Online detection and prevention method applied to forestry diseases and insect pests

Abstract

The present disclosure provides an online detection and prevention method applied to forestry diseases and insect pests, and relates to the field of data processing. The method comprises: performing multiple inspections on a to-be-inspected forest area with the same inspection rule to obtain hyperspectral images of a detection area and a control area in the to-be-inspected forest area; wherein the control area is an area containing only Pinus species and having been subjected to prevention and control of Monochamus alternatus larvae, and the detection area is other areas except the control area; constructing a reference waveband curve of Pinus species based on the hyperspectral images of the control area; comparing the hyperspectral curve of each pixel point in the hyperspectral images of the detection area with the reference waveband curve to identify abnormal areas; calculating the pest possibility index of each abnormal area, and determining the disease and insect pest area in each abnormal area based on the time series data of the pest possibility index; generating a warning index of each disease and insect pest area, and outputting warning information of each disease and insect pest area based on the warning index. The present disclosure can obtain accurate disease and insect pest areas and give warnings.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, in particular to an online detection and prevention method applied to forestry diseases and insect pests. BACKGROUND

[0002] Forestry diseases and insect pests refer to diseases and insect pests that harm forests and trees, which can have a serious impact on forest ecosystems and forestry production. Through the detection and prevention of forestry diseases and insect pests, signs of diseases and insect pests can be detected and prevented in time, so that the harm can be controlled in a small range, which is conducive to protecting the ecological balance of forests and ensuring the sustainable development of forestry economy. The pine brown longhorn beetle larva usually eats the phloem and xylem of the tree trunk and carries the pine wood nematode into the pine tree, causing the terminal of water and nutrient transport of the pine tree, and eventually causing the whole pine tree to die. In the detection and prevention of the above-mentioned forestry diseases and insect pests, through the monitoring of pine brown longhorn beetle larva diseases, the trees suffering from diseases and insect pests can be sprayed during the pine brown longhorn beetle larva period, so as to achieve the purpose of preventing pine brown longhorn beetle and pine wood nematode disease.

[0003] In the related art, the above-mentioned monitoring of pine brown longhorn beetle larva diseases can be implemented as follows: obtaining a hyperspectral detection result of trees in a forest area based on an unmanned aerial vehicle-mounted hyperspectral camera; constructing a fixed model to extract the spectral characteristics of the pine tree disease area to realize intelligent identification of pine brown longhorn beetle larva disease phenomena. However, during the operation of the unmanned aerial vehicle, external factors may cause errors in the hyperspectral detection result obtained by the unmanned aerial vehicle; and the types of trees in the forest area are relatively chaotic, and the reflectance spectral characteristics of different tree species are different, so there may be an overlap between the spectrum of a healthy tree and the spectrum curve of a pine brown longhorn beetle larva disease tree, which may be misjudged as an infected tree, so the above-mentioned method cannot accurately identify pine brown longhorn beetle larva diseases. SUMMARY

[0004] In order to solve the problem in the related art that the attitude change caused by external factors during the operation of the unmanned aerial vehicle may cause errors in the hyperspectral detection result; and the types of trees in the forest area are relatively chaotic, and the reflectance spectral characteristics of different tree species are different, so there may be an overlap between the spectrum of a healthy tree and the spectrum curve of a pine brown longhorn beetle larva disease tree, which may be misjudged as an infected tree, and the pine brown longhorn beetle larva disease cannot be accurately identified, the present application provides an online detection and prevention method applied to forestry diseases and insect pests, and the technical solution adopted is as follows:

[0005] The same inspection rule is used to inspect the to-be-inspected forest area multiple times to obtain hyperspectral images of a detection area and a control area in the to-be-inspected forest area; wherein the control area is an area in the to-be-inspected forest area that only contains pine species and has been prevented and controlled for pine brown longhorn beetle larvae, and the detection area is all other areas in the to-be-inspected forest area except the control area;

[0006] constructing a reference waveband curve of the Pinus species based on the hyperspectral image of the control region;

[0007] comparing the hyperspectral curve of each pixel point in the hyperspectral image of the detection region with the reference waveband curve to identify an abnormal region in the detection region;

[0008] calculating a pest possibility index of the pine beetle larva in each abnormal region, and determining a pest and disease region in each abnormal region based on time sequence data of the pest possibility index corresponding to each abnormal region;

[0009] generating a warning index of each pest and disease region, and outputting warning information of each pest and disease region based on the warning index; wherein the warning index is used to indicate the pest spread degree of the pest and disease region.

[0010] Further, the same inspection rule is used for multiple inspections of the forest area to be inspected to obtain the hyperspectral images of the detection region and the control region in the forest area to be inspected, comprising:

[0011] determining a detection period;

[0012] In each detection period, the detection region is inspected multiple times at fixed time points to obtain the hyperspectral images of the detection region; wherein the hyperspectral images of the detection region obtained at adjacent time points in a single inspection process have overlapping regions;

[0013] In each detection period, the control region is inspected multiple times at the fixed time points to obtain the hyperspectral images of the control region; wherein the hyperspectral images of the control region obtained at adjacent time points in a single inspection process have overlapping regions.

[0014] Further, after obtaining the hyperspectral images of the detection region and the control region in the forest area to be inspected, the method further comprises:

[0015] For each hyperspectral image of the detection region and the control region obtained in a single inspection, a tree region in the hyperspectral image is marked based on a semantic recognition algorithm;

[0016] For each single waveband gray image in each hyperspectral image, an edge point of the tree region is obtained based on an edge detection algorithm;

[0017] Based on the curvature and gray scale gradient amplitude of the edge point, a feature point in the edge point is determined;

[0018] Based on the feature points, image stitching is performed on the hyperspectral images acquired at different time points in a single inspection, to obtain a complete image of the detection area and a complete image of the control area.

[0019] Further, the feature points among the edge points are determined based on the curvature and the gray scale gradient amplitude of the edge points, including:

[0020] The curvature of each edge point, the gray scale gradient amplitude of each edge point, and the mean value of the gray scale gradient amplitude of all edge points are calculated.

[0021] Based on the curvature, the gray scale gradient amplitude of each edge point, and the mean value of the gray scale gradient amplitude of all edge points, a recognition degree index of each edge point is determined, and the edge points with the recognition degree index greater than a first preset threshold are marked as recognizable points.

[0022] The first frequency of each edge point being marked as a recognizable point in the gray scale images of all wavebands of the hyperspectral image, and the second frequency of all edge points being marked as recognizable points in the gray scale images of all wavebands of the hyperspectral image are determined.

[0023] The ratio of the first frequency and the second frequency is determined, and when the ratio of the first frequency and the second frequency is greater than a second preset threshold, the edge points corresponding to the first frequency are marked as feature points.

[0024] Further, the reference waveband curve of the Pinus species is constructed based on the hyperspectral images of the control area, including:

[0025] For the complete image of the control area, the reflectance extreme value ratio of each pixel point in the tree area at each waveband, and the slope extreme value ratio of the hyperspectral curve of all pixel points are calculated.

[0026] Based on the reflectance extreme value ratio and the slope extreme value ratio, a waveband consistency index of each waveband is determined.

[0027] The wavebands with the waveband consistency index greater than a third preset threshold are marked, and the wavebands that are marked in each inspection are determined as similar wavebands.

[0028] The mean value of the reflectance of all pixel points in the similar wavebands in the control area is determined as the normal reflectance reference value of the similar waveband.

[0029] The normal reflectance reference values of each similar waveband are arranged in order of wavelength, and a wavelength-reflectance curve is drawn to obtain the reference waveband curve.

[0030] Further, the method further comprises:

[0031] determining a detection sensitive wave band;

[0032] calculating a first tangent slope and a first reflectivity of each pixel point in the tree region in the detection sensitive wave band of the corresponding hyperspectral curve in the complete image of the detection region determined for a single inspection;

[0033] obtaining a second tangent slope and a second reflectivity of the detection sensitive wave band in the reference wave band curve;

[0034] determining an abnormality possibility of the corresponding pixel point based on the first tangent slope, the first reflectivity, the second tangent slope and the second reflectivity;

[0035] when the abnormality possibility is greater than a fourth preset threshold, marking the corresponding pixel point as a detection abnormal point, and determining the abnormal region based on the detection abnormal point.

[0036] Further, the method further comprises:

[0037] for each detection period, obtaining a minimum number of times that each pixel point in each abnormal region is marked as the detection abnormal point, a first average value of the abnormality possibility of all the pixel points in each abnormal region in multiple inspections, and a second average value of the number of times that the pixel points in all the abnormal regions are marked as the detection abnormal point;

[0038] determining the pest possibility index of the pine brown longicorn larva existing in each abnormal region based on the minimum number, the first average value and the second average value;

[0039] obtaining time sequence data of the pest possibility index, determining an effective continuous difference sequence length, a mean value of positive difference and an area expansion ratio of the abnormal region of the time sequence data;

[0040] determining a pest region feature index corresponding to the abnormal region based on the effective continuous difference sequence length, the mean value of the positive difference and the area expansion ratio;

[0041] when the pest region feature index is greater than a fifth preset threshold, determining the corresponding abnormal region as the pest region.

[0042] Further, the determination of the effective continuous difference sequence length of the time series data, the mean of the difference positive value and the area expansion ratio of the abnormal area comprises:

[0043] First-order difference is performed on the time series data of the pest possibility index, and values greater than 0 in the first-order difference result are marked; if there is a continuous difference marked, the number of continuous marks is determined as the effective continuous difference sequence length;

[0044] The average value of the first-order difference result with a positive calculation result is obtained as the mean of the difference positive value;

[0045] The first area of the abnormal area corresponding to the first detection period when the abnormal area first appears is obtained, and the second area of the abnormal area corresponding to the target detection period is obtained, and the area expansion ratio is determined based on the first area and the second area.

[0046] Further, the generation of the early warning index of each pest and disease area comprises:

[0047] The first pest and disease area characteristic index corresponding to the initial appearance of the pest and disease area, the second pest and disease area characteristic index corresponding to the target detection period of the pest and disease area, and the duration between the initial appearance of the pest and disease area and the target detection period are determined.

[0048] The first pest and disease area characteristic index, the second pest and disease area characteristic index and the duration are used to determine the early warning index of each pest and disease area.

[0049] Further, the output of the early warning information of each pest and disease area based on the early warning index comprises:

[0050] In the detection period, the early warning information of each pest and disease area is output in the order from large to small according to the early warning index, and the early warning information includes the area position, the pest and disease transmission path and the control suggestion of the corresponding pest and disease area.

[0051] The present application can have the following parts or all of the beneficial effects:

[0052] In the online detection and prevention method for forestry diseases and pests provided by the application, the same inspection rule is used for multiple inspections of a to-be-inspected forest area, and hyperspectral images of a detection area and a control area in the to-be-inspected forest area are obtained; the control area is an area in the to-be-inspected forest area that only contains Pinus species and has been subjected to prevention and control of Monochamus alternatus larvae, and the detection area is all other areas in the to-be-inspected forest area except the control area; a reference band curve of Pinus species is constructed based on the hyperspectral images of the control area; the hyperspectral curves of each pixel point in the hyperspectral images of the detection area are compared with the reference band curve to identify abnormal areas in the detection area; the possibility index of Monochamus alternatus larvae in each abnormal area is calculated, and the disease and pest area in each abnormal area is determined based on the time series data of the possibility index corresponding to each abnormal area; a warning index of each disease and pest area is generated, and warning information of each disease and pest area is output based on the warning index; wherein the warning index is used to indicate the degree of pest spread in the disease and pest area. The reference band curve of Pinus trees is constructed by the hyperspectral images of the control area obtained by multiple inspections of the to-be-inspected forest area, and the hyperspectral curves of each pixel point in the hyperspectral images of the detection area are compared with the reference band curve to preliminarily identify abnormal areas in the detection area, so that the disease and pest area can be determined according to the change trend of each abnormal area with time, the misjudgment of healthy trees can be screened out, and the accuracy of Monochamus alternatus disease tree identification is improved. In addition, the warning information of each disease and pest area can be output based on the warning index, so that the staff can prevent and control the disease and pest area according to the harm degree, which is conducive to improving the prevention and control effect.

[0053] It should be understood that the foregoing general description and the following detailed description are only exemplary and explanatory, and cannot limit the present disclosure. BRIEF DESCRIPTION OF DRAWINGS

[0054] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present application or prior art, the drawings needed in the embodiments or prior art description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0055] Figure 1 A flowchart of the online detection and prevention method for forestry diseases and pests according to an exemplary embodiment of the present disclosure is shown;

[0056] Figure 2 A schematic diagram of the hyperspectral detection result obtained at a certain position in the single inspection process in the online detection and prevention method for forestry diseases and pests according to an exemplary embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0057] In order to further clarify the technical means and effects taken by the present application to achieve the predetermined object of the application, the specific implementation, structure, characteristics and effects of the online detection and prevention method for forestry diseases and pests according to the present application are described in detail below in combination with the drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0058] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0059] The specific scheme of the online detection and prevention method for forestry diseases and pests provided by the present application is described in detail below in combination with the drawings.

[0060] Please refer to Figure 1 , which shows the method flowchart of the online detection and prevention method for forestry diseases and pests provided by one embodiment of the present application, as shown in Figure 1 , the online detection and prevention method for forestry diseases and pests specifically includes the following steps:

[0061] S110: multiple inspections are performed on the to-be-inspected forest area with the same inspection rule to obtain hyperspectral images of the detection area and the control area in the to-be-inspected forest area; wherein the control area is an area in the to-be-inspected forest area that only contains Pinus species and has been subjected to prevention and control of Monochamus alternatus larva, and the detection area is all other areas in the to-be-inspected forest area except the control area;

[0062] S120: a reference waveband curve of Pinus species is constructed based on the hyperspectral image of the control area;

[0063] S130: the hyperspectral curve of each pixel point in the hyperspectral image of the detection area is compared with the reference waveband curve to identify abnormal areas in the detection area;

[0064] S140: the pest possibility index of Monochamus alternatus larva in each abnormal area is calculated, and the disease and pest area in each abnormal area is determined based on the time series data of the pest possibility index corresponding to each abnormal area;

[0065] S150: an early warning index of each disease and pest area is generated, and early warning information of each disease and pest area is output based on the early warning index; wherein the early warning index is used to indicate the pest spread degree of the disease and pest area.

[0066] The present application constructs the reference band curve of Pinus trees by the hyperspectral image of the control area obtained by multiple inspections of the to-be-tested forest area, and preliminarily identifies the abnormal area in the detection area by comparing the hyperspectral curve of each pixel point in the hyperspectral image of the detection area with the reference band curve, so that the disease and pest area can be determined according to the change trend of each abnormal area with time sequence, the misjudgment of healthy trees can be screened out, and the accuracy of the pine brown longicorn larva disease tree identification is improved. In addition, the present application can output early warning information of each disease and pest area based on the early warning index, so that the staff can prevent and control the disease and pest area according to the harm degree, which is helpful to improve the prevention and control effect.

[0067] Next, each step of the above-mentioned online detection and prevention method applied to forestry diseases and pests will be described in detail:

[0068] In step S110, the to-be-tested forest area is inspected multiple times by the same inspection rule to obtain the hyperspectral image of the detection area and the control area in the to-be-tested forest area; wherein the control area is a region in the to-be-tested forest area that only contains Pinus species and has been prevented and controlled for pine brown longicorn larvae, and the detection area is all other regions in the to-be-tested forest area except the control area.

[0069] In the embodiment of the present application, the above-mentioned to-be-tested forest area is a forest area for pine brown longicorn larva disease monitoring based on the online detection and prevention method applied to forestry diseases and pests provided by the present application.

[0070] In the embodiment of the present application, the above-mentioned control area is used to provide a reference standard for the detection area. By selecting a healthy tree area and processing its hyperspectral image, the spectral characteristics and other data of normal trees can be obtained, providing a clear reference for the detection area. By comparing the spectral data and other data of the detection area with the control area, it can be judged whether the trees in the detection area have health problems or other abnormal conditions. In addition, the above-mentioned control area can also exclude irrelevant factor interference. By setting the same environmental conditions for the data acquisition process of the detection area and the control area (i.e. obtaining the hyperspectral images of the control area and the detection area in the to-be-tested forest area by the same inspection rule), the interference of irrelevant factors such as unmanned aerial vehicle flight height, lighting conditions, and atmospheric environment can be excluded to a certain extent, so that the detection result can more truly reflect the actual situation of the trees.

[0071] Since the online detection and prevention method for forestry diseases and insect pests provided in the embodiments of the present application is mainly aimed at the Monochamus alternatus larva disease, and the harm of the Monochamus alternatus larva disease is mainly aimed at the Pinus tree species, the selection method of the above-mentioned control area can be as follows: a small range area with only Pinus tree species in the to-be-detected forest area is selected, and the Monochamus alternatus larva prevention and control is performed in advance to obtain a healthy tree area as a control area to provide a reference for the detection area. The implementation of the above-mentioned Monochamus alternatus larva prevention and control can be as follows: the small range area with only Pinus tree species selected is uniformly sprayed with a white rot fungus spore suspension with a suitable concentration to avoid the influence of the Monochamus alternatus larva disease on the trees.

[0072] In the embodiments of the present application, the detection area is the area in the to-be-detected forest area that needs to be detected for the Monochamus alternatus larva disease. Exemplarily, other areas in the to-be-detected forest area except the control area can be set as the detection area.

[0073] In the embodiments of the present application, in order to exclude the interference of irrelevant factors such as the flight height of the unmanned aerial vehicle, the light condition, and the atmospheric environment, the detection area and the control area need to be inspected based on the same inspection rule. Exemplarily, the above-mentioned multiple inspections of the to-be-inspected forest area based on the same inspection rule to obtain the hyper-spectral images of the detection area and the control area in the to-be-inspected forest area can be implemented as follows: a detection period is determined; in each detection period, the detection area is inspected multiple times at a fixed time point to obtain the hyper-spectral images of the detection area; wherein the hyper-spectral images of the detection area obtained at adjacent time points in a single inspection process have an overlapping area; in each detection period, the control area is inspected multiple times at a fixed time point to obtain the hyper-spectral images of the control area; wherein the hyper-spectral images of the control area obtained at adjacent time points in a single inspection process have an overlapping area.

[0074] Specifically, in an actual detection scenario, the hyper-spectral images of the detection area can be obtained by the following process: the unmanned aerial vehicle-borne hyper-spectral camera is used to perform multiple inspections on the detection area at a fixed height according to a fixed path to obtain the hyper-spectral images of the detection area. In the above-mentioned inspection process, one day can be taken as one detection period, and in each detection period, the inspection is performed at a fixed time of each day (for example, in the time period from 11 am to 1 pm), and multiple inspections are performed each day (at least 3 times), and in each inspection, the hyper-spectral images of different positions are obtained by the unmanned aerial vehicle-borne hyper-spectral camera at a uniform speed and vertically to the canopy to ensure that there is an overlapping area in the hyper-spectral images obtained at adjacent time points during the inspection.

[0075] In order to exclude irrelevant factors, the control area also needs to be collected by the same inspection rule, that is, the control area also needs to be inspected at the fixed time every day (for example, in the period from 11 am to 1 pm), multiple times a day (at least 3 times), and the hyperspectral images of different positions in the control area are obtained by the unmanned aerial hyperspectral camera at a uniform speed and vertically to the canopy, and the overlapping area in the hyperspectral images obtained at adjacent times is ensured. Preferably, in order to avoid the interference of the white film formed by the Beauveria bassiana suspension on the spectral detection result, the control area needs to be inspected multiple times within 12-24 hours after spraying the Beauveria bassiana suspension.

[0076] After obtaining the hyperspectral images of the above-mentioned detection area and control area, the obtained data can be further stored and preprocessed. The above-mentioned preprocessing process includes spectral radiation correction and geometric correction of the hyperspectral images. The spectral radiation correction is used to eliminate the influence of factors such as sensor characteristics and atmospheric conditions on spectral radiation measurement, so as to obtain data that truly reflects the spectral radiation characteristics of the ground object. The geometric correction is used to eliminate the geometric distortion of the hyperspectral images caused by various factors in the acquisition process, so that the ground object position in the hyperspectral images matches the actual geographical position.

[0077] In the embodiment of the present application, the above-mentioned hyperspectral image is optical image data containing hundreds of continuous narrow wave bands, covering the visible light (400-700 nm), near-infrared (700-1300 nm) and short-wave infrared (1300-2500 nm) ranges, and each pixel point corresponds to a complete spectral curve, which can reflect the fine spectral characteristics of the ground object, such as Figure 2 As shown in FIG. 5, it is a schematic diagram of the hyperspectral detection result obtained at a position in a single inspection process.

[0078] Since the to-be-tested forest area usually has a large range, the hyperspectral image obtained by the UAV in a single time usually only contains part of the to-be-tested forest area. Therefore, in order to obtain the complete hyperspectral image of the detection area and the control area, the embodiments of the present application can use the complementary information of the overlapping area between two hyperspectral images obtained at adjacent times in a single inspection process to splice the images obtained by the UAV at different positions, and obtain the complete hyperspectral image of the detection area and the control area. Exemplarily, the process can be implemented as follows: for each hyperspectral image of the detection area and the control area obtained in a single inspection, the tree area in the hyperspectral image is marked based on a semantic recognition algorithm; for the gray scale image of each single band in each hyperspectral image, the edge points of the tree area are obtained based on an edge detection algorithm; the feature points in the edge points are determined based on the curvature and the gray scale gradient amplitude of the edge points; the hyperspectral images obtained at different times in a single inspection are spliced based on the feature points, and the complete image of the detection area of the detection area and the complete image of the control area of the control area are obtained.

[0079] In the above, the feature points are determined based on the curvature and the gray scale gradient amplitude of the edge points, which can be implemented as follows: the curvature of each edge point, the gray scale gradient amplitude of each edge point, and the mean value of the gray scale gradient amplitude of all edge points are calculated; the recognition degree index of each edge point is determined based on the curvature, the gray scale gradient amplitude of each edge point, and the mean value of the gray scale gradient amplitude of all edge points, and the edge point with the recognition degree index greater than a first preset threshold is marked as a recognizable point; the first frequency of each edge point being marked as a recognizable point in the gray scale image of all bands of the hyperspectral image is determined, and the second frequency of all edge points being marked as a recognizable point in the gray scale image of all bands of the hyperspectral image is determined; the ratio of the first frequency and the second frequency is determined, and when the ratio of the first frequency and the second frequency is greater than a second preset threshold, the edge point corresponding to the first frequency is marked as a feature point.

[0080] In the following, the process of selecting feature points and image splicing in a specific embodiment is described in detail:

[0081] Taking the kth hyperspectral image obtained in a single inspection as an example, the process of determining the feature points can be implemented as follows:

[0082] S1-1: Mark the tree area in the hyperspectral image based on a semantic recognition algorithm.

[0083] In the embodiments of the present application, the hyperspectral image is segmented, and the tree area in the hyperspectral image is marked by using a semantic recognition algorithm, and the remaining area is masked. The semantic recognition algorithm can be implemented by using a 3D-UNet model (three-dimensional U-shaped network model), and the marking result of the tree area can be mapped to the gray scale image of all bands of the hyperspectral image.

[0084] S2-1: For any single-band grayscale image of the hyperspectral image, obtain the edge points in the above-mentioned tree region based on the edge detection algorithm.

[0085] In this application embodiment, for example, the Canny operator can be used to obtain the edge points of tree regions in any single-band grayscale image.

[0086] S3-1: Determine feature points in edge points based on the curvature and gray-level gradient magnitude of edge points.

[0087] In this embodiment, the aforementioned feature points are used to stitch together hyperspectral images from a single inspection, and need to reflect the morphological and spectral features of the tree region in the current image. For example, the aforementioned morphological features are related to the curvature of edge points; the greater the curvature of an edge point, the greater the likelihood that it is a turning point or a point of abrupt texture change at the edge of a tree region. The aforementioned spectral features are related to the gray-level gradient amplitude of edge points; that is, the greater the gray-level gradient amplitude in the gray-level image corresponding to the current band, the greater the likelihood that an edge point is a ground feature edge point, and the higher the identifiability of that point. Therefore, feature points in edge points can be determined based on the curvature and gray-level gradient amplitude of edge points. Specifically, in this embodiment, the identifiability index of each edge point can be calculated using the curvature and gray-level gradient amplitude of the edge points. Taking the i-th edge point in the j-th tree region of the current single-band gray-level image as an example, the aforementioned identifiability index can be calculated using the following formula:

[0088]

[0089] Where, δ i,j c represents the discrimination index of the i-th edge point in the j-th tree region of the current single-band grayscale image. i,j G represents the curvature of that edge point. i,j This represents the grayscale gradient magnitude at the edge point. This represents the gray-level gradient magnitude g of all edge points in the current single-band grayscale image. i,j The mean; The larger the value, the greater the change in reflectance of the edge point relative to other edge points in the current single-band grayscale image. The sigmoid function is a normalization function with a range of (0,1).

[0090] After calculating the identifiability index of each edge point in the current single-band grayscale image using the above formula, the identifiable points among each edge point can be determined based on this identifiability index. For example, δ can be... i,j Edge points with a value greater than 0.5 (where 0.5 is the value of the first preset threshold mentioned above) are marked as identifiable points in the current single-band grayscale image.

[0091] To improve the stability of the selected feature points, the embodiments of the present application can determine the feature points by the frequency of each edge point being marked as a recognizable point in the single-band gray images of all bands of the hyperspectral image, thereby improving the stability of the feature points. The higher the frequency of a certain edge point being marked as a recognizable point in all bands, the more stable the edge point is, and the greater the possibility of the edge point being a feature point is. Exemplarily, the edge points with a frequency of being marked as a recognizable point in the single-band gray images of all bands of the hyperspectral image greater than 0.5 (i.e., the second preset threshold) can be determined as the feature points. (where 0.5 is the value of the second preset threshold) are the feature points, where n i,j (i.e., the first frequency) represents the frequency of the i-th edge point being marked as a recognizable point, (i.e., the second frequency) represents the frequency of all edge points being marked as recognizable points.

[0092] Specifically, assuming that the hyperspectral image has a total of 10 bands, in the single-band gray images corresponding to each band, the recognition degree index of each edge point is determined based on the above steps, if the i-th edge point is marked as a recognizable point in the single-band gray images corresponding to 7 bands and the recognition degree index of the i-th edge point in the single-band gray images corresponding to the 7 bands is calculated by the above steps, and the value of the recognition degree index is greater than 0.5 (i.e., the i-th edge point is marked as a recognizable point in 7 bands), then the value of n i,j is 7; if the marking of all edge points in 10 bands is counted, it is found that the total number of times that all edge points are marked as recognizable points in 10 bands is 500, then the value of n i,j is 10; if the marking of all edge points in 10 bands is counted, it is found that the total number of times that all edge points are marked as recognizable points in 10 bands is 500, then the value of n At this time, the value of n Since 0.014 < 0.5, the stability of the i-th edge point as a feature point is relatively low, and it cannot be selected as a feature point.

[0093] S4-1: Based on the feature points determined in steps S1 to S3, the hyperspectral images of adjacent time points in a single inspection process are spliced to obtain a complete image of the detection region or a complete image of the control region corresponding to the inspection process.

[0094] In the embodiments of the present application, the process of image splicing based on feature points can be implemented based on the SuperGlue algorithm (SuperGlue algorithm), which is a deep learning algorithm for image feature matching, used to match feature points in two images and determine feature points representing the same physical position in different images. After using the SuperGlue algorithm to match the extracted feature points and finding the corresponding feature points in the hyperspectral images of different time points, the transformation matrix between the images can be calculated according to the matched feature points, and the hyperspectral images of different time points in the inspection process can be spliced according to the calculated transformation matrix.

[0095] In step S120, a reference band curve of the poplar species is constructed based on the hyperspectral image of the control region.

[0096] In the embodiment of the present application, the reference waveband curve is used to characterize the reference curve of the spectral characteristics of healthy Pinus trees, and provides a reliable comparison standard for pest detection. The reference waveband curve can be generated by statistically analyzing the hyperspectral data of healthy Pinus trees, screening wavebands with high consistency, and calculating the average reflectivity thereof. The hyperspectral data of healthy Pinus trees can be the complete hyperspectral image of the control area generated based on the above steps (i.e., the complete image of the control area).

[0097] For example, the reference waveband curve of Pinus species constructed based on the hyperspectral image of the control area can be implemented as follows: for the complete image of the control area, the reflectivity extreme ratio of each pixel in the tree region at each waveband and the slope extreme ratio of the hyperspectral curve of all pixels are calculated; the waveband consistency index of each waveband is determined based on the reflectivity extreme ratio and the slope extreme ratio; the wavebands with a waveband consistency index greater than a third preset threshold are marked, and the wavebands marked in each inspection are determined as similar wavebands; the average reflectivity of the pixels in all tree regions in the control area at the similar wavebands is determined as the normal reflectivity reference value of the similar wavebands; the normal reflectivity reference values of the similar wavebands are arranged in order of wavelength, and a wavelength-reflectivity curve is drawn to obtain the reference waveband curve.

[0098] In the embodiment of the present application, the similar waveband refers to a waveband with high consistency in reflectivity and spectral morphology, which is screened out by analyzing the hyperspectral data of healthy Pinus trees in the control area, and can stably characterize the optical properties of healthy Pinus trees, and provide a reliable reference for subsequent detection of pine brown longhorn beetle larva pests in the detection area.

[0099] In the above steps, the Pinus trees in the control area are protected from pine brown longhorn beetle larvae pests, and the hyperspectral curves of different Pinus trees have a high overlap rate in the overall spectral morphology. Therefore, the normal reflectivity reference value of Pinus trees at similar wavebands in the detection forest environment can be detected based on the consistency analysis of the hyperspectral data of Pinus trees in the control area, and the reference waveband curve can be determined based on the normal reflectivity reference value. Specifically, taking the complete image of the control area obtained by splicing a single inspection of the control area as an example, the process of marking similar wavebands and determining the reference waveband curve is described in detail as follows:

[0100] S1-2: Calculate the reflectivity extreme ratio of each pixel in the tree region of the complete image of the control area at each waveband.

[0101] In the embodiments of the present application, the reflectance extreme value ratio is an index for characterizing the similarity of the reflectance of the tree region pixel points of the complete image of the control region in the corresponding waveband. Specifically, taking the zth waveband of the complete image of the control region as an example, the reflectance extreme value ratio a of the tree region pixel points of the complete image of the control region in the zth waveband is calculated according to the following formula: z The reflectance extreme value ratio a can be calculated according to the following formula:

[0102]

[0103] wherein r z,min is the minimum value of the reflectance of the hyperspectral curve corresponding to all the pixel points of the tree region of the complete image of the control region in the zth waveband, and r z,max is the maximum value of the reflectance of the hyperspectral curve corresponding to all the pixel points of the tree region of the complete image of the control region in the zth waveband. z The greater the reflectance extreme value ratio a, the more similar the reflectance of all the trees in the zth waveband.

[0104] S2-2: Calculate the slope extreme value ratio of the hyperspectral curve of all the pixel points in each waveband.

[0105] In the embodiments of the present application, the slope extreme value ratio is an index for characterizing the similarity of the reflectance change trend of the pixel points of the complete image of the control region in the corresponding waveband. Specifically, taking the zth waveband of the complete image of the control region as an example, the slope extreme value ratio b of the pixel points of the complete image of the control region in the zth waveband is calculated according to the following formula: z The slope extreme value ratio b can be calculated according to the following formula:

[0106]

[0107] wherein k z,min is the minimum value of the tangent slope in the zth waveband in the hyperspectral curve corresponding to all the pixel points of the complete image of the control region, and k z,max is the maximum value of the tangent slope in the zth waveband in the hyperspectral curve corresponding to all the pixel points of the complete image of the control region.

[0108] S3-2: Determine the waveband consistency index of each waveband based on the reflectance extreme value ratio and the slope extreme value ratio.

[0109] In the embodiments of the present application, the waveband consistency index is used to evaluate the overall stability of each waveband in combination with the reflectance extreme value ratio and the slope extreme value ratio. Specifically, taking the zth waveband of the complete image of the control region as an example, the waveband consistency index u of the zth waveband is calculated according to the following formula: z The waveband consistency index u can be calculated according to the following formula:

[0110] u z = premnmx(a z×b z )

[0111] Among them, u z is the band consistency index for the z-th band, used to represent the consistency of reflectance of all tree region pixels in the z-th band in the complete image of the current control area; premnmx is the normalization function with a value range of [-1,1].

[0112] S4-2: Based on the above band consistency index, determine the similar bands in the complete image of the control area.

[0113] In this embodiment, the determination of similar bands in the complete image of the comparison region based on the aforementioned band consistency index can be achieved as follows: bands with a band consistency index greater than a third preset threshold are marked, and bands marked in each inspection are determined as similar bands. Specifically, taking the z-th band of the complete image of the comparison region as an example, u z Bands with a normalization result greater than 0.5 (where 0.5 is the value of the third preset threshold mentioned above) are marked. If the z-th band is marked in every inspection process, the z-th band is marked as a similar band to the pine trees in the forest area to be inspected, and the average reflectance of the pixels in the z-th band in all tree areas in the control area is taken as the normal reflectance reference value of the z-th band.

[0114] S5-2: Determine the reference band curve based on the normal reflectance reference values ​​of each similar band.

[0115] In this embodiment of the application, the normal reflectance reference values ​​of all similar bands are plotted at the corresponding positions in the wavelength (horizontal axis)-reflectance (vertical axis) coordinate system to obtain the reference band curves of the pine trees in the forest area to be inspected.

[0116] In step S130, the hyperspectral curves of each pixel in the hyperspectral image of the detection area are compared with the reference band curve to identify abnormal areas in the detection area.

[0117] In this embodiment, the aforementioned abnormal area refers to the region within the detection area where pine sawyer beetle larvae may be present. Because pine sawyer beetle larvae damage the internal structure of pine trees, causing imbalances in water metabolism and decreased photosynthesis, the hyperspectral curves of the trees change. Furthermore, due to different growing environments, the spectral curves of different trees also vary. The aforementioned reference band curve serves as a baseline curve characterizing the spectral features of healthy pine trees in the tested forest area. Therefore, by comparing the differences between the hyperspectral curves of pine trees within the detection area and the aforementioned reference band curve, the likelihood of pine sawyer beetle larvae infestation in the corresponding trees can be analyzed, and the aforementioned abnormal area can be identified accordingly.

[0118] Exemplarily, the above-mentioned comparison of the hyperspectral curve of each pixel point in the hyperspectral image of the detection region with the reference waveband curve to identify the abnormal region in the detection region can be implemented as follows: determining a detection sensitive waveband; calculating, for the complete image of the detection region determined in a single inspection, a first tangent slope and a first reflectivity of each pixel point of the tree region in the complete image of the detection region at the detection sensitive waveband of the corresponding hyperspectral curve; obtaining a second tangent slope and a second reflectivity of the detection sensitive waveband in the reference waveband curve; determining an abnormal possibility of the corresponding pixel point based on the first tangent slope, the first reflectivity, the second tangent slope and the second reflectivity; and when the abnormal possibility is greater than a fourth preset threshold, marking the corresponding pixel point as a detection abnormal point, and determining an abnormal region based on the detection abnormal point.

[0119] In the embodiments of the present application, the above-mentioned detection sensitive waveband refers to a specific wavelength interval in which the reflectivity of the vegetation suffering from pests and diseases changes significantly and can be detected. That is, in the comparison of the normal reference spectrum of healthy vegetation with the spectrum of vegetation suffering from pests and diseases, the waveband with the most significant reflectivity difference. Since the larvae of the Monochamus alternatus Hope bore into the pine trees, causing the chlorophyll of the pine leaves to decrease, the reflectivity of the near-infrared waveband shows a downward trend, so for the Pinus species suffering from the Monochamus alternatus Hope pest and disease, the above-mentioned detection sensitive waveband is concentrated in the near-infrared waveband (700nm-1300nm).

[0120] Next, in a specific embodiment, the process of determining the abnormal region is described in detail:

[0121] In the embodiments of the present application, the reference waveband curve obtained based on the above-mentioned step S120 is taken as a reference benchmark to analyze the hyperspectral curve corresponding to the pixel point at different positions in the detection region. The greater the difference between a certain hyperspectral curve and the normal reference waveband curve, the greater the possibility that the position of the pixel point corresponding to the hyperspectral curve is abnormal. Specifically, taking a single pixel point of the tree region in the complete image of the detection region obtained in the single inspection process as an example, if there is a near-infrared waveband in the reference waveband curve, the corresponding waveband (i.e., the waveband with a wavelength in the range of 700nm-1300nm in the reference waveband curve) is marked as a detection sensitive waveband. Taking the i-th pixel point in the tree region in the complete image of the detection region as an example, the abnormal possibility v i which can be calculated by the following formula:

[0122]

[0123] wherein v i represents the possibility of the i-th pixel point in the tree region in the complete image of the detection region being abnormal, r i,z' represents the first reflectivity of the i-th pixel point at the z 'The reflectance of each detection-sensitive band (i.e., the first reflectance mentioned above), k i,z' The hyperspectral curve corresponding to the i-th pixel at the z-th position ' The tangent slope of each detection sensitive band (i.e., the first tangent slope mentioned above), r z' k represents the reflectance of the z'-th detection sensitive band in the reference band curve (i.e., the second reflectance mentioned above). z' Indicates the z-th band in the reference band curve ' The tangent slope of each detection sensitive band (i.e., the second tangent slope mentioned above); n z' This indicates the total number of sensitive bands detected.

[0124] In the above formula, The larger the value, the more likely the i-th pixel is to be at the z-th position. ' The reflectance of the z-th sensitive band in the detection band is compared to that of the reference band curve. ' The smaller the reflectance reference value for each sensitive wavelength band, the better; |k i,z' -k z' The larger the | value, the more significant the hyperspectral curve corresponding to the i-th pixel at the z-th pixel. ' The greater the difference between the shape of the morphology at the detection sensitive band and the shape of the reference band curve, the greater the possibility that the ith pixel has a detection anomaly. If the hyperspectral curve of the ith pixel has a smaller reflectance in the detection sensitive band compared to the reference band curve, and the slope of the tangent at the corresponding position of the ith pixel differs more from that of the detection sensitive band, the greater the possibility that the ith pixel has a detection anomaly.

[0125] After calculating the probability of anomalies for each pixel in the tree region of the complete image of the detection area using the above formula, the detection anomaly points can be determined based on the probability of anomalies. Specifically, the probability of anomalies for each pixel can be normalized to (0,1), and pixels with a normalization result greater than 0.5 (where 0.5 is the value of the fourth preset threshold mentioned above) can be marked as detection anomalies.

[0126] In this embodiment of the application, after determining the abnormal points in each pixel of the tree region of the complete image of the detection area through the above process, abnormal regions in the tree region of the complete image of the detection area can also be determined based on the above-mentioned abnormal points.

[0127] For example, the above-mentioned identification of abnormal regions can be achieved as follows:

[0128] S1-3: Extract the connected components containing all detected outlier points as the outlier detection regions.

[0129] Since the UAV is affected by outdoor environment interference and air flow when flying outdoors, the attitude of the UAV may be unstable, which may change the incident angle of the hyperspectral detection, making the hyperspectral detection result inaccurate, and finally leading to the detection abnormal area obtained by multiple inspections being different.

[0130] To solve this problem, after extracting all the detection abnormal points in the connected domain as the detection abnormal area, the following steps are performed to determine the abnormal area based on the detection abnormal area:

[0131] S2-3: Match the detection area complete image obtained by each inspection in a single detection period based on the selected feature points, wherein the image size of the detection area complete image obtained by each splicing is ensured to be the same when splicing the hyperspectral image obtained by a single inspection in step S110.

[0132] S3-3: Establish an empty image with an initial gray value of 0 for all pixel points, mark the pixel points of all detection abnormal areas in the empty image, and add 1 to the gray value of the pixel points at the corresponding position for each marking.

[0133] S4-3: Obtain the foreground area by threshold segmentation as the abnormal area of the current detection period.

[0134] In the embodiments of the present application, the threshold value can be selected according to the misjudgment tolerance and detection requirements. For example, if 3 inspections are performed in a detection period and the threshold value is set to 2, a certain pixel point needs to be marked as a detection abnormal point in at least 2 detections to be segmented to the foreground area.

[0135] In step S140, the possibility index of the pine beetle larva in each abnormal area is calculated, and the pest damage area in each abnormal area is determined based on the time sequence data of the possibility index corresponding to each abnormal area.

[0136] Since the hyperspectral detection results of the remaining healthy tree species are similar to the hyperspectral detection curve of the pine tree suffering from the pine beetle larva in the actual detection process, the healthy tree species may be misjudged as the pine beetle larva in the process of determining the abnormal area. Therefore, the present application further distinguishes the above-mentioned situation based on the trend of the abnormal degree of each abnormal area changing with time sequence after determining the abnormal area by the above-mentioned process to determine the pest damage area in the abnormal area.

[0137] In the embodiments of the present application, the pest possibility index is used to represent the index of the possibility of the existence of the Monochamus alternatus larvae in the corresponding abnormal area in the current detection period, and the abnormal area can be distinguished as a healthy tree species or a pest and disease area of the Monochamus alternatus larvae based on the pest possibility index.

[0138] In the development process of the Monochamus alternatus larvae disaster, the abnormal degree of the abnormal area in the initial boring stage of the larvae may fluctuate to a certain extent, but as the larvae penetrate into the xylem, the abnormal degree will continue to increase significantly, and the state of the healthy trees of the remaining tree species that are misjudged as abnormal areas is relatively stable, and the fluctuation of the corresponding spectral curve at different times is small, that is, the healthy trees of the remaining tree species that are misjudged as abnormal areas will maintain a relatively consistent abnormal degree for a long time. Based on the above analysis, the pest and disease area of the Monochamus alternatus larvae can be determined by the consistency of the abnormal degree of each abnormal area.

[0139] For example, the calculation of the pest possibility index of the existence of the Monochamus alternatus larvae in each abnormal area and the determination of the pest and disease area in each abnormal area based on the time sequence data of the pest possibility index of each abnormal area can be implemented as follows: for each detection period, the minimum number of times that each pixel point in each abnormal area is marked as a detection abnormal point, the first average value of the abnormal possibility of all pixel points in each abnormal area in multiple inspections, and the second average value of the number of times that the pixel points in all abnormal areas are marked as detection abnormal points are obtained; the pest possibility index of the existence of the Monochamus alternatus larvae in each abnormal area is determined based on the minimum number, the first average value and the second average value; the time sequence data of the pest possibility index is obtained, the effective continuous difference sequence length of the time sequence data, the average value of the positive difference value and the area expansion ratio of the abnormal area are determined; the pest and disease area characteristic index corresponding to the abnormal area is determined based on the effective continuous difference sequence length, the average value of the positive difference value and the area expansion ratio; when the pest and disease area characteristic index is greater than a fifth preset threshold, the corresponding abnormal area is determined as a pest and disease area.

[0140] The determination of the effective continuous difference sequence length, the average value of the positive difference value and the area expansion ratio of the abnormal area of the time sequence data can be implemented as follows: first-order difference is performed on the time sequence data of the pest possibility index, and the values greater than 0 of the first-order difference result are marked; if there is a continuous difference marked, the number of continuous marks is determined as the effective continuous difference sequence length; the average value of the first-order difference result with a positive result is calculated to obtain the average value of the positive difference value; the first area corresponding to the detection period when the abnormal area first appears is obtained, and the second area corresponding to the target detection period of the abnormal area is obtained, and the area expansion ratio is determined based on the first area and the second area.

[0141] In the following, in a specific embodiment, the process of determining the pest and disease area is described in detail.

[0142] S1-4: Based on the feature point pairs, the detection area in the adjacent detection period corresponding to the detection area complete image is matched, and the abnormal area pair with the largest overlapping area in the adjacent detection period is obtained.

[0143] In the embodiments of the present application, taking the Tth period and the (T-1)th period as an example, the above process can be implemented as follows: a matching method based on a graph neural network is used to match the feature point pairs of the Tth period and the (T-1)th period, and the coordinates of the matched point pairs are output; a random sample consensus (RANSAC) algorithm is used to estimate an affine transformation matrix, and the Tth period image is aligned to the coordinate system of the (T-1)th period; all abnormal area pairs corresponding to the abnormal areas of the (T-1)th period and the Tth period are determined, and the overlapping area between the two abnormal areas in each abnormal area pair is calculated; and the abnormal area pair with the largest overlapping area in all abnormal area pairs of the (T-1)th period and the Tth period is obtained.

[0144] S2-4: The pest possibility index of each abnormal area existing in the Monochamus alternatus larva is calculated.

[0145] In the embodiments of the present application, taking the jth abnormal area as an example, for a single detection period, the more times the pixel points in the jth abnormal area are marked as detection abnormal points, the greater the abnormal possibility v i of the pixel points in the multiple inspection processes of the detection period, and the greater the possibility of the jth abnormal area existing in the Monochamus alternatus larva. Therefore, the pest possibility index of the jth abnormal area can be calculated by the following formula:

[0146]

[0147] wherein q j represents the possibility of the jth abnormal area existing in the Monochamus alternatus larva in the current detection period (i.e., the above pest possibility index), n j,min represents the minimum value of the number of times each pixel point in the jth abnormal area is marked as a detection abnormal point, represents the average value of the number of times each pixel point in all abnormal areas is marked as a detection abnormal point (i.e., the above second average value), represents the average value of the average abnormal possibility v of all pixel points in the jth abnormal area in each inspection process (i.e., the above first average value); The greater the number of times the pixel points in the jth abnormal area are marked, the greater the number of times the pixel points in the jth abnormal area are marked.

[0148] Specifically, if there are 100 pixel points in the jth abnormal area, 20 pixel points are marked as detection abnormal points for 3 times, 40 pixel points are marked as detection abnormal points for 1 time, and 40 pixel points are marked as detection abnormal points for 2 times in 3 inspections in the current detection period, then n j,min = 1; if the detection area has 3 abnormal areas, each abnormal area contains 50, 100, and 150 pixel points, respectively, and the total number of marks of each pixel point in each abnormal area is 200, 300, and 450 times, respectively, then n If there are 100 pixel points in the jth abnormal area, the average abnormality possibility of 20 pixel points is 0.7, and the average abnormality possibility of 80 pixel points is 0.6 in 3 inspections in the current detection period, then

[0149] S3-4: Determine the pest and disease area in each abnormal area based on the time series data of the pest possibility index corresponding to each abnormal area.

[0150] In an embodiment of the present application, taking the jth abnormal area as an example, if the abnormal development of the jth abnormal area from the first occurrence of the jth abnormal area to the Tth period meets the following conditions: (1) the pest possibility index q j of the jth abnormal area continues to grow; (2) the pest possibility index growth rate is large and the time is long; (3) the area of the jth abnormal area expands to a large extent; then it is determined that the jth abnormal area meets the development trend of the larva of the monochromatic brown beetle, and the jth abnormal area is more likely to be a pest and disease area of the larva of the monochromatic brown beetle.

[0151] Specifically, taking the jth abnormal area appearing in the first detection period and ending in the third detection period as an example, based on the above determination conditions, the effective continuous difference sequence length of the time series data of the pest possibility index q j , the average value of the positive difference, and the area expansion ratio of the abnormal area are as follows: the time series data of the pest possibility index of the jth abnormal area from the first detection period to the third detection period is obtained, and it is assumed that the pest possibility indexes of the first detection period to the third detection period are 0.6, 0.7, and 0.8, respectively. Then the time series data of the pest possibility index q j{0.6, 0.7, 0.8 | i = 1, 2, 3}; the first-order difference of the pest possibility index time series data of the jth abnormal area from the first detection period to the third detection period is: from the first detection period to the second detection period, the pest possibility index increases from 0.6 to 0.7, the first-order difference is 0.1, from the second detection period to the third detection period, the pest possibility index increases from 0.7 to 0.8, the first-order difference is 0.1, and the effective continuous difference sequence length of the above pest possibility index time series data is 2; the above first-order difference with a positive value includes the first-order difference from the first detection period to the second detection period and from the second detection period to the third detection period, and both are 0.1, and the mean of the above difference positive value is (0.1+0.1) / 2 = 0.1; assuming that the area of the jth abnormal area initially appearing is 200 pixel points, and the area as of the third detection period is 350 pixel points, the above area expansion ratio is 350 / 200 = 1.75.

[0152] In the embodiments of the present application, the pest possibility index q j of the abnormal area is determined, and the effective continuous difference sequence length, the mean of the difference positive value, and the area expansion ratio of the abnormal area are determined.

[0153]

[0154] wherein w j,T represents the index of the jth abnormal area in the Tth detection period conforming to the disease and pest area characteristics of the Monochamus alternatus larva (i.e., the disease and pest area characteristic index), n j represents the effective continuous difference sequence length of the pest possibility index time series q j of the jth abnormal area (i.e., the effective continuous difference sequence length), n T represents the number of periods as of the current detection period, represents the mean of all difference values greater than 0 (i.e., the mean of the difference positive value); The greater the value of w j , the longer the time during which the pest possibility index q j of the jth abnormal area continues to increase; s j,T represents the area (number of pixel points) of the jth abnormal area in the Tth detection period (i.e., the second area area), s j,1 represents the area of the jth abnormal area when it initially appears (i.e., the first area area), The greater the value of s j,T , the greater the expansion degree of the jth abnormal area.

[0155] Specifically, the effective continuous difference sequence length n j is 2, the average of the positive difference values is 0.1, the area expansion ratio is 1.75, for example, the period number n T corresponding to the current detection period from the first detection period to the third detection period j,T = 3, then the pest area feature index w j,T of the jth abnormal area = (2*0.1) / 3*1.75 ≈ 0.12.

[0156] After obtaining the pest area feature index w j,T of the abnormal area, the pest area feature index can be normalized, and whether the abnormal area is a pest area is determined based on the normalized result. Specifically, taking the jth abnormal area as an example, after calculating the pest area feature index w j,T of the jth abnormal area, w j,T can be normalized to (0, 1), and if the normalized result is greater than 0.5 (where 0.5 is the value of the fifth preset threshold), the jth abnormal area is determined to be a pest area of the monochromous longicorn beetle larvae.

[0157] In step S150, a warning index of each pest area is generated, and warning information of each pest area is output based on the warning index; wherein the warning index is used to indicate the pest spread degree of the pest area.

[0158] In the embodiments of the present application, the warning index is used to indicate the pest spread degree of the pest area of the monochromous longicorn beetle larvae, and the warning information of the pest area of the monochromous longicorn beetle larvae can be set according to the warning index. For example, if the growth rate of the degree of the pest area meeting the characteristics of the monochromous longicorn beetle larvae is faster, the degree of the tree in the pest area being eroded is greater and faster, and the pest area needs to be treated more.

[0159] In the embodiments of the present application, the warning information includes the area position of the pest area, the pest spread path and the control suggestion. Taking a pest area of the monochromous longicorn beetle larvae in the Tth period as an example, the pest spread path can be obtained by the following method: marking the coordinate center point of the pest area of the monochromous longicorn beetle larvae in the Tth period, and the spread path is the line connecting the coordinate center point of the T-1th period to the coordinate center point of the Tth period; repeating the above steps to obtain the historical spread path of the pest area.

[0160] Exemplarily, the generation of the early warning indexes of each pest and disease area can be implemented as follows: determining a first pest and disease area characteristic index corresponding to the initial appearance of the pest and disease area, a second pest and disease area characteristic index corresponding to the target detection period of the pest and disease area, and a duration between the initial appearance of the pest and disease area and the target detection period; determining the early warning index of each pest and disease area based on the first pest and disease area characteristic index, the second pest and disease area characteristic index and the duration. Specifically, if the jthabnormal area is determined to be a pest and disease area, taking the jthpest and disease area as an example, the early warning index can be calculated by the following formula:

[0161]

[0162] wherein p j,T represents the early warning index required by the jthpest and disease area in the Tthdetection period (i.e., the target detection period), w j,1 represents the wcorresponding to the initial appearance of the jthpest and disease area (i.e., the first pest and disease area characteristic index), w j,T represents the wcorresponding to the jthpest and disease area in the Tthdetection period (i.e., the second pest and disease area characteristic index); t j represents the duration from the initial appearance of the jthpest and disease area to the Tthperiod (i.e., the duration).

[0163] In the embodiments of the present application, after determining the early warning indexes of each pest and disease area, the output of the early warning information of each pest and disease area based on the early warning indexes can be implemented as follows: in the current detection period, the early warning information of each pest and disease area is output in the order from large to small according to the early warning indexes, wherein the early warning information includes the area position, the pest and disease transmission path and the control suggestion of the corresponding pest and disease area. Specifically, the early warning indexes calculated above can be normalized to (0, 1), and the position information, the transmission path and the control suggestion of the corresponding pest and disease area are output in the order from large to small according to the early warning indexes in the current period. j,T

[0164] The present application constructs the reference waveband curve of the pine tree by constructing the hyperspectral image of the control area obtained by multiple inspections of the to-be-measured forest area, and preliminarily identifies the abnormal area in the detection area by comparing the hyperspectral curve of each pixel point in the detection area hyperspectral image with the reference waveband curve, so that the pest and disease area can be determined according to the change trend of the pest and disease characteristics of each abnormal area with time, the misjudgment of healthy trees can be screened out, and the accuracy of the pine beetle larva disease tree identification is improved. In addition, the present application can output the early warning information of each pest and disease area based on the early warning index, so that the staff can prevent and control the pest and disease area according to the harm degree, which is conducive to improving the prevention and control effect.​

[0165] The above mainly introduces the scheme provided by the embodiments of the application from the perspective of method. In order to realize the above functions, it contains the hardware structure and / or software module corresponding to the execution of each function. Those skilled in the art should easily realize that, in combination with the units and algorithm steps of each example described in the embodiments disclosed in the present text, the application can be realized in the form of hardware or the combination of hardware and computer software. Whether a certain function is executed in the form of hardware or computer software driven hardware depends on the specific application and design constraints of the technical solution. The skilled person can use different methods to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the application.

[0166] It should be noted that: the above sequence of the embodiments of the application is only for description, and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are also possible or can be advantageous.

[0167] Each of the embodiments in the specification is described in a progressive manner, and the same or similar parts between each embodiment can be referred to each other. Each embodiment mainly describes the difference from other embodiments.

Claims

1. An online detection and prevention method applied to forestry diseases and pests, characterized in that, The method comprises: multiple inspections are performed on the to-be-inspected forest area according to the same inspection rule, and hyperspectral images of a detection area and a control area in the to-be-inspected forest area are obtained; the control area is an area in the to-be-inspected forest area that only contains Pinus species and has been subjected to prevention and control of Monochamus alternatus larva, and the detection area is all other areas in the to-be-inspected forest area except the control area; a reference waveband curve of the Pinus species is constructed based on the hyperspectral images of the control area; hyperspectral curves of each pixel point in the hyperspectral images of the detection area are compared with the reference waveband curve, and an abnormal area in the detection area is identified; a pest infestation possibility index of the Monochamus alternatus larva in each abnormal area is calculated, and a pest infestation area in each abnormal area is determined based on time sequence data of the pest infestation possibility index corresponding to each abnormal area; a warning index of each pest infestation area is generated, and warning information of each pest infestation area is output based on the warning index; the warning index is used to indicate the degree of pest infestation spread in the pest infestation area.

2. The online detection and prevention method for forestry diseases and pests according to claim 1, characterized in that, The multiple inspections are performed on the to-be-inspected forest area according to the same inspection rule, and the hyperspectral images of the detection area and the control area in the to-be-inspected forest area are obtained, comprising: a detection period is determined; in each detection period, the detection area is inspected multiple times at fixed time points, and the hyperspectral images of the detection area are obtained; the hyperspectral images of the detection area obtained at adjacent time points in a single inspection process have an overlapping area; in each detection period, the control area is inspected multiple times at the fixed time points, and the hyperspectral images of the control area are obtained; the hyperspectral images of the control area obtained at adjacent time points in a single inspection process have an overlapping area.

3. The online detection and prevention method for forestry diseases and pests according to claim 2, characterized in that, After the hyperspectral images of the detection area and the control area in the to-be-inspected forest area are obtained, the method further comprises: for each hyperspectral image of the detection area and the control area obtained in a single inspection, a tree area in the hyperspectral image is marked based on a semantic recognition algorithm; for each gray scale image of each single waveband in each hyperspectral image, an edge point of the tree area is obtained based on an edge detection algorithm; a feature point in the edge point is determined based on the curvature and the gray scale gradient amplitude of the edge point; the hyperspectral images of the detection area obtained at different time points in a single inspection are spliced based on the feature point, to obtain a detection area complete image of the detection area and a control area complete image of the control area.

4. The online detection and prevention method for forestry diseases and pests according to claim 3, characterized in that, The feature point in the edge point is determined based on the curvature and the gray scale gradient amplitude of the edge point, comprising: the curvature of each edge point, the gray scale gradient amplitude of each edge point, and the mean value of the gray scale gradient amplitude of all edge points are calculated; a recognition degree index of each edge point is determined based on the curvature, the gray scale gradient amplitude of each edge point, and the mean value of the gray scale gradient amplitude of all edge points, and the edge point with the recognition degree index greater than a first preset threshold is marked as a recognizable point. determining a first frequency of each of the edge points being marked as the identifiable point in the gray-scale image of each of the bands of the hyperspectral image, and a second frequency of all the edge points being marked as the identifiable point in the gray-scale image of each of the bands of the hyperspectral image, all the edge points corresponding to a second frequency; determining a ratio of the first frequency and the second frequency, each of the edge points corresponding to a ratio, and marking the edge point corresponding to the first frequency as a feature point when the ratio of the first frequency and the second frequency is greater than a second preset threshold.

5. The online detection and prevention method for forestry diseases and pests according to claim 3, characterized in that, The constructing the reference band curve of the Pinus species based on the hyperspectral image of the control area comprises: calculating, for the complete image of the control area, a reflectivity extreme value ratio of each of the pixel points in the tree area in each band, and a slope extreme value ratio of a hyperspectral curve of all the pixel points, each of the pixel points in the complete image of the control area corresponding to a slope extreme value ratio in a band; determining a band consistency index of each band based on the reflectivity extreme value ratio and the slope extreme value ratio, each of the pixel points in the tree area corresponding to a band consistency index in each band; marking the band whose band consistency index is greater than a third preset threshold, and determining a band that is marked in each inspection as a similar band; determining a mean reflectivity of the pixel points in all the tree areas in the control area in the similar band as a normal reflectivity reference value of the similar band; arranging the normal reflectivity reference values of each of the similar bands in a wavelength order to draw a wavelength-reflectivity curve, thereby obtaining the reference band curve.

6. The online detection and prevention method for forestry diseases and pests according to claim 5, characterized in that, The comparing the hyperspectral curve of each of the pixel points in the hyperspectral image of the detection area with the reference band curve to identify an abnormal area in the detection area comprises: determining a detection sensitive band; calculating, for the complete image of the detection area determined in a single inspection, a first tangent slope and a first reflectivity of each of the pixel points in the tree area in the complete image of the detection area in the detection sensitive band of the corresponding hyperspectral curve; obtaining a second tangent slope and a second reflectivity of the detection sensitive band in the reference band curve; determining an abnormal possibility of the corresponding pixel point based on the first tangent slope, the first reflectivity, the second tangent slope and the second reflectivity; when the abnormal possibility is greater than a fourth preset threshold, marking the corresponding pixel point as a detection abnormal point, and determining the abnormal area based on the detection abnormal point.

7. The online detection and prevention method for forestry diseases and pests according to claim 6, characterized in that, The calculating a pest possibility index of the pine beetle larva in each of the abnormal areas, and determining a pest area in each of the abnormal areas based on time sequence data of the pest possibility index corresponding to each of the abnormal areas, comprises: For each detection cycle, the minimum number of times each pixel point in each abnormal region is marked as the detection abnormal point, the first average value of the abnormal possibility of all pixel points in each abnormal region in multiple inspections, and the second average value of the number of times the pixel points in all abnormal regions are marked as the detection abnormal point are obtained; Based on the minimum number, the first average value and the second average value, the pest possibility index of each abnormal region existing the pine bark beetle larvae is determined; The time sequence data of the pest possibility index is obtained, the effective continuous difference sequence length, the average value of the difference positive value and the area expansion ratio of the abnormal region are determined; Based on the effective continuous difference sequence length, the average value of the difference positive value and the area expansion ratio, the pest region feature index corresponding to the abnormal region is determined; When the pest region feature index is greater than a fifth preset threshold, the corresponding abnormal region is determined as the pest region.

8. The online detection and prevention method for forestry diseases and pests according to claim 7, characterized in that, The determination of the effective continuous difference sequence length, the average value of the difference positive value and the area expansion ratio of the abnormal region includes: First-order difference is performed on the time sequence data of the pest possibility index, and values greater than 0 in the first-order difference result are marked. If there is a continuous difference marked, the number of continuous marks is determined as the effective continuous difference sequence length; The average value of the first-order difference result with a positive result is calculated to obtain the average value of the difference positive value; The first area corresponding to the detection cycle when the abnormal region first appears is obtained, and the second area corresponding to the target detection cycle is obtained, and the area expansion ratio is determined based on the first area and the second area. 9.The online detection and prevention method for forestry diseases and pests according to claim 7, characterized in that, The generation of the early warning index of each pest region includes: The first pest region feature index corresponding to the initial appearance of the pest region, the second pest region feature index corresponding to the target detection cycle of the pest region, and the continuous time length from the initial appearance of the pest region to the target detection cycle are determined; Based on the first pest region feature index, the second pest region feature index and the continuous time length, the early warning index of each pest region is determined.

10. The online detection and prevention method for forestry diseases and pests according to claim 9, characterized in that, The output of the early warning information of each pest region based on the early warning index includes: In the detection cycle, the early warning information of each pest region is output in the order from large to small according to the early warning index. The early warning information includes the region position, pest transmission path and control suggestion of the corresponding pest region.

Images