Scale parameter optimization method for remote sensing detection of phyllostachys pubescens forest phyllostachys pubescens

Abstract

The invention provides a phyllostachys pubescens forest phyllostachys pubescens moth hazard remote sensing detection scale parameter optimization method, which comprises the following steps: (1) resampling hyperspectral data to a multi-source satellite sensor spectral scale through a Gaussian spectral response function, and analyzing and quantifying the response stability of spectral indexes to insect pests under different spectral scales by adopting a variable coefficient, screening spectral indexes with sensitive response and high stability; (2) based on the remote sensing images with different spatial resolutions, extracting moso bamboo forest and insect pest features by utilizing a machine learning algorithm and combining a recursive feature elimination technology, and determining an optimal spatial resolution through precision evaluation; (3) constructing an insect pest time sequence set, analyzing an insect pest grade conversion path through dynamic attitude analysis and a hazard transfer matrix, and determining a suitable monitoring period; and (4) synthesizing the spectral scale, the spatial resolution and the time period parameters output in the steps (1)-(3) to generate an optimized scale combination for remote sensing detection of the harm of the phyllostachys nigra.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of forest pest remote sensing monitoring, and particularly relates to a scale parameter optimization method for remote sensing detection of Phryganidia californica harm in Phyllostachys edulis forests. BACKGROUND

[0002] Phyllostachys edulis forests, as important ecological and economic resources in subtropical regions, play a key role in carbon sequestration and emission reduction and water and soil conservation, but are prone to pest infestation, which leads to a decrease in shoot emergence rate, a weakening of growth momentum, a brittle bamboo wall, and poor material quality, greatly affecting ecological and economic values. Phryganidia californica is an important leaf-eating pest of Phyllostachys edulis, which has the characteristics of mass occurrence, periodicity, and extremely serious harm, and has a destructive power comparable to "smokeless forest fire". Traditional manual ground investigation is low in efficiency and limited in coverage, and is difficult to meet the needs of large-scale dynamic monitoring, while existing remote sensing technologies have improved pest identification capability through spectral feature analysis, multi-feature fusion, and "sky-ground" integrated monitoring system, but the precision improvement is restricted due to insufficient scale effect research. Specifically, there is blindness in spectral scale selection, different sensor band configurations lead to unstable vegetation index response, and wide-band data easily masks subtle changes in pests; the contradiction between spatial resolution and precision is prominent, high-resolution images are disturbed by noise and have a large processing load, and low-resolution data cause mixed pixel problems; there is insufficient time scale adaptability, and it is difficult to balance short-term disturbance and long-term trend with fixed observation period; multi-scale coupling analysis is missing, existing methods optimize a single dimension in isolation, and do not solve the problem of the collaborative effect of spectrum, space, and time. In addition, existing technologies focus on single sensor or algorithm improvement, and do not break through the bottleneck of multi-scale parameter collaborative optimization, resulting in poor model universality and insufficient quantification of pest grade conversion rules. SUMMARY

[0003] In view of the defects and deficiencies of the prior art, the application provides a scale parameter optimization method and system for remote sensing detection of Phryganidia californica harm in Phyllostachys edulis forests. The existing pest monitoring technology lacks coupling of spectral, spatial, and temporal scale effects, leading to problems such as unstable response caused by sensor band differences, mixed pixels and noise interference, and difficulty in balancing short-term disturbance and long-term trend with fixed observation period. Therefore, the application first creates a time-space-spectrum three-dimensional collaborative optimization mechanism:

[0004] At the spectral scale level, the hyperspectral data is resampled to the multisource satellite sensor scale using the Gaussian spectral response function, breaking through the spectral drift problem caused by the difference in sensor wavebands. The sensitivity of the insect damage is quantified through coefficient of variation analysis: the maximum spectral deviation is calculated based on the extreme value deviation of the spectral index under different damage levels, and the stability is evaluated through the scale effect variation rate: the spectral difference of each sensor and the reference scale is compared to generate a standardized variation parameter, and the sensitive waveband is selected by combining the single factor variance analysis method. This technical path significantly improves the spectral response stability in the embodiment, and in the test and verification of the preferred embodiment, the Sentinel-2 sensor has the lowest variation degree, becoming the optimal spectral scale selection.

[0005] Based on the output of spectral scale optimization, a hierarchical feature extraction architecture is constructed at the spatial scale level: first, the gray level co-occurrence matrix texture feature and the random forest algorithm are fused, and the bamboo forest response features are dynamically selected through recursive feature elimination (RFE); on this basis, the random forest, support vector machine or extreme gradient boosting algorithm is used to extract the insect damage information, and RFE is used for secondary optimization of the insect damage features. After accuracy verification (overall accuracy ≥ 90.45%, Kappa coefficient ≥ 0.921), the optimal solution of noise suppression and mixed pixel balance is realized at the spatial resolution of 10 meters, and the boundary of the insect damage information is clear and the ground object interference is minimal.

[0006] Further, a dynamic analysis mechanism is introduced at the time scale level: a time series set is constructed with 15 days as the minimum analysis period, the change rate of a specific insect damage level is quantified through a single dynamic, the overall change trend of the system is analyzed comprehensively, and the level conversion path is analyzed combined with the hazard transfer matrix. The 30-45 day period transfer matrix clearly shows the typical conversion path of "healthy→mild damage→severe damage", effectively balancing short-term disturbance and long-term trend capture.

[0007] Finally, the present scheme generates a time-space-spectrum coupled scale combination through a three-dimensional cooperative output module, integrating the optimal parameters of spectrum (such as multispectral sensor), space (sub-ten-meter resolution), and time (30-45 day period). The optimal combination is used to directly guide the configuration of the remote sensing detection model, and in the embodiment, the insect damage recognition accuracy is significantly improved. The supporting hardware system executes the above optimization process through the processor, providing a standardized and reusable parameter configuration scheme for large-scale insect damage monitoring of bamboo forests.

[0008] The technical scheme specifically adopted by the present application to solve its technical problems is:

[0009] A scale parameter optimization method for detecting the harm of the bamboo borer to the bamboo forest, comprising:

[0010] (1) Resample the hyperspectral data to the multisource satellite sensor spectral scale by the Gaussian spectral response function, use the coefficient of variation analysis to quantify the response stability of the spectral index to the insect pest under different spectral scales, and screen the spectral index with high response sensitivity and stability;

[0011] (2) Based on remote sensing images with different spatial resolutions, use machine learning algorithms combined with recursive feature elimination technology to extract Phyllostachys edulis forest and insect pest features, and determine the optimal spatial resolution through accuracy evaluation;

[0012] (3) Construct a time series set of insect pests, analyze the grade conversion path of insect pests through dynamic degree analysis and hazard transfer matrix, and determine the appropriate monitoring period;

[0013] (4) Combine the spectral scale, spatial resolution and time period parameters output by steps (1)-(3) to generate the optimal scale combination for detecting the damage of the Phyllostachys edulis moth.

[0014] Further, the coefficient of variation analysis comprises:

[0015] Based on the maximum and minimum values of the spectral index under different insect pest grades, calculate the maximum spectral index deviation rate caused by insect pests;

[0016] The sensitivity of the spectral index to the insect pest is quantitatively described by the deviation rate.

[0017] Further, in step (1), the single factor variance analysis method is used to screen the sensitive spectral band.

[0018] Further, the response stability is evaluated by the scale effect variation rate, specifically comprising:

[0019] For the same insect pest grade, calculate the maximum difference value of the spectral index under each spectral scale and the reference scale spectral index;

[0020] Divide the maximum difference value by the maximum value of the reference scale spectral index to generate the scale effect variation rate.

[0021] Further, in step (2), the extraction of Phyllostachys edulis forest information comprises:

[0022] Use the gray level co-occurrence matrix to extract texture features;

[0023] Combine the random forest algorithm to establish a detection model;

[0024] Screen the response features by the recursive feature elimination method.

[0025] Further, in step (2), the extraction of insect pest information comprises:

[0026] Use random forest, support vector machine or extreme gradient boosting algorithm to establish an insect pest detection model;

[0027] Screening pest response features based on recursive feature elimination algorithm.

[0028] Further, the precision evaluation includes calculating overall precision, Kappa coefficient and F1 score.

[0029] Further, the dynamic degree analysis includes:

[0030] Single dynamic degree: calculating the rate of change of a specific pest grade;

[0031] Comprehensive dynamic degree: calculating the rate of change of the overall system.

[0032] Further, the hazard transition matrix is an n*n matrix P representing the transition proportion of pest grade from i to j ij .

[0033] And a scale parameter optimization system for detecting the pine caterpillar hazard of Phyllosticta bambusicola in a bamboo forest by remote sensing, comprising a memory, a processor and a computer program stored in the memory, characterized in that when the processor executes the program, the method steps as described above are realized.

[0034] Compared with the prior art, the present application and its preferred schemes at least include the following beneficial effects:

[0035] 1. Breakthrough of the inherent bottleneck of scale effect coupling missing

[0036] Through the time-space-spectrum three-dimensional cooperative optimization mechanism, the linkage analysis of spectral, spatial and temporal scales is realized for the first time. On the spectral scale, Gaussian resampling combined with variation coefficient quantification technology significantly improves the response consistency of multi-source sensor data, overcoming the spectral drift problem caused by differences in waveband configuration in traditional methods; on the spatial scale, hierarchical feature screening driven by recursive feature elimination (RFE) effectively balances the contradiction between high-resolution noise and low-resolution mixed pixels; on the time scale, dynamic degree matrix and grade conversion path analysis solve the problem that fixed monitoring period is difficult to adapt to pest dynamic changes. This multi-scale coupling architecture fills the gap of isolated optimization in the prior art and provides a universal framework for pest monitoring in complex ecological scenarios.

[0037] 2. Synergistic improvement of spectral response stability and sensitivity

[0038] Based on spectral scale effect variation rate analysis, the sensor difference quantification evaluation and pest sensitivity screening are innovatively combined. By comparing the spectral deviation of different sensor data and reference scale, a standardized stability parameter is generated, and sensitive bands are screened by combining single factor variance analysis.

[0039] 3. Spatial feature step-by-step optimization improves detection accuracy

[0040] A hierarchical extraction strategy of "Phyllostachys edulis forest information → pest information" is adopted: first, the texture features of Phyllostachys edulis forest are extracted by fusing the gray level co-occurrence matrix and the random forest, and then the pest response features are screened based on recursive feature elimination (RFE). This step-by-step architecture not only retains spatial details, but also suppresses redundant noise, so that the boundary recognition accuracy of pest detection at sub-meter resolution is significantly optimized.

[0041] 4. Visual analysis of dynamic evolution law of pests

[0042] The innovation application hazard transfer matrix analyzes the grade conversion path, captures the change rate of a specific pest grade through a single dynamic degree, and quantifies the overall evolution trend of the system. This mechanism first realizes the visualization tracking of key paths such as "healthy → mild hazard → severe hazard", providing quantitative basis for pest outbreak warning.

[0043] 5. Standardized output of parameter combination and system integration

[0044] The generated optimal scale combination of time, space and spectrum (such as multispectral sensor + sub-meter resolution + 30-45 day cycle) forms a standardized configuration scheme, which can be directly migrated to different regional pest monitoring scenarios. The supporting hardware system realizes the closed-loop support from parameter optimization to detection landing through the integration of spectral resampling, spatial feature optimization and time dynamic analysis module.

[0045] The present application breaks through the limitation of multi-scale fragmentation optimization from the technical principle, improves the response robustness through the spectral quantization mechanism, guarantees the feature effectiveness through the hierarchical spatial strategy, enhances the trend prediction ability through the dynamic time analysis, and finally promotes the pest monitoring from "experience-driven" to "data-driven" with the standardized parameter combination output. BRIEF DESCRIPTION OF DRAWINGS

[0046] The present application will be further described in detail below in combination with the drawings and specific embodiments:

[0047] Figure 1 is a method implementation flowchart of an embodiment of the present application.

[0048] Figure 2 is a regional remote sensing image of an embodiment of the present application.

[0049] Figure 3 is a variation degree graph of spectral index under each spectral scale of an embodiment of the present application.

[0050] Figure 4 is a Phyllostachys edulis forest Neotoxopteryx regularis pest hazard information graph under different spatial scales of an embodiment of the present application.

[0051] Figure 5 is a Phyllostachys edulis forest Neotoxopteryx regularis pest hazard level transfer matrix graph of each time scale of an embodiment of the present application.

[0052] Figure 6 Figure 1 is a remote sensing detection model of bamboo forest C. pyralis damage according to an embodiment of the present application. DETAILED DESCRIPTION

[0053] In order to make the features and advantages of the present application more apparent, the following embodiments are specifically described, and the details are described as follows:

[0054] It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the present application. Unless otherwise indicated, all technical and scientific terms used in the present description have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0055] It should be noted that the terms used herein are only for the purpose of describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, the singular form is intended to include the plural form unless the context clearly indicates otherwise, and furthermore, it should be understood that when the terms "comprise" and / or "include" are used in the present description, they indicate the presence of a feature, step, operation, device, component, and / or combination thereof.

[0056] The present application provides a scale parameter optimization method for remote sensing detection of bamboo forest C. pyralis damage, which provides a solid technical foundation for efficient and accurate pest detection. The implementation process includes the following steps:

[0057] (1) Spectral scale effect analysis: The satellite hyperspectral remote sensing image is simulated into spectral data of different satellite sensors using the Gaussian spectral response function, and the response ability and stability of spectral index to C. pyralis damage under different spectral scales are analyzed;

[0058] (2) Spatial scale effect analysis: Based on the spectral scale effect analysis, the bamboo forest information and C. pyralis damage information are extracted using machine learning algorithms based on remote sensing images of different spatial resolutions, and the response features are screened and the information detection model is established using the recursive feature elimination algorithm, and the model is evaluated for accuracy;

[0059] (3) Time scale effect analysis: Considering the spectral and spatial scale effects of remote sensing detection of bamboo forest C. pyralis damage, a time series set of C. pyralis damage is constructed, and the trend of pest information under different time scales is analyzed;

[0060] (4) Model construction: A remote sensing detection model of bamboo forest C. pyralis damage based on time-space-spectral scale effect analysis is constructed, and the appropriate combination of time, space, and spectral scale for detecting C. pyralis damage is determined.

[0061] As a preferred scheme of the embodiment, in step (1), the hyperspectral remote sensing image is simulated into spectral data of different satellite sensors by using a Gaussian spectral response function, and the response capability and stability of the spectral index to the pine caterpillar damage at different spectral scales are analyzed. The characteristic band characteristic spectral index, difference vegetation index, and enhanced vegetation index are selected, the response capability of the spectral index to the insect damage is analyzed by using single factor variance analysis combined with the least significant difference method, the sensitivity of the spectral index to the pine caterpillar damage is quantitatively described by analyzing the variation coefficient of the spectral index with the change of the insect damage grade, and the calculation formula is as follows:

[0062]

[0063] In the formula, SI represents the spectral index, SI max and SI min respectively represent the spectral index under different insect damage degrees.

[0064] Similarly, the spectral scale effect of the spectral index is quantitatively described by using the variation degree of the spectral index at different spectral scales under the pine caterpillar damage, and the formula is as follows:

[0065]

[0066] In the formula, i represents the corresponding spectral scale; j is the degree of the pine caterpillar damage; SI i,j represents the spectral index of each satellite at the corresponding spectral scale under different degrees of the pine caterpillar damage, SI ZY,j represents the spectral index of the satellite at the corresponding spectral scale under different degrees of the insect damage.

[0067] The stability of the response capability of the spectral index to the pine caterpillar damage at different bands is analyzed.

[0068] As a preferred scheme of the embodiment, in step (2), the spectral scale effect of the pine caterpillar damage in the bamboo forest is considered, the bamboo forest information and the pine caterpillar damage information are extracted by using a machine learning algorithm based on remote sensing images with different spatial resolutions, the response features are screened by using a recursive feature elimination algorithm, the information detection model is established, the model is subjected to accuracy evaluation, and the appropriate spatial resolution is determined based on the spatial scale effect analysis. The steps are as follows:

[0069] a) The bamboo forest information is extracted by using a gray level co-occurrence matrix and a random forest algorithm, the response features of the bamboo forest information are screened based on a recursive feature elimination method, the bamboo forest information detection model is established and subjected to accuracy evaluation;

[0070] b) On the basis of the Phyllostachys edulis forest information, random forest, support vector machine and extreme gradient boosting algorithm are used to extract the Orgyia postica harm information respectively, recursive feature elimination algorithm is used for harm information response feature screening, an Orgyia postica harm information detection model is established and precision evaluation is carried out;

[0071] c) Based on the spatial scale effect analysis of the Phyllostachys edulis forest information and the Orgyia postica harm information under different spatial scales, the suitable spatial resolution for Orgyia postica harm detection is determined.

[0072] As a preferred scheme of the embodiment, in step (3), the spectral and spatial scale effects of the Orgyia postica harm detection of the Phyllostachys edulis forest are comprehensively considered, a time series set of the Orgyia postica harm is constructed, and the Orgyia postica harm dynamic degree and the Orgyia postica harm transfer matrix method are used to analyze the change trend of the Orgyia postica harm information under different time scales, find out the transition path between each pest grade, and determine the suitable time scale.

[0073] Among them, the dynamic degree analysis is used to measure the overall change rate of all pest grades (comprehensive dynamic degree) and the change rate of a specific pest grade (single dynamic degree) in a certain time scale of the whole Phyllostachys edulis forest system in a certain time scale; and the Orgyia postica harm transfer matrix is used to describe the change of the stress level of the Phyllostachys edulis forest under the harm of the Orgyia postica under different time scales, and clearly show the trend and intensity of the dynamic change of the pest at each time scale. The calculation formula is as follows:

[0074] Single dynamic degree:

[0075] Comprehensive dynamic degree:

[0076] Harm transfer matrix:

[0077] As a preferred scheme of the embodiment, in step (4), a Phyllostachys edulis forest Orgyia postica harm remote sensing detection model based on time-space-spectral scale effect analysis is constructed, and the suitable time, space and spectral scale combination is determined.

[0078] The application innovatively comprehensively analyzes the scale effect of the Orgyia postica harm remote sensing detection of the Phyllostachys edulis forest from the spectrum, space and time three scales, provides a new perspective and method for accurately extracting pest information, clearly determines the suitable spectral, spatial and temporal scales for the Orgyia postica remote sensing detection of the Phyllostachys edulis forest, and provides a key basis for optimizing the detection scheme; the field measurement data and the hyperspectral remote sensing image are combined to construct a remote sensing detection model, and the comprehensive ability of remote sensing detection is improved.

[0079] The above schemes of the application are more specifically demonstrated and introduced through a specific application example as follows:

[0080] For example,Figure 1 As shown, the scale parameter optimization method for detecting Phryganidia californica harm in Phyllostachys edulis forest adopted in the embodiment includes the following steps:

[0081] (1) Spectral scale effect analysis: The satellite hyperspectral remote sensing image is simulated into spectral data of different satellite sensors by using the Gaussian spectral response function, and the response ability and stability of spectral index to Phryganidia californica harm under different spectral scales are analyzed;

[0082] (2) Spatial scale effect analysis: On the basis of spectral scale effect analysis, based on remote sensing images of different spatial resolutions, machine learning algorithm is used to extract Phyllostachys edulis forest information and Phryganidia californica harm information, and recursive feature elimination algorithm is used to screen response features and establish information detection model, and the model is evaluated in accuracy;

[0083] (3) Time scale effect analysis: The spectral and spatial scale effects of remote sensing detection of Phryganidia californica harm in Phyllostachys edulis forest are comprehensively considered, a time series set of Phryganidia californica harm is constructed, and the change trend of pest information under different time scales is analyzed;

[0084] (4) Model construction: A remote sensing detection model of Phryganidia californica harm in Phyllostachys edulis forest based on time-space-spectral scale effect analysis is constructed, and the appropriate combination of time, space and spectral scale for detecting Phryganidia californica harm is determined.

[0085] The following takes the embodiment area of the application as shown in the embodiment of the application as an example to further illustrate the related contents involved in the method. Figure 2

[0086] (1) Spectral scale effect analysis

[0087] Based on the pest points obtained by field investigation, the hyperspectral data of the pest points corresponding to the Phyllostachys edulis forest are extracted; the ZY-1 02D AHSI hyperspectral remote sensing data is resampled by using the Gaussian spectral response function, and is simulated into the corresponding spectral resolution of Sentinel-2 MSI, Landsat8 OLI and GF-1 PMS1; according to the change rule of spectral reflectance of Phyllostachys edulis forest under different Phryganidia californica harm levels, it is found that the spectral reflectance of Phyllostachys edulis forest under different pest levels has differences at characteristic wavebands such as 480, 560, 670 and 880 nm, and 15 spectral indices are selected according to the characteristic wavebands; the 15 selected spectral indices are calculated based on the simulated Phyllostachys edulis forest spectral data of each satellite sensor obtained by resampling the hyperspectral data, and the Pearson correlation analysis method is used to analyze the correlation between the spectral indices under different spectral scales and the Phryganidia californica harm, and it is found that 13 spectral indices have significant correlation; the sensitivity of the spectral indices to the Phryganidia californica harm under different spectral scales is analyzed, and the variation degree of the spectral indices under different spectral scales is analyzed based on the spectral scale effect. Figure 3 ​The results show that most spectral indices at the corresponding scales of Sentinel-2 MSI and GF-1 PMS1 can sensitively distinguish different levels of pests. GF-1 PMS1 has a better ability to identify differences between pest levels, while the spectral indices at the corresponding scales of Sentinel-2 MSI have stronger stability.

[0088] (2) Spatial scale effect analysis

[0089] Based on the spectral scale effect analysis of the damage caused by the tussock moth in bamboo forests, and using remote sensing images at different spatial resolutions, random forest, support vector machine, and extreme gradient boosting algorithms were employed to extract information on bamboo forests and tussock moth damage at different spatial resolutions. Figure 4 By employing a recursive feature elimination method, spectral indices and texture features related to bamboo forest information and the damage caused by the bamboo tussock moth were screened. An information detection model was established and its accuracy was evaluated to determine the appropriate spatial resolution. Ultimately, a spatial resolution of 10m was found to be superior to other resolutions in extracting bamboo forest information, with an OA of 90.45%, a Kappa coefficient of 0.921, and an F1 score of 0.931 at this resolution.

[0090] (3) Time scale effect analysis

[0091] Taking into account the spectral and spatial scale effects of the detection of tussock moth damage in bamboo forests, a time series set of tussock moth damage was constructed to analyze the dynamic changes in pest information at different time scales. Specifically, this included calculating the comprehensive dynamic degree analysis and single dynamic degree analysis of tussock moth damage at T=1 (15-day interval), T=2 (30-day interval), T=3 (45-day interval), and T=4 (60-day interval) with a 15-day period; and constructing the horizontal transfer matrix of tussock moth damage at T=1, T=2, T=3, and T=4 respectively. Figure 5 To analyze the transition paths between different pest levels, it was ultimately determined that T=2 and T=3 are more suitable time scales, which can provide stable pest level transition information and reduce the impact of short-term disturbances.

[0092] (4) Construction of remote sensing detection model for damage caused by bamboo tussock moth

[0093] Constructing a remote sensing detection model for the damage caused by the tussock moth in moso bamboo forests based on spatiotemporal-spectral scale effect analysis ( Figure 6 The appropriate combination of temporal, spatial, and spectral scales was determined. Ultimately, Sentinel-2 satellite imagery, a spatial resolution of 10m, and a period of 30 or 45 days were selected as the suitable temporal-spatial-spectral scales for detecting tussock moth damage in bamboo forests.

[0094] The embodiment also provides a system for detecting Phaedna robustella harm in Phyllostachys edulis forests based on time-space-spectrum scale effect analysis, comprising a memory, a processor and computer program instructions stored in the memory and capable of being executed by the processor, and when the processor executes the computer program instructions, the above method steps can be realized.

[0095] Based on the same inventive concept, the present application further provides a computer device, comprising one or more processors, and a memory for storing one or more computer programs; the program comprises program instructions, and the processor is configured to execute the program instructions stored in the memory. The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc., which are the computing core and control core of the terminal, and are configured to implement one or more instructions, and are specifically configured to load and execute one or more instructions in the computer storage medium to realize the above method.

[0096] It needs to be further explained that based on the same inventive concept, the present application further provides a computer storage medium, which stores a computer program, and the computer program is executed by the processor to execute the above method. The storage medium can adopt any combination of one or more computer readable media. The computer readable medium can be a computer readable signal medium or a computer readable storage medium. The computer readable storage medium can be, but is not limited to, an electrical, magnetic, optical, infrared or semiconductor system, device or component, or any combination of the above. More specific examples (non-exhaustive list) of the computer readable storage medium include: an electrical connection with one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, the computer readable storage medium can be any tangible medium containing or storing a program, which can be used by or in combination with an instruction execution system, device or component.

[0097] It should be noted that, unless otherwise defined, technical terms or scientific terms used in the present application shall have the common meaning understood by one of ordinary skill in the art to which the present application pertains. The terms "first", "second", and similar terms used in the present application do not denote any order, quantity, or importance, but are used to distinguish different components. The terms "comprise", "comprising", and similar terms mean that the elements or objects before the term encompass the elements or objects listed after the term and their equivalents, and do not exclude other elements or objects. The terms "connected" or "connected" and similar terms are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. The terms "up", "down", "left", "right", and the like are only used to represent relative positional relationships, and when the absolute positions of the described objects change, the relative positional relationships can also change accordingly.

[0098] The above description is only the preferred embodiments of the present application, and does not limit the other forms of the present application. Any person skilled in the art can modify or change the above-mentioned disclosed technology into equivalent embodiments with equivalent changes. However, any simple modification, equivalent change, and modification of the above-mentioned embodiments without departing from the technical solution of the present application, and according to the technical essence of the present application, still belongs to the protection scope of the present application.

[0099] The present application is not limited to the above-mentioned best mode, and anyone can derive other various forms of a scale parameter optimization method for remote sensing detection of Phyllosoma sinica harm in a bamboo forest under the inspiration of the present application. Any equivalent change and modification made within the scope of the present application shall be covered by the present application.

Claims

1. A scale parameter optimization method for detecting the harm of the bamboo forest by the bamboo shoot moth through remote sensing, characterized in that, The method comprises the following steps: (1) Resampling hyperspectral data to multisource satellite sensor spectral scale by Gaussian spectral response function, quantifying the response stability of spectral index to insect pests at different spectral scales by coefficient of variation analysis, and screening spectral index with sensitive response and high stability; (2) Based on remote sensing images with different spatial resolutions, extracting the characteristics of bamboo forests and insect pests by machine learning algorithm combined with recursive feature elimination technology, and determining the optimal spatial resolution through accuracy evaluation; (3) Constructing a time series set of insect pests, analyzing the grade conversion path of insect pests by dynamic degree analysis and hazard transfer matrix, and determining the appropriate monitoring period; (4) Generating the optimal scale combination of the detection of the damage of the moth to the bamboo by remote sensing by comprehensively outputting the spectral scale, spatial resolution and time period parameters output by steps (1)-(3).

2. The scale parameter optimization method for the detection of the damage of the moth to the bamboo in the bamboo forest according to claim 1, wherein the coefficient of variation analysis comprises: calculating the maximum spectral index deviation rate caused by insect pests based on the maximum value and the minimum value of the spectral index under different insect pest grades; quantitatively describing the sensitivity of the spectral index to insect pests through the deviation rate. In step (1), the single factor variance analysis method is used to screen the sensitive spectral band.

3. The scale parameter optimization method for detecting the damage of N. menoniis caused by the N. menonius on the bamboo forest by remote sensing according to claim 1, characterized in that:

4. The scale parameter optimization method for the detection of the damage of the moth to the bamboo in the bamboo forest according to claim 1, wherein the response stability is evaluated by the spectral scale effect variation rate, specifically comprising: for the same insect pest grade, calculating the maximum difference value of the spectral index at each spectral scale and the spectral index of the reference scale; dividing the maximum difference value by the maximum value of the spectral index of the reference scale to generate the scale effect variation rate.

5. The scale parameter optimization method for the detection of the damage of the moth to the bamboo in the bamboo forest according to claim 1, wherein in step (2), the extraction of bamboo forest information comprises: extracting texture features by using the gray level co-occurrence matrix; combining the random forest algorithm to establish a detection model; screening the response features by the recursive feature elimination method.

6. The scale parameter optimization method for the detection of the damage of the moth to the bamboo in the bamboo forest according to claim 1, wherein in step (2), the extraction of insect pest information comprises: establishing an insect pest detection model by using the random forest, support vector machine or extreme gradient boosting algorithm; screening the insect pest response features based on the recursive feature elimination algorithm. The accuracy evaluation comprises calculating the overall accuracy, Kappa coefficient and F1 score.

8. The scale parameter optimization method for the detection of the damage of the moth to the bamboo in the bamboo forest according to claim 1, wherein the dynamic degree analysis comprises: single dynamic degree: calculating the change rate of a specific insect pest grade; comprehensive dynamic degree: calculating the change rate of the overall system.

7. The scale parameter optimization method for remote sensing detection of Camptochromium herbivora hazard in a bamboo forest according to claim 1, characterized in that: The processor executes the program to realize the method steps of any one of claims 1-9. ​ ​ ​ ​ 9. The scale parameter optimization method for detecting the damage of N. menoniis caused by the N. menonius on the bamboo forest by remote sensing according to claim 1, characterized in that: The hazard transition matrix is an n x n matrix P representing the proportion of transition of the pest damage level from i to j ij .

10. A scale parameter optimization system for detecting the harm of Bupalus piniarius Linn. in Phyllostachys edulis forest by remote sensing, comprising a memory, a processor and a computer program stored in the memory, characterized in that, ​

Images