Intelligent identification method and system for diseases in early growth stage of sugarcane

Abstract

The invention discloses an intelligent identification method and system for diseases in the early growth stage of sugarcane, and relates to the technical field of sugarcane planting, and the method comprises the steps: obtaining multispectral image data of leaves of sampled sugarcane in the early growth stage, and obtaining the spectral data of each preset wave band and a corresponding grayscale image from the multispectral image data; according to the quality of the spectral data of each preset wave band, endowing the spectral data of each preset wave band with a weight; performing weighting processing on the corresponding grayscale image by using the weight of the spectral data of each preset wave band to obtain a target grayscale image corresponding to the spectral data of each preset wave band; and under a preset constraint condition, fusing all the target grayscale images to obtain a fused image, and obtaining a disease identification result according to the fused image and the trained disease intelligent identification model. According to the invention, the accuracy of sugarcane disease identification can be improved.

Description

Technical Field

[0001] This invention relates to the field of sugarcane planting technology, and in particular to a method and system for intelligent identification of early-stage diseases in sugarcane growth. Background Technology

[0002] Currently, manual observation and judgment are often used to identify diseases in early-stage sugarcane. This method is easily affected by subjective factors, making it difficult to guarantee the accuracy and consistency of the judgment results, and it is also inefficient. In addition, when using image recognition algorithms to identify diseases in early-stage sugarcane, the recognition accuracy will decrease under complex environments such as different lighting and backgrounds. Summary of the Invention

[0003] The technical problem to be solved by this invention is to address the shortcomings of existing technologies, specifically by providing a method and system for intelligent identification of early-stage sugarcane diseases, as detailed below:

[0004] 1) In a first aspect, the present invention provides an intelligent identification method for early-stage sugarcane diseases, the specific technical solution of which is as follows:

[0005] Acquire multispectral image data of sugarcane leaves in the early stage of growth. The multispectral image data includes spectral data of multiple bands and grayscale images corresponding to the spectral data of each band.

[0006] From multispectral image data, obtain spectral data for each preset band, and obtain the grayscale image corresponding to the spectral data for each preset band;

[0007] Weights are assigned to the spectral data of each preset band based on the quality of the spectral data of each preset band.

[0008] The corresponding grayscale image is weighted by the weight of the spectral data of each preset band to obtain the target grayscale image corresponding to the spectral data of each preset band.

[0009] Under preset constraints, all target grayscale images are fused to obtain a fused image. The preset constraints are: the positions of the sampled sugarcane leaves in all target grayscale images coincide.

[0010] Based on the fused image and the trained intelligent disease identification model, the disease identification results are obtained.

[0011] The beneficial effects of the intelligent identification method for early-stage sugarcane diseases provided by this invention are as follows:

[0012] Sugarcane leaves contain rich information in their spectral data across different bands. By assigning weights to different bands, key disease-related information can be highlighted, improving identification accuracy. For example, certain bands may be more sensitive to the difference in reflectance between healthy and diseased leaves; by appropriately assigning weights, these differences can be better captured. The quality of spectral data varies across different bands; some bands may be less affected by noise, resulting in clearer and more stable spectral data that more accurately reflects the characteristics of sugarcane leaves. Assigning greater weights to these high-quality bands allows them to play a greater role in the fused image, thereby improving the accuracy of sugarcane disease identification. Conversely, reducing the weights of spectral data in bands with high noise and poor data quality can reduce their interference with the fused image, preventing errors introduced by low-quality data from affecting the final identification result. By appropriately assigning weights and then performing image fusion, the system can better adapt to complex natural environments, reduce interference from factors such as lighting and background, effectively enhance the robustness of the trained intelligent disease identification model to different environmental conditions, and improve its reliability in practical applications. Moreover, in the early stages of disease, the changes in the appearance of the leaves may not be obvious, but the spectral data may show abnormalities. This method can help detect diseases in the early stages of sugarcane growth and take timely control measures to reduce losses.

[0013] Based on the above scheme, the intelligent identification method for early sugarcane growth diseases of the present invention can be further improved as follows.

[0014] Furthermore, after obtaining the fused image, the process also includes:

[0015] Illumination correction is performed on the fused image to obtain the corrected fused image;

[0016] Based on the fused image and the trained intelligent disease identification model, the disease identification results are obtained, including:

[0017] Based on the corrected fused image and the trained intelligent disease identification model, the disease identification result is obtained.

[0018] The beneficial effects of adopting the above-mentioned further scheme are as follows: On the one hand, illumination correction can adjust the brightness and contrast of the fused image, making the details in the fused image more clearly visible, and making the fused image have a consistent visual effect under different lighting conditions, reducing the visual differences caused by changes in illumination. On the other hand, the fused image after correction can more realistically reflect the spectral characteristics of ground objects, which helps the trained intelligent disease identification model to extract features related to sugarcane diseases more accurately. Moreover, illumination correction can make the fused image after correction relatively stable under different lighting conditions, reducing the impact of changes in illumination on the trained intelligent disease identification model, thereby improving the generalization ability of the trained intelligent disease identification model in different scenarios.

[0019] Furthermore, after acquiring the spectral data for each preset band, the process also includes:

[0020] Radiometric and atmospheric corrections are performed on the spectral data of each preset band to obtain the corrected spectral data of each preset band.

[0021] Weights are assigned to the spectral data of each preset band based on the quality of the spectral data in the multispectral image data, including:

[0022] Weights are assigned to the corrected spectral data of each preset band based on the quality of the corrected spectral data of each preset band.

[0023] The corresponding grayscale image is weighted using the weights of the spectral data for each preset band to obtain the target grayscale image corresponding to the spectral data of each preset band, including:

[0024] The corresponding grayscale image is weighted by using the weights of the corrected spectral data of each preset band to obtain the target grayscale image corresponding to the corrected spectral data of each preset band.

[0025] The beneficial effects of adopting the above-mentioned further scheme are: radiation correction can eliminate errors caused by factors such as atmospheric attenuation and surface reflection characteristics, thus more accurately reflecting the radiation characteristics of sugarcane; atmospheric correction can eliminate the influence of atmospheric scattering and absorption, and more accurately reflect the spectral characteristics of sugarcane, providing a reliable data foundation for subsequent analysis and application.

[0026] Furthermore, it also includes:

[0027] When the disease identification results include diseased areas, the boundaries of the diseased areas are enhanced.

[0028] The beneficial effect of adopting the above-mentioned further scheme is that by enhancing the boundary of the lesion area, it is easier to view it intuitively.

[0029] 2) Secondly, the present invention also provides an intelligent identification system for early-stage sugarcane diseases, the specific technical solution of which is as follows:

[0030] It includes a multispectral image data acquisition module, a data filtering module, a weight allocation module, a weighted processing module, an image fusion module, and a disease identification module;

[0031] The multispectral image data acquisition module is used to acquire multispectral image data of sampled sugarcane leaves in the early stage of growth. The multispectral image data includes spectral data of multiple bands and grayscale images corresponding to the spectral data of each band.

[0032] The data filtering module is used to: obtain spectral data for each preset band from multispectral image data, and obtain the grayscale image corresponding to the spectral data of each preset band;

[0033] The weight allocation module is used to assign weights to the spectral data of each preset band based on the quality of the spectral data of each preset band.

[0034] The weighted processing module is used to: perform weighted processing on the corresponding grayscale image using the weights of the spectral data of each preset band, so as to obtain the target grayscale image corresponding to the spectral data of each preset band;

[0035] The image fusion module is used to fuse all target grayscale images under preset constraints to obtain a fused image. The preset constraints are: the positions of the sampled sugarcane leaves in all target grayscale images coincide.

[0036] The disease identification module is used to obtain disease identification results based on the fused image and the trained intelligent disease identification model.

[0037] Based on the above scheme, the intelligent identification system for early-stage sugarcane diseases of the present invention can be further improved as follows.

[0038] Furthermore, it also includes an image correction module, which is used for:

[0039] After obtaining the fused image, illumination correction is performed on the fused image to obtain the corrected fused image;

[0040] The disease identification module is specifically used to obtain disease identification results based on the corrected fused image and the trained intelligent disease identification model.

[0041] Furthermore, it also includes a spectral data correction module, which is used for:

[0042] After acquiring the spectral data for each preset band, radiometric and atmospheric corrections are performed on the spectral data for each preset band to obtain the corrected spectral data for each preset band.

[0043] The weight allocation module is specifically used to assign weights to the corrected spectral data of each preset band based on the quality of the corrected spectral data of each preset band.

[0044] The weighted processing module is specifically used to: use the weights of the corrected spectral data of each preset band to perform weighted processing on the corresponding grayscale image, so as to obtain the target grayscale image corresponding to the corrected spectral data of each preset band.

[0045] Furthermore, it also includes an enhancement processing module, which is used for:

[0046] When the disease identification results include diseased areas, the boundaries of the diseased areas are enhanced.

[0047] 3) In a third aspect, the present invention also provides an electronic device, the electronic device including a processor coupled to a memory, the memory storing at least one computer program, the at least one computer program being loaded and executed by the processor, so that the electronic device realizes any of the above-mentioned intelligent identification methods for early sugarcane growth diseases.

[0048] 4) In a fourth aspect, the present invention also provides a computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements any of the above-mentioned methods for intelligent identification of early-stage sugarcane diseases.

[0049] It should be noted that the beneficial effects of the technical solutions of the second to fourth aspects of the present invention and their corresponding possible implementations can be found in the above description of the technical effects of the first aspect and its corresponding possible implementations, and will not be repeated here. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments of the present invention will be briefly introduced below:

[0051] Figure 1 This is a flowchart illustrating an intelligent identification method for early-stage sugarcane diseases according to an embodiment of the present invention.

[0052] Figure 2 This is a schematic diagram of the structure of an intelligent identification system for early-stage sugarcane diseases according to an embodiment of the present invention;

[0053] Figure 3 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0054] The principles and features of the present invention are described below. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.

[0055] The technical solution of the present invention and how the technical solution of the present invention solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of the present invention will now be described with reference to the accompanying drawings.

[0056] like Figure 1 As shown in the figure, an intelligent identification method for early-stage sugarcane diseases according to an embodiment of the present invention includes the following steps:

[0057] S1. Acquire multispectral image data of sampled sugarcane leaves in the early stage of growth. The multispectral image data includes spectral data of multiple bands and grayscale images corresponding to the spectral data of each band.

[0058] The early growth stages of sugarcane include the germination stage, seedling stage, and tillering stage.

[0059] The specific process for determining the sugarcane to be sampled is as follows:

[0060] Using a drone equipped with a multispectral camera, each sub-region of the target sugarcane planting area is periodically scanned (the target sugarcane planting area is divided into multiple sub-regions in advance according to the actual situation). The morphology of the initial sampled sugarcane in each sub-region (the initial sampled sugarcane in any sub-region can be any sugarcane in that sub-region) is extracted. The morphology of the sugarcane includes leaf area index, chlorophyll content and plant height.

[0061] LoRa sensors are deployed to monitor parameters such as soil temperature, humidity, pH, and nitrogen, phosphorus, and potassium content in each sub-region in real time, and to mark abnormal areas, including drought areas and waterlogged areas.

[0062] Simultaneously collect macro-environmental data such as temperature, humidity, wind speed, and light intensity for each sub-region.

[0063] Federated learning is used to perform cross-modal alignment of images acquired by multispectral cameras, data acquired by LoRa sensors, and data acquired by weather stations to generate a spatiotemporal feature matrix for each initial sample of sugarcane. The elements in the spatiotemporal feature matrix include the leaf area index, chlorophyll content, plant height, soil temperature and humidity, pH value, nitrogen, phosphorus and potassium content, temperature, humidity, wind speed, and light intensity of the initial sample of sugarcane. The arrangement of the elements can be set according to the actual situation.

[0064] Cluster analysis was performed on all spatiotemporal feature matrices to obtain multiple groups. At least one initial sample sugarcane was selected from each group as the sample sugarcane, so that the selected sample sugarcane is more representative and can more comprehensively cover the sugarcane growth of the target sugarcane planting area.

[0065] Optionally, in the above technical solution, features are generated to characterize virtual disaster scenarios (such as persistent drought and insect outbreaks). These features and the target spatiotemporal feature matrix corresponding to any group are then input into the sugarcane survival probability prediction model to obtain the sugarcane survival probability of that group. This process is repeated until the sugarcane survival probability of each group is obtained. Based on the sugarcane survival probability of each group, the number of sugarcane samples to be drawn from each group is determined. Generally speaking, the higher the sugarcane survival probability of any group, the fewer the number of sugarcane samples in that group; conversely, the lower the sugarcane survival probability of any group, the more sugarcane samples in that group. This allows for the acquisition of more data on sugarcane samples that are susceptible to disasters, enabling more accurate irrigation and fertilization plans to be provided.

[0066] Where N = (1-P) × 100, where N is the number of sugarcane samples and P is the survival probability of sugarcane (the survival probability of sugarcane can be kept to two digits). Alternatively, a functional relationship between the number of sugarcane samples and the survival probability of sugarcane can be set according to the actual situation.

[0067] The present invention utilizes an intelligent identification method for early-stage sugarcane diseases to process multispectral image data of each sampled sugarcane to obtain disease identification results for each sampled sugarcane. The disease identification result of any sampled sugarcane can be used as the overall disease identification result of all sugarcanes associated with that sampled sugarcane, so as to facilitate subsequent planting management. Here, all sugarcanes associated with any sampled sugarcane refer to: all sugarcanes in the sub-region where the sampled sugarcane is located, as well as sugarcanes in each sub-region of the group to which the sampled sugarcane belongs.

[0068] The preset range corresponding to any sampled sugarcane can be a rectangular area or a circular area centered on the sampled sugarcane. The size of the rectangular area and the circular area can be set according to the actual situation.

[0069] The process involved using a high-resolution multispectral camera to photograph the leaves of sampled sugarcane, aiming to acquire high-quality multispectral image data. For example, selecting a high-resolution multispectral camera with a wavelength range of 400-1000 nanometers enabled the capture of the spectral reflectance characteristics of the sampled sugarcane leaves at different wavelengths.

[0070] S2. Obtain the spectral data of each preset band from the multispectral image data, and obtain the grayscale image corresponding to the spectral data of each preset band.

[0071] Prioritizing the different responses of sugarcane leaves to different wavelengths during growth is a good approach. Analysis revealed that the spectral reflectance of sugarcane leaves in the red light band is closely related to nitrogen content, while the spectral reflectance in the green light band is related to leaf health. Therefore, focusing on the red and green light bands improves the effectiveness and relevance of the data. Multiple preset wavelengths, including 450nm-500nm (blue light), 550nm-600nm (green light), and 650nm-700nm (red light), help focus on the spectrally sensitive regions of sugarcane leaves, reducing data processing workload.

[0072] S3. Assign weights to the spectral data of each preset band based on the quality of the spectral data of each preset band.

[0073] Specifically, the quality of spectral data for each preset band can be evaluated using signal-to-noise ratio and information entropy calculation methods.

[0074] Calculate the signal-to-noise ratio (SNR) and information entropy of the spectral data for each preset band. Perform Min-Max normalization on the SNR and information entropy of the spectral data for each preset band. Multiply the normalized SNR and normalized information entropy of the spectral data for any preset band to obtain the comprehensive index of the spectral data for that preset band. Repeat this process until the comprehensive index corresponding to the spectral data for each preset band is obtained. Normalize all comprehensive indices again using the following formula to obtain the weight of the spectral data for each preset band:

[0075]

[0076] Among them, W i Represents: the weight of the spectral data of the i-th preset band, C i C represents the comprehensive index of the spectral data for the i-th preset band. j Represents the weight of the spectral data in the j-th preset band, and n represents the number of spectral data in the preset band.

[0077] In the process of obtaining the comprehensive index by multiplication, the multiplication method can ensure that the subsequent weights are significantly increased. Through two normalizations, the dimensional differences of the original data (spectral data of each preset band) are eliminated, avoiding the dominance of the spectral data of a single preset band in weight allocation. In particular, the extreme value normalization based on the original data itself does not require preset parameters (such as signal-to-noise ratio threshold and information entropy threshold), which improves the applicability in different scenarios. Moreover, it only requires basic mathematical operations, has low complexity, and is suitable for real-time processing.

[0078] S4. The corresponding grayscale image is weighted using the weights of the spectral data for each preset band to obtain the target grayscale image corresponding to the spectral data of each preset band. Specifically:

[0079] The weights of the spectral data for any preset band are used to perform multi-scale decomposition on the corresponding grayscale image, generating multiple low-frequency components and multiple high-frequency components. The weights of the spectral data for the preset band are multiplied by each low-frequency component to obtain multiple weighted low-frequency components. The weights of the spectral data for the preset band are multiplied by the square root of each high-frequency component to obtain multiple weighted high-frequency components. All weighted low-frequency components and all weighted high-frequency components are reconstructed by inverse transformation to obtain the target grayscale image corresponding to the spectral data for the preset band, until the target grayscale image corresponding to the spectral data for each preset band is obtained.

[0080] The process of obtaining the weighted low-frequency components involves linear weighting of the low-frequency components, which preserves the overall structure of the grayscale image. The process of obtaining the weighted high-frequency components involves nonlinear weighting of the high-frequency components, which effectively suppresses low-weight noise while preserving high-weight details. This solves the trade-off problem between smooth and detailed regions in traditional weighting methods. Furthermore, by combining the multi-scale characteristics of the target grayscale image, it avoids information loss caused by single spatial domain weighting (such as edge blurring), thereby improving the accuracy of the obtained disease identification results.

[0081] Specifically, when wavelet transform is used to weight the spectral data of the preset band and decompose the corresponding grayscale image at multiple scales, the inverse transform reconstruction process is the inverse transform reconstruction of wavelet transform, i.e., wavelet inverse transform. When pyramid decomposition is used to weight the spectral data of the preset band and decompose the corresponding grayscale image at multiple scales, the inverse transform reconstruction process is the inverse transform reconstruction of pyramid decomposition, i.e., pyramid inverse transform.

[0082] The overall structure of the grayscale image is described as follows:

[0083] In image processing, low-frequency components typically correspond to smooth areas, large background areas, or slowly changing parts of an image. These parts form the basic grayscale levels and overall structure of the image. Low-frequency components capture the main structure and shape of the image, such as the background and the outline of large objects. They are the parts of the image that change more slowly and usually correspond to large background areas or smooth areas.

[0084] S5. Under the preset constraints, all target grayscale images are fused to obtain a fused image. The preset constraints are: the positions of the sampled sugarcane leaves in all target grayscale images coincide.

[0085] S6. Based on the fused image and the trained intelligent disease identification model, the disease identification result is obtained.

[0086] The disease identification results include the affected area, disease name, and disease severity, which can be set according to the actual situation.

[0087] The process of acquiring the trained intelligent disease identification model includes:

[0088] S60. Construct the target deep learning model, which includes: an encoder, a data processing intermediate layer, and a decoder. Specifically:

[0089] 1) The encoder consists of four layers arranged sequentially. Each layer includes a Dynamic Multi-Scale Feature Fusion (DMSF) module and an AFA module. Each DMSF module includes four parallel branches. The first branch includes a convolutional layer with a kernel size of 1×1, which can preserve the input data to the greatest extent. (The input to the DMSF module of the first layer is the historical fused image in the samples. The input to the DMSF module of the second layer is the output data of the AFA module of the first layer.) The input to the feature fusion module is the output data of the AFA module of the second layer group, and the input data to the dynamic multi-scale feature fusion module of the fourth layer group is the output data of the AFA module of the third layer group. The second branch includes a dilated convolutional layer with a kernel size of 3×3, which can effectively increase the receptive field. The third branch includes a depthwise separable convolutional layer with a kernel size of 5×5, which can effectively reduce the computational cost. The fourth branch includes a global average pooling layer and a deconvolutional layer set in sequence. The kernel size of the deconvolutional layer can be 3×3, which can effectively capture the global context.

[0090] The encoder also calculates the information entropy and signal-to-noise ratio of the output data for each branch, and then uses the Softmax function to calculate the information entropy and signal-to-noise ratio of the output data for each branch to obtain the fusion weight of each branch, using the formula... The output data from each branch is weighted and fused to obtain the weighted fusion result F. k Represents: the fusion weight of the k-th branch, g k This represents the output data of the k-th branch.

[0091] Each AFA module processes the received data as follows:

[0092] The received data (output data from the dynamic multi-scale feature fusion module within the same layer group) undergoes a Fast Fourier Transform (FFT) to obtain frequency domain data. Energy distribution is calculated based on this frequency domain data. The energy distribution result is then processed by a multilayer perceptron to generate a frequency domain mask. Based on the frequency domain mask, the frequency domain data is filtered. Finally, the filtered frequency domain data undergoes an Inverse Fast Fourier Transform (IFFT) to obtain spatial domain features. These spatial domain features are the output data of the AFA module. By directly suppressing noise bands in the frequency domain, the target feature bands—the key features used for disease identification—are enhanced, thus improving the accuracy of the trained intelligent disease identification model.

[0093] 2) The data processing intermediate layer includes Spatial and Temporal Memory Units (STMs). The output data of the last dynamic multi-scale feature fusion module of the encoder is used as the input data of the Spatial and Temporal Memory Units, and the output data of the Spatial and Temporal Memory Units is input to the decoder. By introducing memory units and gating mechanisms, the Spatial and Temporal Memory Units can effectively prevent the gradient from gradually decreasing in long-term sequences. Therefore, it can better analyze the historical fused images of sugarcane and help improve the recognition accuracy of the trained disease intelligent recognition model.

[0094] 3) The decoder receives the output data from the data processing intermediate layer, processes the output data, and obtains the disease identification results.

[0095] The output data of the spatiotemporal memory unit is an encoded high-dimensional feature vector, which contains spatiotemporal and contextual information of the historical fused images. The decoder is based on an attention mechanism and combines linear transformation operation, activation function and normalization operation to obtain the disease identification result.

[0096] S61. Construct multiple samples based on multiple historical fused images and corresponding disease identification results. Train the target deep learning model based on the multiple samples to obtain a trained intelligent disease identification model.

[0097] Optionally, in the above technical solution, after obtaining the fused image, the method further includes: performing illumination correction on the fused image to obtain a corrected fused image, then:

[0098] In S6, based on the fused image and the trained intelligent disease recognition model, the disease recognition result is obtained, including: based on the corrected fused image and the trained intelligent disease recognition model, the disease recognition result is obtained.

[0099] The specific process for performing illumination correction on the fused image is as follows:

[0100] ① The fused image is decomposed into low-frequency and high-frequency components using guided filtering. The low-frequency components include information such as the global brightness distribution and shadow areas of the fused image. A low-frequency illumination layer is generated based on the low-frequency components. The high-frequency components include detailed information such as the texture and edges of the fused image. A high-frequency detail layer is generated based on the high-frequency components.

[0101] In the guided filtering method, the radius of the filter kernel is dynamically adjusted based on local contrast. Specifically, the local contrast of the region surrounding each pixel in the fused image is calculated. Local contrast can be the standard deviation or variance of pixel values. Based on the overall contrast distribution of the fused image, a local contrast threshold can be determined through histogram analysis or an adaptive thresholding algorithm to distinguish between high-contrast and low-contrast regions. For each pixel, if its local contrast is greater than the local contrast threshold, a smaller filter kernel radius is used; otherwise, a larger filter kernel radius is used. Both the smaller and larger filter kernel radii are preset.

[0102] ② The low-frequency illumination layer is nonlinearly stretched to generate an initial illumination map, and the initial illumination map is optimized based on gradient domain constraints to obtain a corrected illumination map.

[0103] ③ The trained disease intelligent recognition model is used for the first time to recognize the fused image. The information entropy and signal-to-noise ratio of the output data of each branch calculated by the encoder are obtained. The confidence mask M is generated using the information entropy and signal-to-noise ratio of the output data of each branch. Specifically, it is implemented by the following formula:

[0104]

[0105] Where λ1 and λ2 are coefficients, for example, λ1 = 0.6 and λ2 = 0.4. In practical use, λ1 and λ2 can be determined by cross-validation. S1 represents the signal-to-noise ratio of the output data of the first branch, S2 represents the signal-to-noise ratio of the output data of the second branch, S3 represents the signal-to-noise ratio of the output data of the third branch, S4 represents the signal-to-noise ratio of the output data of the fourth branch, E1 represents the information entropy of the output data of the first branch, E2 represents the information entropy of the output data of the second branch, E3 represents the information entropy of the output data of the third branch, and E4 represents the information entropy of the output data of the fourth branch.

[0106] Indicates: will Substitute the Sigmoid function.

[0107] In this step, the information entropy and signal-to-noise ratio of the output data of each branch are directly used as prior knowledge, which can reduce computational redundancy, improve efficiency, and perform adaptive correction according to different fused images, which is more in line with reality and can effectively improve the accuracy of the corrected fused image.

[0108] ④ The fused image is corrected using a confidence mask M to obtain a corrected fused image. Specifically, the confidence mask M is used as a weight map and weighted with the fused image to obtain the corrected fused image.

[0109] Then, the trained disease intelligent identification model is used a second time to analyze the corrected fused image to obtain the disease identification result. This disease identification result is the final disease identification result, so that users can carry out planting management based on the disease identification result.

[0110] In another embodiment, the Retinex algorithm is used to perform illumination correction on the fused image.

[0111] Optionally, in the above technical solution, after acquiring the spectral data for each preset band, the method further includes:

[0112] Perform radiometric and atmospheric corrections on the spectral data for each preset band to obtain the corrected spectral data for each preset band. Then:

[0113] 1) In S3, weights are assigned to the spectral data of each preset band based on the quality of the spectral data of each preset band in the multispectral image data, including: assigning weights to the corrected spectral data of each preset band based on the quality of the corrected spectral data of each preset band.

[0114] 2) In S4, the corresponding grayscale image is weighted using the weight of the spectral data of each preset band to obtain the target grayscale image corresponding to the spectral data of each preset band, including: weighting the corresponding grayscale image using the weight of the corrected spectral data of each preset band to obtain the target grayscale image corresponding to the corrected spectral data of each preset band.

[0115] The process of radiometric correction for the spectral data of each preset band is as follows:

[0116] Based on the characteristics of the multispectral camera, the coefficients required for radiometric correction are calculated, and radiometric correction is performed on the spectral data of each preset band according to the coefficients required for radiometric correction. Radiometric correction can also be achieved through other methods.

[0117] The process of performing atmospheric correction on the spectral data for each preset band is as follows:

[0118] Atmospheric correction is performed on the spectral data of each preset band using physical model-based or statistical methods.

[0119] Before performing radiometric and atmospheric corrections, the process also includes format conversion, data cleaning, and interpolation of the spectral data for each preset band to ensure data integrity and processability.

[0120] Optionally, the above technical solution also includes:

[0121] S7. When the disease identification result includes a lesion area, the boundary of the lesion area is enhanced. Specifically, the lesion area undergoes contrast stretching using histogram equalization to improve the contrast between the lesion area and the normal area. Further sharpening is performed using a Butterworth high-pass filter to enhance clarity. Then, edge enhancement is achieved using the Sobel operator to enhance the boundary of the lesion area.

[0122] In the above embodiments, although the steps are numbered S1, S2, etc., they are only specific embodiments given by the present invention. Those skilled in the art can adjust the execution order of S1, S2, etc. according to the actual situation, which is also within the protection scope of the present invention. It can be understood that in some embodiments, some or all of the above embodiments may be included.

[0123] like Figure 2 As shown, an intelligent identification system 200 for early-stage sugarcane diseases according to an embodiment of the present invention includes a multispectral image data acquisition module 201, a data filtering module 202, a weight allocation module 203, a weighted processing module 204, an image fusion module 205, and a disease identification module 206.

[0124] The multispectral image data acquisition module 201 is used to: acquire multispectral image data of sampled sugarcane leaves in the early stage of growth. The multispectral image data includes spectral data of multiple bands and grayscale images corresponding to the spectral data of each band.

[0125] The data filtering module 202 is used to: obtain spectral data of each preset band from multispectral image data, and obtain the grayscale image corresponding to the spectral data of each preset band;

[0126] The weight allocation module 203 is used to: assign weights to the spectral data of each preset band according to the quality of the spectral data of each preset band;

[0127] The weighted processing module 204 is used to: perform weighted processing on the corresponding grayscale image using the weights of the spectral data of each preset band, so as to obtain the target grayscale image corresponding to the spectral data of each preset band;

[0128] Image fusion module 205 is used to: fuse all target grayscale images under preset constraints to obtain a fused image. The preset constraints are: the positions of the sampled sugarcane leaves in all target grayscale images overlap.

[0129] The disease identification module 206 is used to obtain disease identification results based on the fused image and the trained intelligent disease identification model.

[0130] Optionally, the above technical solution further includes an image correction module, which is used for:

[0131] After obtaining the fused image, illumination correction is performed on the fused image to obtain the corrected fused image;

[0132] The disease identification module 206 is specifically used to obtain disease identification results based on the corrected fused image and the trained intelligent disease identification model.

[0133] Optionally, the above technical solution also includes a spectral data correction module, which is used for:

[0134] After acquiring the spectral data for each preset band, radiometric and atmospheric corrections are performed on the spectral data for each preset band to obtain the corrected spectral data for each preset band.

[0135] The weight allocation module 203 is specifically used to: assign weights to the corrected spectral data of each preset band according to the quality of the corrected spectral data of each preset band;

[0136] The weighted processing module 204 is specifically used to: perform weighted processing on the corresponding grayscale image using the weights of the corrected spectral data of each preset band, so as to obtain the target grayscale image corresponding to the corrected spectral data of each preset band.

[0137] Optionally, the above technical solution further includes an enhancement processing module, which is used for:

[0138] When the disease identification results include diseased areas, the boundaries of the diseased areas are enhanced.

[0139] It should be noted that the beneficial effects of the intelligent identification system for early-stage sugarcane diseases provided in the above embodiments are the same as those of the intelligent identification method for early-stage sugarcane diseases described above, and will not be repeated here. Furthermore, the system provided in the above embodiments is only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the system can be divided into different functional modules according to the actual situation to complete all or part of the functions described above. In addition, the system and method embodiments provided in the above embodiments belong to the same concept, and their specific implementation process is detailed in the method embodiments, and will not be repeated here.

[0140] The intelligent identification system for early-stage sugarcane diseases of the present invention can be a computer program (including program code) running on a computer device. For example, the intelligent identification system for early-stage sugarcane diseases of the present invention is an application software that can be used to execute the corresponding steps in the intelligent identification method for early-stage sugarcane diseases of the present invention.

[0141] In some embodiments, the intelligent identification system for early-stage sugarcane diseases of the present invention can be implemented using a combination of hardware and software. As an example, the intelligent identification system for early-stage sugarcane diseases of the present invention can be a processor in the form of a hardware decoding processor, which is programmed to execute the intelligent identification method for early-stage sugarcane diseases of the present invention. For example, the processor in the form of a hardware decoding processor can be one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), or other electronic components.

[0142] The modules described in the embodiments of this invention can be implemented in software or hardware. The names of the modules are not, in some cases, limiting the scope of the module itself.

[0143] An electronic device according to an embodiment of the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements any of the above-mentioned intelligent identification methods for early-stage sugarcane diseases. That is, an electronic device according to an embodiment of the present invention may include, but is not limited to: a processor and a memory; the memory is used to store the computer program; the processor is used to execute the intelligent identification method for early-stage sugarcane diseases according to any embodiment of the present invention by calling the computer program.

[0144] In one alternative embodiment, an electronic device is provided, such as Figure 3 As shown, Figure 3 The illustrated electronic device 4000 includes a processor 4001 and a memory 4003. The processor 4001 and the memory 4003 are connected, for example, via a bus 4002. Optionally, the electronic device 4000 may further include a transceiver 4004, which can be used for data interaction between the electronic device and other electronic devices, such as sending and / or receiving data. It should be noted that in practical applications, the transceiver 4004 is not limited to one type, and the structure of the electronic device 4000 does not constitute a limitation on the embodiments of the present invention.

[0145] Processor 4001 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this invention. Processor 4001 may also be a combination that implements computational functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0146] Bus 4002 may include a path for transmitting information between the aforementioned components. Bus 4002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. Bus 4002 can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 3The bus 4002 is represented by only one thick line, but this does not mean that there is only one bus or one type of bus.

[0147] The memory 4003 may be ROM (Read Only Memory) or other types of static storage devices capable of storing static information and instructions, RAM (Random Access Memory) or other types of dynamic storage devices capable of storing information and instructions, or EEPROM (Electrically Erasable Programmable Read Only Memory), CD-ROM (Compact Disc Read Only Memory) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto.

[0148] The memory 4003 stores the application code (computer program) for executing the present invention, and its execution is controlled by the processor 4001. The processor 4001 executes the application code stored in the memory 4003 to implement the content shown in the foregoing method embodiments.

[0149] Among them, electronic devices can also be terminal devices, which can be any device that can install applications, including at least one of smartphones, tablets, laptops, desktop computers, smart speakers, smartwatches, smart TVs, and smart in-vehicle devices.

[0150] It should be noted that, Figure 3 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of the present invention.

[0151] An embodiment of the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements any of the above-described methods for intelligent identification of early-stage sugarcane diseases.

[0152] Alternatively, the computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a compact disc read-only memory (CD-ROM), magnetic tape, a floppy disk, and an optical data storage device, etc.

[0153] In an exemplary embodiment, a computer program product or computer program is also provided, which includes computer instructions stored in a computer-readable storage medium. The processor of an electronic device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the electronic device to perform any of the above-described intelligent identification methods for early-stage sugarcane diseases.

[0154] Computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof. These programming languages ​​include object-oriented programming languages—such as Java, Smalltalk, and C++—and conventional procedural programming languages—such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0155] It should be understood that the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of methods and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0156] The computer-readable storage medium provided in this invention can be, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EEPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0157] The aforementioned computer-readable storage medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to perform the method shown in the above embodiments.

[0158] The above description is merely a preferred embodiment of the present invention and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of disclosure in this invention is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this invention.

[0159] It should be noted that the terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and represent a limitation on a specific order or sequence. Where appropriate, the order of use for similar objects can be interchanged so that the embodiments of this application described herein can be implemented in an order other than that shown or described.

[0160] Those skilled in the art will recognize that this invention can be implemented as a system, method, or computer program product. Therefore, this invention can be specifically implemented in the following forms: it can be entirely hardware, entirely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software, generally referred to herein as a "circuit," "module," or "system." Furthermore, in some embodiments, this invention can also be implemented as a computer program product contained in one or more computer-readable media, which includes computer-readable program code.

[0161] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for intelligent identification of early-stage diseases in sugarcane, characterized in that, include: Acquire multispectral image data of sugarcane leaves in the early stage of growth. The multispectral image data includes spectral data of multiple bands and grayscale images corresponding to the spectral data of each band. From the multispectral image data, obtain the spectral data of each preset band, and obtain the grayscale image corresponding to the spectral data of each preset band; Weights are assigned to the spectral data of each preset band based on the quality of the spectral data of each preset band. The corresponding grayscale image is weighted by the weight of the spectral data of each preset band to obtain the target grayscale image corresponding to the spectral data of each preset band. Under preset constraints, all target grayscale images are fused to obtain a fused image. The preset constraints are: the positions of the sampled sugarcane leaves in all target grayscale images overlap. Based on the fused image and the trained intelligent disease recognition model, the disease recognition result is obtained.

2. The intelligent identification method for early-stage sugarcane diseases according to claim 1, characterized in that, After obtaining the fused image, the process also includes: The fused image is then subjected to illumination correction to obtain a corrected fused image; Based on the fused image and the trained intelligent disease identification model, the disease identification results are obtained, including: Based on the corrected fused image and the trained intelligent disease recognition model, the disease recognition result is obtained.

3. The intelligent identification method for early-stage sugarcane diseases according to claim 1, characterized in that, After acquiring the spectral data for each preset band, the process also includes: Radiometric and atmospheric corrections are performed on the spectral data of each preset band to obtain the corrected spectral data of each preset band. Weights are assigned to the spectral data of each preset band based on the quality of the spectral data in the multispectral image data, including: Weights are assigned to the corrected spectral data of each preset band based on the quality of the corrected spectral data of each preset band. The corresponding grayscale image is weighted using the weights of the spectral data for each preset band to obtain the target grayscale image corresponding to the spectral data of each preset band, including: The corresponding grayscale image is weighted by using the weights of the corrected spectral data of each preset band to obtain the target grayscale image corresponding to the corrected spectral data of each preset band.

4. A method for intelligent identification of early-stage sugarcane diseases according to any one of claims 1 to 3, characterized in that, Also includes: When the disease identification result includes a lesion area, the boundary of the lesion area is enhanced.

5. An intelligent identification system for early-stage sugarcane diseases, characterized in that, It includes a multispectral image data acquisition module, a data filtering module, a weight allocation module, a weighted processing module, an image fusion module, and a disease identification module; The multispectral image data acquisition module is used to: acquire multispectral image data of sampled sugarcane leaves in the early stage of growth, wherein the multispectral image data includes spectral data of multiple bands and grayscale images corresponding to the spectral data of each band. The data filtering module is used to: obtain spectral data of each preset band from the multispectral image data, and obtain the grayscale image corresponding to the spectral data of each preset band; The weight allocation module is used to: assign weights to the spectral data of each preset band according to the quality of the spectral data of each preset band; The weighted processing module is used to: perform weighted processing on the corresponding grayscale image using the weights of the spectral data of each preset band, so as to obtain the target grayscale image corresponding to the spectral data of each preset band. The image fusion module is used to: fuse all target grayscale images under preset constraints to obtain a fused image. The preset constraints are: the positions of the sampled sugarcane leaves in all target grayscale images overlap. The disease identification module is used to obtain disease identification results based on the fused image and the trained intelligent disease identification model.

6. The intelligent identification system for early-stage sugarcane diseases according to claim 5, characterized in that, It also includes an image correction module, which is used for: After obtaining the fused image, illumination correction is performed on the fused image to obtain the corrected fused image; The disease identification module is specifically used to: obtain disease identification results based on the corrected fused image and the trained intelligent disease identification model.

7. The intelligent identification system for early-stage sugarcane diseases according to claim 5, characterized in that, It also includes a spectral data correction module, which is used for: After acquiring the spectral data for each preset band, radiometric and atmospheric corrections are performed on the spectral data for each preset band to obtain the corrected spectral data for each preset band. The weight allocation module is specifically used to: assign weights to the corrected spectral data of each preset band according to the quality of the corrected spectral data of each preset band. The weighted processing module is specifically used to: perform weighted processing on the corresponding grayscale image using the weights of the corrected spectral data of each preset band, so as to obtain the target grayscale image corresponding to the corrected spectral data of each preset band.

8. A smart identification system for early-stage sugarcane diseases according to any one of claims 5 to 7, characterized in that, It also includes an enhancement processing module, which is used for: When the disease identification result includes a lesion area, the boundary of the lesion area is enhanced.

9. An electronic device, characterized in that, The method includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the intelligent identification method for early-stage sugarcane diseases as described in any one of claims 1 to 4.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the intelligent identification method for early-stage sugarcane diseases as described in any one of claims 1 to 4.

Images