infrared thermal imager, and improve the resolution of the infrared image using a pre-trained super-resolution model to obtain a super-resolution image.

[0060] Step S1012: Using a pre-trained leaf recognition model, the leaf region in the super-resolution image is identified to obtain a target infrared image.

[0061] Optionally, increasing the resolution of infrared images using a super-resolution model can improve the accuracy of the subsequent leaf recognition model. This distinguishes potato leaves from plant organs such as branches and flowers, as well as the image background, eliminating interference from other organs and background. This facilitates the subsequent identification and location of diseased areas within the leaves, improving the accuracy of diseased area location. The leaf segmentation model is a machine learning or deep learning model.

[0062] In one example, the disease segmentation model in step S101 is obtained by training an infrared image of potato leaves with pre-marked diseased areas, the ambient temperature corresponding to the image, and the temperature of each pixel in the image to obtain the disease segmentation model.

[0063] Optionally, the training process of the disease segmentation model includes:

[0064] (1) Obtain training data for the disease segmentation model. The training data includes infrared images of potato leaves with pre-marked diseased areas, the ambient temperature corresponding to the infrared images, and the temperature of each pixel in the image.

[0065] (2) After obtaining the training data, the training data is preprocessed. The preprocessing includes normalization, edge enhancement, and pseudo-color processing of the infrared image. The preprocessing operation improves the recognizability of the neural network.

[0066] (3) The deep learning model is trained using the preprocessed training data. When the number of training folds is greater than or equal to the preset cross-validation folds, or the accuracy of the deep learning model is greater than or equal to the preset accuracy threshold, the iteration is stopped to obtain a disease segmentation model.

[0067] In this way, the disease segmentation model was trained, and the ambient temperature was added as a training feature during the training process to correct the segmentation threshold based on the ambient temperature. This prevented the influence of ambient temperature on disease segmentation and improved the accuracy of the initial segmentation.

[0068] Step S102 : determining the moisture content of the target infrared image according to the leaf spectrum information using a pre-acquired moisture content prediction model.

[0069] Optionally, the pre-acquired leaf spectral data and the corresponding leaf moisture content are used as training data for training to obtain a moisture content prediction model. The leaf spectral information corresponding to the target infrared image is input into the moisture content prediction model for prediction to obtain the moisture content of the target infrared image.

[0070] Step S103 , using a pre-acquired temperature correction model, the leaf emissivity characteristics are calculated according to the moisture content, and the temperature of the abnormal area is corrected based on the emissivity characteristics and the moisture content characteristics, wherein the temperature correction model is trained based on potato leaf characteristics.

[0071] Alternatively, based on the physical properties of potato leaves and the principles of heat and mass transfer, considering that leaf temperature anomalies may be due to interference from the high moisture content of potato leaves, and that leaf moisture content affects the leaf emissivity parameter, the temperature correction model uses the leaf moisture characteristics to derive the emissivity characteristic. The emissivity characteristic and moisture content characteristics are then used to calibrate the temperature of each pixel in the abnormal region. The emissivity parameter measures the ability of an object's surface to emit thermal radiation, and different materials have different emissivities.

[0072] In this way, combined with the physical properties of potato leaves, the temperature in the abnormal area is corrected to avoid abnormal temperature interference caused by leaf moisture content, thereby improving the accuracy of locating the diseased area of ​​potato leaves.

[0073] In one example, the temperature correction model in step S103 is configured to be obtained in the following manner:

[0074] Step S1031, obtaining training data, the training data including an infrared image of a potato leaf with a pre-marked diseased area, the temperature value of each pixel in the image, the moisture content of the leaf, the position of the pixel in the diseased area on the leaf, and the actual temperature corresponding to each pixel in the corrected image.

[0075] Optionally, considering that potato diseases often begin in their early stages at key locations such as leaf tips, leaf margins, and leaf veins, the positional features of pixels in the diseased area are introduced for training when training the temperature correction model. The pixel temperature is corrected using leaf moisture characteristics and pixel positional features. The positions of pixels in the diseased area on the leaf include: pixels at key locations such as the leaf tip, leaf veins, and leaf margins. Furthermore, before training, the images in the training data are normalized and resized to a uniform resolution, and the diseased area annotations are converted into binary masks.

[0076] Step S1032: Obtain the moisture content feature of the leaf in each image and the position feature of each pixel in the diseased spot area, determine the emissivity feature based on the moisture content feature, and use the moisture content feature and the emissivity feature as global features.

[0077] Step S1033 , predicting the corrected predicted temperature of the corresponding pixel based on the moisture content feature, the emissivity feature, and the position feature of each pixel in the lesion area.

[0078] Step S1034 , calculating the error between the predicted temperature and the actual temperature, and in response to the error being less than a preset threshold, stopping the iteration to obtain a temperature correction model.

[0079] Optionally, the temperature correction model can be any deep learning model that, during training, extends leaf moisture content and leaf emissivity as global features to all pixels on the leaf. Taking the U-Net model as an example, the input is an infrared image of a potato leaf, the temperature value of each pixel in the image, the leaf moisture content, and the position of the pixel in the lesion area on the leaf. Attribute features such as position features and moisture content features are attached to each pixel. At the end of the encoder, all feature vectors are broadcast to the same size as the image feature map, concatenated, and input into the decoder. The image features and attribute features are dynamically weighted using an attention mechanism to calculate and output the predicted temperature. The temperature correction model is obtained by iteratively training the model using the error between the predicted temperature and the actual temperature as an evaluation metric.

[0080] In one example, step S103 includes:

[0081] Step S1033: using a pre-trained position recognition model, identify the relative position of each pixel in the abnormal area in the leaf.

[0082] Step S1034 , in response to the relative position being within the preset target position range, the temperature of the abnormal area is corrected according to the water content and the relative position through the temperature correction model.

[0083] Optionally, considering that potato diseases often begin early in key locations such as the leaf tip, leaf margin, and leaf veins, a position recognition model is used to determine the relative position of pixels in the abnormal region within the leaf before temperature correction. Relative position refers to whether the pixel is located at a key location such as the leaf tip, leaf vein, leaf margin, or other non-critical locations; the target position range includes key locations such as the leaf tip, leaf vein, and leaf margin. If the pixel in the abnormal region is located at a key location on the leaf, temperature correction is performed on the abnormal region using the moisture content, relative position, and temperature correction model.

[0084] In this way, taking into account the moisture content characteristics of potato leaves and the location characteristics of potato leaf diseases, the moisture content characteristics and disease location characteristics are introduced to correct the pixel temperature in the abnormal area, thereby achieving accurate identification of potato leaf temperature, which is conducive to improving the accuracy of subsequent disease area positioning.

[0085] Step S104: using a pre-trained disease recognition model, identify the diseased area from the temperature-corrected abnormal area.

[0086] Optionally, a disease recognition model is developed by training a pre-labeled infrared image of a potato leaf with diseased areas, the corresponding ambient temperature, and the temperature of each pixel in the image. Ambient temperature is added as a training feature during training to calibrate the threshold for identifying abnormal areas. This reduces the influence of ambient temperature on abnormal area identification and improves recognition accuracy. In this way, by identifying abnormal areas after temperature correction, interference from leaf moisture content is avoided, improving the accuracy of abnormal area identification.

[0087] In one example, obtaining the disease identification area further includes:

[0088] Step S1041 : In response to the relative position being outside the preset target position range, the corresponding pixel is regarded as a first abnormal area.

[0089] Step S1042 , in response to the relative position being within the preset target position range, the corresponding pixel is used as the middle area, and a second abnormal area is identified from the temperature-corrected middle area through a disease recognition model.

[0090] Step S1043: The first abnormal area and the second abnormal area are regarded as diseased areas.

[0091] Alternatively, the relative position indicates that the pixel is located at a key position on the leaf, such as the tip, vein, or edge, or at other non-key positions; the target position range is key positions such as the tip, vein, and edge. The early stages of potato diseases usually occur at key positions such as the tip, vein, and edge. For example, the early stages of potato late blight often start at the tip or edge; the early stages of potato nutrient deficiency (magnesium or iron deficiency) usually start at the veins. Therefore, when the pixel in the diseased area is not at a key position such as the tip, vein, or edge, it indicates that the disease may not be in its early stages and has spread to other non-key positions. In this case, the pixel is directly added to the first abnormal area. When the pixel in the diseased area is at a key position such as the tip, vein, or edge, it indicates that the disease may be in its early stages. It is necessary to further determine whether the temperature anomaly in this area is caused by moisture content or by the early stages of the disease. In this case, the middle area is temperature-corrected using the moisture content, relative position, and temperature correction model, and then the temperature-corrected middle area is identified to obtain the second abnormal area. The combination of the first abnormal area and the second abnormal area is regarded as the diseased area.

[0092] In this way, the pixel locations of the diseased areas obtained after initial segmentation serve as a basis for further accurate identification of abnormal areas. Preliminary identification of the first abnormal area based on pixel locations reduces the number of pixels involved in subsequent temperature correction, thereby reducing the computational complexity of the temperature correction model. Furthermore, by incorporating moisture content and disease location characteristics to correct the temperature of the pixels in the abnormal area, accurate identification of potato leaf temperature is achieved, thereby improving the accuracy of diseased area location.

[0093] In summary, the present application uses a disease segmentation model to preliminarily segment potato leaves into normal areas and abnormal areas, and during the segmentation process, the segmentation threshold is corrected by the ambient temperature to avoid the influence of ambient temperature on disease segmentation and improve the accuracy of initial disease segmentation. In addition, taking into account the water content characteristics of potato leaves and the location characteristics of potato leaf disease, the water content characteristics and disease location characteristics are introduced to correct the pixel temperature of the abnormal area obtained after the preliminary segmentation, thereby achieving accurate identification of potato leaf temperature and improving the accuracy of subsequent disease area positioning. At the same time, the present application uses a high-sensitivity thermal imager, which can effectively capture the weak temperature difference signals generated by abnormal physiological metabolism in the early stages of potato leaf diseases, thereby achieving early detection and early warning of potato diseases.

[0094] In a second aspect, an embodiment of the present application provides a potato leaf disease area positioning system, which is deployed in a potato leaf disease area positioning device.

[0095] Optionally, the potato leaf disease area positioning system provided in this application can be deployed in different locations. (1) Deployed on local devices. For example, local deployment can be achieved through a neural network processor (NPU) and ARM architecture, thereby achieving completely offline real-time intelligent recognition and ensuring low latency and data privacy. (2) Deployed on the cloud server. Model calculations are performed through the computing power of the cloud server, and hardware limitations are broken through through data upload and result return mechanisms, thereby improving recognition accuracy in complex scenarios. (3) Deployed on local devices and servers at the same time for hierarchical processing. First, the data is initially screened locally, and only the key information is uploaded to the cloud for in-depth analysis. This combines the instant response of the local end with the computing scalability of the cloud, achieving a dynamic balance between resources and efficiency.

[0096] Figure 2 FIG. 1 is a structural block diagram of a potato leaf disease area positioning system according to an embodiment of the present application. Figure 2 As shown, the system includes:

[0097] Segmentation module 100: used to segment the target infrared image of potato leaves into normal areas and abnormal areas according to the ambient temperature using a pre-acquired disease segmentation model.

[0098] The moisture content prediction module 200 is used to determine the moisture content of the target infrared image according to the leaf spectrum information using a pre-acquired moisture content prediction model.

[0099] Temperature correction module 300: used to calculate the leaf emissivity characteristics according to the moisture content through a pre-acquired temperature correction model, and correct the temperature of the abnormal area based on the emissivity characteristics and the moisture content characteristics, wherein the temperature correction model is trained based on the leaf characteristics of potatoes.

[0100] The diseased area positioning module 400 is used to identify the diseased area from the abnormal area after temperature correction using a pre-trained disease recognition model.

[0101] In one example, the temperature correction model in the temperature correction module 300 is configured to be obtained in the following manner:

[0102] The training data includes infrared images of potato leaves with pre-marked diseased areas, the temperature value of each pixel in the image, the water content of the leaf, the position of the pixel in the diseased area on the leaf, and the actual temperature corresponding to each pixel in the corrected image.

[0103] The water content characteristics of the leaves in each image and the position characteristics of each pixel in the lesion area are obtained, the emissivity characteristics are determined based on the water content characteristics, and the water content characteristics and emissivity characteristics are used as global features.

[0104] The corrected predicted temperature of the corresponding pixel is predicted based on the moisture content characteristics, emissivity characteristics and position characteristics of each pixel in the lesion area.

[0105] The error value between the predicted temperature and the actual temperature is calculated, and in response to the error value being less than a preset threshold, the iteration is stopped to obtain a temperature correction model.

[0106] In one example, the temperature correction module 300 includes:

[0107] It is used to identify the relative position of each pixel in the leaf in the abnormal area through a pre-trained position recognition model.

[0108] In response to the relative position being within the preset target position range, the temperature of the abnormal area is corrected according to the water content and the relative position through the temperature correction model.

[0109] In one example, obtaining the disease identification area in the disease area positioning module 400 further includes:

[0110] Used to respond to the relative position being outside the preset target position range and to take the corresponding pixel as the first abnormal area.

[0111] In response to the relative position being within the preset target position range, the corresponding pixel is used as the middle area, and the second abnormal area is identified from the temperature-corrected middle area through the disease recognition model.

[0112] The first abnormal area and the second abnormal area are regarded as diseased areas.

[0113] In one example, acquiring the target infrared image in the segmentation module 100 includes:

[0114] It is used to obtain infrared images captured by an infrared thermal imager, improve the resolution of the infrared image through a pre-trained super-resolution model, and obtain a super-resolution image.

[0115] The target infrared image is obtained by identifying the leaf area in the super-resolution image using the pre-trained leaf recognition model.

[0116] In one example, the disease segmentation model in the segmentation module 100 is configured to be acquired by training an infrared image of potato leaves with pre-marked diseased areas, the ambient temperature corresponding to the image, and the temperature of each pixel in the image to obtain the disease segmentation model.

[0117] In summary, the present application uses a disease segmentation model to preliminarily segment potato leaves into normal areas and abnormal areas, and during the segmentation process, the segmentation threshold is corrected by the ambient temperature to avoid the influence of ambient temperature on disease segmentation and improve the accuracy of initial disease segmentation. In addition, taking into account the water content characteristics of potato leaves and the location characteristics of potato leaf disease, the water content characteristics and disease location characteristics are introduced to correct the pixel temperature of the abnormal area obtained after the preliminary segmentation, thereby achieving accurate identification of potato leaf temperature and improving the accuracy of subsequent disease area positioning. At the same time, the present application uses a high-sensitivity thermal imager, which can effectively capture the weak temperature difference signals generated by abnormal physiological metabolism in the early stages of potato leaf diseases, thereby achieving early detection and early warning of potato diseases.

[0118] In a third aspect, the present invention provides a device for locating diseased areas on potato leaves. Figure 3 This is a structural block diagram of a potato leaf disease area positioning device provided according to an embodiment of the present application, such as Figure 3 As shown, the device includes an infrared thermal image acquisition module 10 , a spectrum detection module 20 and an image processing module 30 .

[0119] The acquisition module 10 is used to acquire infrared images of the blades. The resolution of the acquisition module 10 is greater than 640×480, and the thermal sensitivity of the acquisition module 10 is ≤50mk.

[0120] Optionally, a highly sensitive infrared thermal imaging acquisition module 10 can be used to detect temperature differences, effectively detecting early-stage diseased areas. The high-resolution acquisition module 10 improves image resolution, allowing for clear visualization of leaf temperature differences and improving the accuracy of locating diseased areas. Figure 4 is a schematic diagram of a device structure according to an exemplary embodiment. Figure 4 As shown, the dimensions of the device are 286 mm × 265.501 mm × 225.248 mm. Figure 5 FIG. 1 is a schematic diagram showing the size and structure of an infrared thermal imaging acquisition module according to an exemplary embodiment. Figure 5 As shown, the total length of the infrared thermal image acquisition module 10 is 99.5 mm (3.92 inches), the diameter of the lens is 44 mm (1.73 inches), and there are two M3-sized mounting holes on the acquisition module.

[0121] The spectrum detection module 20 is used to obtain the spectrum information of the blade.

[0122] The image processing module 30 is used to process the infrared image and the leaf spectrum image to implement the potato leaf disease identification method of the first aspect.

[0123] Optionally, the image processing module 30 is equipped with an image processing algorithm that supports image thermal anomaly recognition and pseudo-color processing, and can display the detected image defect area in real time. Statistical analysis of data can also be performed based on the image defect area.

[0124] Continue to refer Figure 3 In one example, the device further includes a data storage module 40 and a power supply module 50 .

[0125] The data storage module 40 is used to store the thermal map data of potato leaves for users to view or synchronize to other devices.

[0126] Optionally, the data storage module 40 can save thermal map data in batches, supporting USB flash drive export or wireless synchronization. The data storage module 40 includes a wireless transmission unit for remote data upload. Local storage expansion solves the problem of data preservation in disconnected environments.

[0127] The power supply module 50 is used to power the device for at least 10 hours. Optionally, the power supply module 50 uses a high-performance lithium battery or solar cell to enable long-term operation of the device, reducing charging frequency and meeting the needs of long-term field inspections.

[0128] In one example, the device's packaging has a protection rating of at least IP65. This is to withstand the dusty and humid environment of farmland, preventing internal components from failing due to moisture or dust intrusion. This high level of protection ensures dust and water resistance, allowing the device to withstand harsh field conditions and enhance its durability.

[0129] In one example, the equipment can also be equipped with a portable transfer box to adapt to transportation and field operations.

[0130] In summary, the present application accurately identifies early diseases through micro-temperature difference sensing, and generates clear thermal images through high-resolution infrared sensors, providing an identification basis for the image processing module. The image processing module 30 combines the temperature change law of potato diseases to quickly and non-destructively detect the diseased area on the leaf surface. Through the lightweight body structure and high-level protection design, it takes into account the convenience of single-person handheld operation and adaptability to complex environments, and can operate stably in humid, dusty and other field scenes. The built-in large-capacity battery supports ultra-long battery life to meet long-term inspection needs. The storage module supports local batch storage and export functions to ensure that the detection data is traceable and easy to analyze, providing data support for potato health monitoring. The potato leaf disease area positioning device provided in this application has the characteristics of high resolution, high sensitivity, long battery life, strong protection, etc., and is suitable for field inspections in the wild.

[0131] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the potato leaf disease area locating method provided in the first aspect.

[0132] The readable storage medium may include, but is not limited to, a portable disk, a hard disk, a random access memory, a read-only memory, an erasable programmable read-only memory, an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0133] In a possible embodiment, the present invention can also be implemented in the form of a program product, which includes program code. When the program product is run on a terminal device, the program code is used to enable the terminal device to execute the steps of the potato leaf disease area locating method provided in the first aspect.

[0134] The program code for executing the present invention may be written in any combination of one or more programming languages, and may be executed entirely on the user device, partially on the user device, as an independent software package, partially on the user device and partially on a remote device, or entirely on the remote device.

[0135] The technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above-mentioned embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0136] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art could make various modifications and improvements without departing from the spirit of the present application, all of which fall within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be determined by the appended claims.

Claims

1. A method for locating potato leaf disease areas, characterized in that: The method is applied to a device for locating a diseased area on potato leaves, and the method comprises: Using the pre-acquired disease segmentation model, the target infrared image of potato leaves is segmented into normal areas and abnormal areas according to the ambient temperature; Determining the moisture content of the target infrared image according to the leaf spectrum information using a pre-acquired moisture content prediction model; Calculating leaf emissivity characteristics according to the moisture content using a pre-acquired temperature correction model, and correcting the temperature of the abnormal area based on the emissivity characteristics and the moisture content characteristics, wherein the temperature correction model is trained based on potato leaf characteristics; The pre-trained disease recognition model is used to identify the diseased area from the temperature-corrected abnormal area.

2. The method for locating potato leaf disease areas according to claim 1, characterized in that: The temperature correction model is configured to be obtained in the following manner: Acquiring training data, the training data including an infrared image of a potato leaf with a pre-marked diseased area, a temperature value of each pixel in the image, a moisture content of the leaf, a position of the pixel in the diseased area on the leaf, and a true temperature corresponding to each pixel in the corrected image; Obtaining the water content feature of the leaf in each image and the position feature of each pixel in the lesion area, determining the emissivity feature based on the water content feature, and using the water content feature and the emissivity feature as global features; Predicting a corrected predicted temperature of a corresponding pixel based on the moisture content characteristic, the emissivity characteristic, and the position characteristic of each pixel in the lesion area; An error value between the predicted temperature and the actual temperature is calculated, and in response to the error value being less than a preset threshold, iteration is stopped to obtain the temperature correction model.

3. The method for locating potato leaf disease areas according to claim 2, characterized in that: The correcting the temperature of the abnormal area includes: Identifying the relative position of each pixel in the abnormal area in the leaf using a pre-trained position recognition model; In response to the relative position being within a preset target position range, the temperature of the abnormal area is corrected according to the water content and the relative position through the temperature correction model.

4. The method for locating potato leaf disease areas according to claim 3, characterized in that: Obtaining disease identification areas also includes: In response to the relative position being outside a preset target position range, taking the corresponding pixel as a first abnormal area; In response to the relative position being within the preset target position range, taking the corresponding pixel as the middle area, and identifying a second abnormal area from the temperature-corrected middle area using the disease recognition model; The first abnormal area and the second abnormal area are regarded as diseased areas.

5. The method for locating potato leaf disease areas according to claim 1, characterized in that: Acquiring target infrared images includes: Obtaining an infrared image of a potato leaf captured by an infrared thermal imager, and improving the resolution of the infrared image using a pre-trained super-resolution model to obtain a super-resolution image; The leaf region in the super-resolution image is identified by using a pre-trained leaf recognition model to obtain a target infrared image of the potato leaf.

6. The method for locating potato leaf disease areas according to claim 1, characterized in that: The acquisition method of the disease segmentation model is configured as follows: The disease segmentation model is obtained by training infrared images of potato leaves with pre-labeled diseased areas, the ambient temperature corresponding to the images, and the temperature of each pixel in the images.

7. A potato leaf disease area positioning system, characterized in that: The system is deployed on a potato leaf disease area positioning device, and the system includes: Segmentation module: Using the pre-acquired disease segmentation model, the target infrared image of potato leaves is segmented into normal areas and abnormal areas according to the ambient temperature; Moisture content prediction module: determines the moisture content of the target infrared image according to the leaf spectrum information using a pre-acquired moisture content prediction model; Temperature correction module: using a pre-acquired temperature correction model, calculating the leaf emissivity characteristics according to the moisture content, and correcting the temperature of the abnormal area based on the emissivity characteristics and the moisture content characteristics, wherein the temperature correction model is trained based on potato leaf characteristics; Disease area positioning module: used to identify the diseased area from the temperature-corrected abnormal area through a pre-trained disease recognition model.

8. A device for locating diseased areas on potato leaves, characterized in that: The device includes an infrared thermal image acquisition module, a spectrum detection module and an image processing module. The acquisition module is used to acquire infrared images of the blades. The resolution of the acquisition module is greater than 640×480, and the thermal sensitivity of the acquisition module is ≤50mk; The spectrum detection module is used to obtain the spectrum information of the blade; The image processing module is used to process the infrared image and the leaf spectral image to implement the potato leaf disease area positioning method according to any one of claims 1 to 6.

9. The device for locating potato leaf disease areas according to claim 8, characterized in that: The device also includes a data storage module and a power supply module. The data storage module is used to store the thermal map data of potato leaves for users to view or synchronize to other devices; The power supply module is used to supply power to the device, and the power supply time is at least 10 hours.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method for locating a potato leaf disease area according to any one of claims 1 to 6 is implemented.

Images