CN114119513B - Time-spectrum mixed early monitoring method for pine wood nematode disease - Google Patents
Time-spectrum mixed early monitoring method for pine wood nematode disease Download PDFInfo
- Publication number
- CN114119513B CN114119513B CN202111341729.1A CN202111341729A CN114119513B CN 114119513 B CN114119513 B CN 114119513B CN 202111341729 A CN202111341729 A CN 202111341729A CN 114119513 B CN114119513 B CN 114119513B
- Authority
- CN
- China
- Prior art keywords
- time
- time sequence
- vegetation index
- image
- predicted
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 201000010099 disease Diseases 0.000 title claims abstract description 55
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 title claims abstract description 55
- 241000243771 Bursaphelenchus xylophilus Species 0.000 title claims abstract description 49
- 238000012544 monitoring process Methods 0.000 title claims abstract description 41
- 238000000034 method Methods 0.000 title claims abstract description 30
- 238000001228 spectrum Methods 0.000 title claims abstract description 29
- 235000011609 Pinus massoniana Nutrition 0.000 claims abstract description 40
- 241000018650 Pinus massoniana Species 0.000 claims abstract description 40
- 238000009826 distribution Methods 0.000 claims abstract description 27
- 238000012549 training Methods 0.000 claims abstract description 15
- 235000008331 Pinus X rigitaeda Nutrition 0.000 claims abstract description 10
- 235000011613 Pinus brutia Nutrition 0.000 claims abstract description 10
- 241000018646 Pinus brutia Species 0.000 claims abstract description 10
- 238000007781 pre-processing Methods 0.000 claims abstract description 10
- 239000002775 capsule Substances 0.000 claims description 28
- 241000894007 species Species 0.000 claims description 12
- 238000004364 calculation method Methods 0.000 claims description 8
- 238000012937 correction Methods 0.000 claims description 8
- 241000238631 Hexapoda Species 0.000 claims description 7
- 238000001514 detection method Methods 0.000 claims description 7
- 238000011835 investigation Methods 0.000 claims description 7
- 238000011176 pooling Methods 0.000 claims description 7
- 239000013598 vector Substances 0.000 claims description 7
- 238000003062 neural network model Methods 0.000 claims description 6
- 238000012952 Resampling Methods 0.000 claims description 4
- 230000015572 biosynthetic process Effects 0.000 claims description 4
- 230000011218 segmentation Effects 0.000 claims description 4
- 238000003786 synthesis reaction Methods 0.000 claims description 4
- 238000013528 artificial neural network Methods 0.000 claims description 3
- 230000008569 process Effects 0.000 claims description 3
- 230000004913 activation Effects 0.000 claims description 2
- 238000006243 chemical reaction Methods 0.000 claims description 2
- 238000010606 normalization Methods 0.000 claims description 2
- 241000607479 Yersinia pestis Species 0.000 abstract description 10
- 230000008859 change Effects 0.000 abstract description 8
- 230000036541 health Effects 0.000 abstract description 3
- 241000244206 Nematoda Species 0.000 abstract 2
- 230000002265 prevention Effects 0.000 description 3
- 230000007480 spreading Effects 0.000 description 3
- 238000003892 spreading Methods 0.000 description 3
- 241000243770 Bursaphelenchus Species 0.000 description 2
- 241000196324 Embryophyta Species 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 230000001066 destructive effect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000002310 reflectometry Methods 0.000 description 2
- 229960002117 triamcinolone acetonide Drugs 0.000 description 2
- YNDXUCZADRHECN-JNQJZLCISA-N triamcinolone acetonide Chemical compound C1CC2=CC(=O)C=C[C@]2(C)[C@]2(F)[C@@H]1[C@@H]1C[C@H]3OC(C)(C)O[C@@]3(C(=O)CO)[C@@]1(C)C[C@@H]2O YNDXUCZADRHECN-JNQJZLCISA-N 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000011065 in-situ storage Methods 0.000 description 1
- 208000015181 infectious disease Diseases 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 230000024241 parasitism Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000000985 reflectance spectrum Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N2021/1793—Remote sensing
- G01N2021/1797—Remote sensing in landscape, e.g. crops
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10032—Satellite or aerial image; Remote sensing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30181—Earth observation
- G06T2207/30188—Vegetation; Agriculture
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Quality & Reliability (AREA)
- Image Processing (AREA)
Abstract
The time-spectrum mixed early monitoring method for pine wood nematode disease comprises the following steps of S1: acquiring a time sequence image and an image to be predicted under the monitoring of Sentinel-2 and preprocessing; s2: calculating a time sequence mean normalized vegetation index based on reflection values of different wave bands under a plurality of time phases, and extracting pinus massoniana distribution; s3: calculating a time sequence multi-type red edge vegetation index sequence and a multi-type red edge vegetation index sequence in a period to be predicted; s4: training a prediction model based on the time sequence multi-type red edge vegetation index sequence; s5: predicting a multi-type red-edge vegetation index sequence in a period to be predicted by using a prediction model, and obtaining a pine wood nematode pest prediction result by combining the distribution condition of masson pine; the multi-time-phase and multi-type red-edge vegetation indexes are introduced to reflect the health change condition of the masson pine, a prediction model is established, the spectrum change rule of the masson pine before and after the pine is suffered from the pine nematode disease is fully excavated, and the problem of early monitoring of the pine nematode disease by utilizing single-period satellite remote sensing is effectively solved.
Description
Technical Field
The invention relates to the technical field of remote sensing image processing, in particular to a time-spectrum mixed early monitoring method for pine wood nematode diseases.
Background
Pine wood nematode disease is a destructive disease which causes rapid death of trees due to pine wood nematode parasitism in pine tree bodies, and is the most dangerous and destructive disease in the global forest ecosystem at present. During the period of fourteen five, the tissues in China develop the action of preventing and controlling the pine wood nematode diseases for 5 years, and the rapid spreading and spreading potential of the pine wood nematode diseases is restrained. Traditional pine wood nematode disease monitoring is generally carried out by manually investigating affected pine on site, and the mode wastes a large amount of manpower and material resources and has low efficiency, especially in mountain areas of mountain high-rise danger, epidemic situation occurrence dynamics can not be timely and comprehensively mastered, and epidemic situation spreading is difficult to effectively control.
At present, domestic production units utilize satellite remote sensing images to monitor pine wood nematode diseases, and establish interpretation marks mainly according to the characteristics of tone, texture and the like shown on high-resolution images after insect damage occurs, kilometer grids are overlapped, and dead tree pattern spots are extracted from grids by a visual interpretation method. Compared with manual field screening, the method has the efficiency advantage, but when the color of the tree changes or even dies, the tree is deeper in disease, the visual interpretation method can only stop damage in time, and early prevention cannot be realized.
Research on monitoring pine wood nematode diseases by using satellite remote sensing images at home and abroad mainly focuses on establishing a prediction model by using single-time-phase remote sensing images and measured data, so that the occurrence condition and the distribution range of the pine wood nematode diseases in a certain period are monitored; or a field spectrometer is used for collecting the field reflectivity spectrum data, and a laboratory analyzes the spectrum change rule so as to judge whether the disease occurs. However, these methods only can respond to the pest situation in the data acquisition period, and cannot monitor or predict early and late development trends. In addition, the measured data is typically field investigation affected pine or field reflectance spectrum data obtained using a field spectrometer. Studies have shown that only slight changes in the spectrum appear at the beginning of the disease, with a 100% difficulty in field investigation. Due to the phenomena of 'foreign matter identical spectrum' and 'identical and different spectrum' on the high-resolution image, the in-situ reflectivity spectrum data is not suitable after the image area and time are changed.
Therefore, how to provide a time-spectrum mixed early monitoring method for pine wood nematode diseases to realize early prediction and dynamic monitoring of pine wood nematode diseases is a problem that needs to be solved by the technicians in the field.
Disclosure of Invention
In view of the above, the invention provides a time-spectrum mixed early monitoring method for pine wood nematode diseases, solves the problems of low automation degree, lack of universality, difficult early prediction, lack of dynamic monitoring and the like of the existing pine wood nematode disease monitoring based on satellite remote sensing, and provides a new idea for early warning and prevention of forestry insect pests.
In order to achieve the above purpose, the present invention adopts the following technical scheme:
A time-spectrum mixed early detection method for pine wood nematode disease comprises the following steps:
s1: acquiring a time sequence image and an image to be predicted under the monitoring of Sentinel-2 and preprocessing;
S2: calculating a time sequence mean normalized vegetation index based on reflection values of different wave bands in a plurality of time phases in a time sequence image, and extracting pinus massoniana distribution from the time sequence image according to the time sequence mean normalized vegetation index to obtain pinus massoniana distribution conditions;
S3: calculating to obtain a time sequence multi-type red-edge vegetation index sequence and a time sequence multi-type red-edge vegetation index sequence to be predicted based on the preprocessed time sequence images and reflection values of each wave band in a near infrared range and each wave band in a red-edge range in the images to be predicted;
S4: training a prediction model based on the time sequence multi-type red edge vegetation index sequence;
S5: and predicting the multi-type red-edge vegetation index sequence in the period to be predicted by using the prediction model, and obtaining a pine wood nematode pest prediction result by combining the distribution condition of masson pine.
Further, in S1, the preprocessing includes atmospheric correction, geometric correction, resampling, and band synthesis.
Furthermore, the time sequence images adopt Sentinel-2 historical monitoring data with annual cloud cover less than 15%, and the images to be predicted adopt Sentinel-2 real-time monitoring data. The image to be predicted refers to image data of a certain period, which is appointed to be used for early prediction of pine wood nematode diseases, in practical application.
Further, in S2, the method for extracting pinus massoniana distribution comprises,
S21: calculating a time sequence mean normalized vegetation index TSMVI based on the B8 band reflection values and the B4 band reflection values in a plurality of time phases in the time sequence image; wherein the B8 wave band is an NIR near infrared wave band, and the B4 wave band is an RED RED light wave band;
S22: removing non-vegetation through time sequence average normalized vegetation index TSMVI;
S23: determining tree species types by combining the three-tone data and the forestry class data, and calculating time sequence average normalized vegetation index differences of various tree species and pinus massoniana;
s23: and setting a threshold according to the time sequence average normalized vegetation index difference of various tree species and pinus massoniana, and extracting pinus massoniana distribution.
Further, S21 includes calculating a time-series mean normalized vegetation index of each land coverage type in the time-series image, obtaining a time-series mean normalized vegetation index difference of vegetation and non-vegetation land coverage types, determining a difference threshold, performing threshold segmentation on the image to be predicted based on the difference threshold, and eliminating non-vegetation.
Further, S21 includes setting a threshold to TSMVI <0.3, culling non-vegetation; s23 comprises setting a threshold TSMVI <0.7, eliminating masson pine, setting a threshold TSMVI >0.6, and eliminating other broad-leaved tree species. The threshold is set such that TSMVI <0.82 culls fir.
Further, the time sequence mean normalized vegetation index calculation formula is:
Wherein TSMVI is a time-series mean normalized vegetation index, n represents the number of time phases in the time-series image, B8 i represents the B8 band of the ith time phase in the time-series image, and B4 i represents the B4 band of the ith time phase in the time-series image.
Further, the step S3 includes:
s31: extracting data in a red edge range from the time sequence image and the image to be predicted;
S32: calculating multi-type red edge vegetation indexes of each wave band in the red edge range of the time sequence image and the image to be predicted;
s33: and respectively and independently carrying out wave band superposition on the multi-type red edge vegetation indexes of the time sequence image and the multi-type red edge vegetation indexes of the period to be predicted to obtain a time sequence multi-type red edge vegetation index sequence and a period to be predicted multi-type red edge vegetation index sequence.
Further, in S32, the multi-type red edge vegetation index calculation formula includes:
Wherein NDVI re1、NDVIre2、NDVIre3、NDVIre4、NDVIre5 and NDVI re6, are represented as 6 different types; b5, B6, B7, B8 and B8a respectively represent the 5 th, 6 th, 7 th, 8 th and 8 th wave bands of the time sequence image after pretreatment or the image to be predicted.
Further, in S4, the prediction model is a capsule neural network model, and the training sample of the capsule neural network includes a time sequence multi-type red edge vegetation index sequence and early-stage triamcinolone acetonide investigation data, and the capsule neural network model is constructed based on the training sample training.
Compared with the prior art, the method for early monitoring of pine wood nematode diseases by time-spectrum mixing is disclosed, a multi-phase and multi-type red-edge vegetation index is introduced to reflect the healthy change condition of masson pine, a time-spectrum mixed pine wood nematode disease prediction model is established, the spectrum change rule of the masson pine before and after suffering from pine wood nematode diseases is fully excavated, the problem of early monitoring of pine wood nematode diseases by single-period satellite remote sensing is effectively solved, early real-time data of the pine wood nematode diseases can be provided for forestry management related departments rapidly and accurately, and effective data support is provided for forestry pest early warning and prevention.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are required to be used in the embodiments or the description of the prior art will be briefly described below, and it is obvious that the drawings in the following description are only embodiments of the present invention, and that other drawings can be obtained according to the provided drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic diagram of an early detection method of pine wood nematode disease by time-spectrum mixing provided by the invention;
Fig. 2 is a diagram showing a CapsNet-based pine wood nematode disease time-spectrum hybrid monitoring model framework.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
As shown in fig. 1, the embodiment of the invention discloses a time-spectrum mixed early detection method for pine wood nematode disease, which comprises the following steps:
S1: data downloading and preprocessing, namely determining a task area, acquiring a time sequence image and an image to be predicted under the monitoring of a Sentinel-2 satellite in the task area, and preprocessing; the data to be predicted refers to image data of a certain period, which is designated by a user and needs to be used for early prediction of the pine wood nematode disease, in actual application.
S2: extracting pinus massoniana distribution, calculating a time sequence mean normalized vegetation index based on reflection values of different wave bands in a plurality of time phases in a time sequence image, and extracting pinus massoniana distribution from a spectrum image in a period to be predicted according to the time sequence mean normalized vegetation index to obtain a pinus massoniana distribution condition;
s3: based on the preprocessed time sequence image in the S1 and reflection values of each wave band in a near infrared range and each wave band in a red range in the image to be predicted, carrying out combination and normalization on vegetation indexes to obtain a time sequence multi-type red vegetation index sequence and a multi-type red vegetation index sequence to be predicted;
s4: constructing a pine wood nematode disease detection model, and constructing a prediction model based on a time sequence multi-type red-edge vegetation index sequence;
s5: and predicting the model, namely predicting the multi-type red-edge vegetation index sequence in the period to be predicted by using a prediction model, and combining the pinus massoniana distribution condition in the time sequence image to obtain a pine wood nematode pest prediction result. And carrying out space superposition analysis on a pinus massoniana distribution map obtained by extracting pinus massoniana distribution and a pine wire insect pest prediction result to obtain the spatial distribution of suspected infected pinus massoniana, thereby realizing early monitoring of pine wood nematode diseases.
Wherein, the masson pine distribution of S2 can be carried out at any time between S1 and S5 without being limited by the sequence.
In order to further implement the above technical solution, in S1, the preprocessing includes atmospheric correction, geometric correction, resampling, and band synthesis.
The data downloading and the wave band synthesis can be realized by using Python voice programming; the preprocessing work such as atmosphere correction, geometric correction, resampling and the like can be performed by adopting an open source toolkit Sen2Cor of the European space agency, and each time sequence data and the data to be predicted after preprocessing are independently stored according to shooting dates.
In order to further implement the technical scheme, the Sentinel-2 time sequence image adopts Sentinel-2 historical monitoring data with annual cloud cover less than 15%, and the image of the period to be predicted of the Sentinel-2 adopts real-time monitoring data of the Sentinel-2. The detection data of Sentinel-2 contains data of three wave bands in the red edge range, which is very effective for monitoring vegetation health information. Therefore, the invention fully utilizes the special vegetation growth condition monitoring attribute of the red-edge wave band to construct the time sequence multi-type red-edge vegetation index as the effective data of the dynamic monitoring of the pine wood nematode disease.
In order to further implement the technical scheme, in S2, the method for extracting the masson pine distribution comprises the steps of,
S21: dividing the ground surface coverage type of the task area into land coverage types in forest lands, bare lands, built-up areas and water bodies 4 according to the field situation of the task area and the data record, and eliminating non-vegetation according to the difference of time sequence average normalized vegetation indexes TSMVI of different land coverage types;
S22: determining tree species types by combining third-time territory investigation data, namely three-time adjustment data and forestry class data, and calculating time sequence average normalized vegetation index differences of various tree species and pinus massoniana;
S23: setting a segmentation threshold according to the time sequence mean normalized vegetation index difference of various tree species and pinus massoniana, and extracting pinus massoniana distribution.
In order to further implement the above technical solution, S21 includes calculating a time-series mean normalized vegetation index of each land coverage type in the time-series image, obtaining a time-series mean normalized vegetation index difference of vegetation and non-vegetation land coverage types, determining a difference threshold, performing threshold segmentation on the image to be predicted based on the difference threshold, and rejecting non-vegetation.
To further implement the above technical solution, S21 includes setting a threshold to TSMVI <0.3, rejecting non-vegetation; s23 comprises setting a threshold TSMVI <0.7, eliminating masson pine, setting a threshold TSMVI >0.6, and eliminating other broad-leaved tree species. The threshold is set such that TSMVI <0.82 culls fir.
In order to further implement the above technical solution, the calculation formula of the time sequence average normalized vegetation index is:
Wherein TSMVI is a time-series mean normalized vegetation index, n represents the number of time phases in the time-series image, B8 i represents the B8 band of the ith time phase in the time-series image, and B4 i represents the B4 band of the ith time phase in the time-series image.
In order to further implement the above technical solution, S3 includes:
s31: extracting data in a red edge range from the time sequence image and the image to be predicted;
S32: calculating multi-type red edge vegetation indexes of each wave band in the red edge range of the time sequence image and the image to be predicted;
s33: and respectively and independently carrying out wave band superposition on the multi-type red edge vegetation indexes of the time sequence image and the multi-type red edge vegetation indexes of the period to be predicted to obtain a time sequence multi-type red edge vegetation index sequence and a period to be predicted multi-type red edge vegetation index sequence.
In order to further implement the above technical solution, in S32, the multi-type red edge vegetation index calculation formula includes:
Wherein, NDVI re1、NDVIre2、NDVIre3、NDVIre4、NDVIre5 and NDVI re6 are expressed as single-time phase multi-type red edge vegetation indexes corresponding to 6 different types; b5, B6, B7, B8 and B8a respectively represent the 5 th, 6 th, 7 th, 8 th and 8 th wave bands of the time sequence image or the image to be predicted of the masson pine distribution characteristics after pretreatment; sequentially calculating the multi-type red edge vegetation indexes of the time sequence image data and the image data to be predicted after the first step of preprocessing according to the above method, performing wave band superposition on each time sequence multi-type red edge vegetation index according to the data shooting time to obtain a time sequence multi-type red edge vegetation index sequence, and performing wave band superposition on the multi-type red edge vegetation indexes to be predicted independently to obtain a multi-type red edge vegetation index sequence to be predicted.
In order to further implement the above technical scheme, in S4, the prediction model is a capsule neural network model, and the training samples of the capsule neural network include a time sequence multi-type red edge vegetation index sequence and early-stage triamcinolone acetonide investigation data, and the capsule neural network model is constructed based on training of the training samples.
As shown in fig. 2, the pre-infection pine investigation data is taken as a true value, and a pine wood nematode disease training data set is obtained through a space matching algorithm to train a prediction model, so that a pine wood nematode disease time-spectrum hybrid monitoring model is obtained. In order to fully mine the time-space change rule of the time sequence multi-type red edge vegetation indexes, the invention selects CapsNet networks, and because CapsNet replaces neurons in the convolutional neural network with vectors, the spatial relationship among the features is maintained, the problem of effective information loss in the pooling process is avoided, and the classification precision is effectively improved. The pine wood nematode disease time-spectrum hybrid monitoring model established by the invention comprises two main layers:
1) A convolution layer comprising two convolution layers and two maximum pooling layers, wherein the convolution kernel is 3 x 3, the step size is set to 1, and the SAME filling mode is adopted, the maximum pooling layer kernel is set to 2 x2, the step size is 2, and the activation function adopts Relu functions. The maximum pooling layer can compress the feature map of the input data, and simplify the complexity of network calculation on the premise of extracting useful information. This layer enables the capsule network to acquire basic information of the input image and pass it into the capsule layer.
2) And a capsule layer. The layer comprises a main capsule layer, a digital capsule layer and a full-connection layer. Wherein the convolution kernel of the main capsule layer is set to 9 multiplied by 9, the step length is set to 2, the channel number is set to 32, and the filling mode is no filling; the main capsule layer encapsulates the vectors transferred by the convolution layer in the previous step to form a plurality of capsule units. The digital capsule layer performs dimension conversion on the capsule units and performs dynamic route updating in the model training process. The full connection layer uses a Routing algorithm to calculate the final output vector and can obtain the probability that the pixel is identified as pest or pest-free based on the output vector.
Inputting the multi-type red edge vegetation index sequence in the period to be predicted into a pine wood nematode disease time-spectrum mixed monitoring model obtained by the previous training to obtain a pine wood nematode disease prediction binary image, wherein a pixel with a pixel value of 0 in the image represents no plant diseases and insect pests, and a pixel with a pixel value of 1 represents plant diseases and insect pests. And carrying out space superposition analysis on a pinus massoniana distribution map obtained by extracting pinus massoniana distribution and a binary image for predicting the pine wood nematode disease to obtain the space distribution of suspected infected pinus massoniana, thereby realizing early detection of the pine wood nematode disease.
According to the invention, the satellite remote sensing image and the deep learning network are utilized to rapidly and automatically monitor the disease condition of the pine wood nematodes, so that early discovery, early report and early treatment are realized, and the control efficiency of the pine wood nematodes is improved; the method utilizes the time sequence image to fully excavate the spectrum change rule before and after the masson pine suffers from the pine wood nematode disease to establish the time-spectrum mixed model, and compared with a method utilizing a single-period image, the method has the advantages of wide monitoring time range, effective dynamic monitoring and early prediction; the vegetation index is considered as an effective index for monitoring green vegetation, and the red-edge band is taken as a dominant band for monitoring the health condition of the green vegetation; according to the invention, the multi-type red-edge vegetation index is calculated by utilizing three red-edge wave bands of Sentinel-2 data, the potential of the multi-time phase and multi-type red-edge wave band vegetation index to reflect the spectrum change rule of affected pine is fully excavated, a theoretical basis is provided for early prediction of pine wood nematode diseases, and the accuracy and reliability of pine wood nematode disease monitoring are improved.
In the present specification, each embodiment is described in a progressive manner, and each embodiment is mainly described in a different point from other embodiments, and identical and similar parts between the embodiments are all enough to refer to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant points refer to the description of the method section.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (5)
1. The time-spectrum mixed early pine wood nematode disease monitoring method is characterized by comprising the following steps of:
s1: acquiring a time sequence image and an image to be predicted under the monitoring of Sentinel-2 and preprocessing;
S2: calculating a time sequence mean normalized vegetation index based on reflection values of different wave bands in a plurality of time phases in a time sequence image, and extracting pinus massoniana distribution from the time sequence image according to the time sequence mean normalized vegetation index to obtain pinus massoniana distribution conditions; the method for extracting the masson pine distribution comprises the following steps:
S21: calculating a time sequence mean normalized vegetation index TSMVI based on the B8 band reflection values and the B4 band reflection values at a plurality of time phases; the time sequence mean value normalization vegetation index calculation formula is as follows:
wherein TSMVI is a time-series mean normalized vegetation index, n represents the number of time phases in the time-series image, B8 i represents the B8 wave band of the ith time phase in the time-series image, and B4 i represents the B4 wave band of the ith time phase in the time-series image;
S22: removing non-vegetation through time sequence average normalized vegetation index TSMVI; the method specifically comprises the following steps: according to the time sequence average normalized vegetation index of each land coverage type in the time sequence image, obtaining the time sequence average normalized vegetation index difference of vegetation and non-vegetation land coverage types, and determining a difference threshold; performing threshold segmentation on the images to be predicted based on the difference threshold, and eliminating non-vegetation;
S23: determining tree species types by combining the three-tone data and the forestry class data, and calculating time sequence average normalized vegetation index differences of various tree species and pinus massoniana;
s23: setting a threshold according to the time sequence average normalized vegetation index difference of various tree species and pinus massoniana, and extracting pinus massoniana distribution;
s3: calculating to obtain a time sequence multi-type red-edge vegetation index sequence and a time sequence multi-type red-edge vegetation index sequence in a period to be predicted based on the preprocessed time sequence images and reflection values of each wave band in a near infrared range and each wave band in a red-edge range in the period to be predicted; the method comprises the following specific steps:
s31: extracting data in a red edge range from the time sequence image and the image to be predicted;
s32: calculating multi-type red edge vegetation indexes of each wave band in the red edge range of the time sequence image and the image to be predicted; the multi-type red edge vegetation index calculation formula comprises:
Wherein NDVI re1、NDVIre2、NDVIre3、NDVIre4、NDVIre5 and NDVI re6, expressed as 6 different types of normalized vegetation indices; b5, B6, B7, B8 and B8a respectively represent the 5 th, 6 th, 7 th, 8 th and 8 th wave bands of the preprocessed time sequence image or the image to be predicted;
s33: performing wave band superposition on each time sequence multi-type red edge vegetation index according to the data shooting time to obtain a time sequence multi-type red edge vegetation index sequence, and performing wave band superposition on the multi-type red edge vegetation indexes in the period to be predicted independently to obtain the multi-type red edge vegetation index sequence in the period to be predicted;
S4: training a prediction model based on the time sequence multi-type red edge vegetation index sequence;
S5: predicting the multi-type red-edge vegetation index sequence in the period to be predicted by using the prediction model, and obtaining a pine wood nematode damage prediction result by combining the distribution condition of masson pine;
the prediction model adopts CapsNet networks, and comprises two main layers:
1) The convolution layer comprises two convolution layers and two maximum pooling layers, wherein the convolution kernel is 3 multiplied by 3, the step length is set to be 1, a SAME filling mode is adopted, the maximum pooling layer kernel is set to be 2 multiplied by 2, the step length is 2, and the activation function is a Relu function; the maximum pooling layer is used for acquiring basic information of an input image and transmitting the basic information into the capsule layer;
2) The capsule layer comprises a main capsule layer, a digital capsule layer and a full-connection layer; wherein the convolution kernel of the main capsule layer is set to 9 multiplied by 9, the step length is set to 2, the channel number is set to 32, and the filling mode is no filling; the main capsule layer encapsulates the vector transmitted by the convolution layer in the previous step to form a plurality of capsule units; the digital capsule layer performs dimension conversion on the capsule unit and performs dynamic route updating in the model training process; the full connection layer obtains a final output vector through calculation by adopting a Routing algorithm, and obtains the probability that the pixel is identified as being insect attack or not insect attack according to the output vector.
2. A method for early detection of pine wood nematode disease according to claim 1, wherein in S1, the pretreatment comprises atmospheric correction, geometric correction, resampling and band synthesis.
3. The method for early monitoring pine wood nematode disease by time-spectrum mixing according to claim 1, wherein the time sequence image adopts Sentinel-2 historical monitoring data with annual cloud cover less than 15%, and the image to be predicted adopts Sentinel-2 real-time monitoring data.
4. The method for early monitoring pine wood nematode disease of claim 1, wherein non-vegetation is removed when TSMVI < 0.3; TSMVI when the content is less than 0.7, removing masson pine; TSMVI >0.6, removing other broad-leaved tree species; TSMVI <0.82, and removing fir.
5. The method for early monitoring pine wood nematode disease according to claim 1, wherein in S4, the prediction model is a capsule neural network model, and the training samples of the capsule neural network comprise a time sequence multi-type red-edge vegetation index sequence and early-stage disease-infected pine investigation data, and the capsule neural network model is trained based on the training samples.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111341729.1A CN114119513B (en) | 2021-11-12 | 2021-11-12 | Time-spectrum mixed early monitoring method for pine wood nematode disease |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111341729.1A CN114119513B (en) | 2021-11-12 | 2021-11-12 | Time-spectrum mixed early monitoring method for pine wood nematode disease |
Publications (2)
Publication Number | Publication Date |
---|---|
CN114119513A CN114119513A (en) | 2022-03-01 |
CN114119513B true CN114119513B (en) | 2024-10-29 |
Family
ID=80379192
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111341729.1A Active CN114119513B (en) | 2021-11-12 | 2021-11-12 | Time-spectrum mixed early monitoring method for pine wood nematode disease |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114119513B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117975299B (en) * | 2024-04-02 | 2024-06-04 | 山东大学 | Pine wood nematode disease monitoring method and system based on slope characteristics of spectrum curve |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105825177A (en) * | 2016-03-09 | 2016-08-03 | 西安科技大学 | Remote-sensing crop disease identification method based on time phase and spectrum information and habitat condition |
CN109142237A (en) * | 2018-09-13 | 2019-01-04 | 航天信德智图(北京)科技有限公司 | A kind of satellite spectral index of the monitoring infection withered masson pine of pine nematode |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10127451B1 (en) * | 2017-04-24 | 2018-11-13 | Peter Cecil Vanderbilt Sinnott | Method of detecting and quantifying sun-drying crops using satellite derived spectral signals |
CN110779879A (en) * | 2019-11-07 | 2020-02-11 | 航天信德智图(北京)科技有限公司 | Pine wood nematode monitoring method based on red-edge vegetation index |
CN111521562A (en) * | 2020-03-19 | 2020-08-11 | 航天信德智图(北京)科技有限公司 | Cotton vegetation index remote sensing detection method based on Sentinel-2 satellite |
CN113408468A (en) * | 2021-07-01 | 2021-09-17 | 中国科学院东北地理与农业生态研究所 | Forest swamp extraction method based on Sentinel satellite image and random forest algorithm |
-
2021
- 2021-11-12 CN CN202111341729.1A patent/CN114119513B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105825177A (en) * | 2016-03-09 | 2016-08-03 | 西安科技大学 | Remote-sensing crop disease identification method based on time phase and spectrum information and habitat condition |
CN109142237A (en) * | 2018-09-13 | 2019-01-04 | 航天信德智图(北京)科技有限公司 | A kind of satellite spectral index of the monitoring infection withered masson pine of pine nematode |
Also Published As
Publication number | Publication date |
---|---|
CN114119513A (en) | 2022-03-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11751499B2 (en) | Computer-implemented methods, computer readable medium and systems for a precision agriculture platform that detects disease in crops | |
US10838936B2 (en) | Computer-implemented methods, computer readable medium and systems for generating an orchard data model for a precision agriculture platform | |
US10748081B2 (en) | Computer-implemented methods, computer readable medium and systems for a precision agriculture platform that identifies generic anomalies in crops | |
US10395355B2 (en) | Computer-implemented methods, computer readable medium and systems for a precision agriculture platform | |
US10410334B2 (en) | Computer-implemented methods, computer readable medium and systems for a precision agriculture platform with a satellite data model | |
CN107067334B (en) | Farmland fire and straw burning monitoring method and system based on multi-source remote sensing data | |
Verbesselt et al. | Forecasting tree mortality using change metrics derived from MODIS satellite data | |
US10943173B2 (en) | Computer-implemented methods, computer readable medium and systems for generating a satellite data model for a precision agriculture platform | |
CN112287886B (en) | Wheat plant nitrogen content estimation method based on hyperspectral image fusion map features | |
CN114199800B (en) | Rice sheath blight disease identification method, system, equipment and medium | |
CN114119513B (en) | Time-spectrum mixed early monitoring method for pine wood nematode disease | |
CN110658211A (en) | Method for extracting spectral characteristics of cotton canopy of aphid pests in cotton bud period and estimating grade | |
CN115665690A (en) | River buffer zone soil restoration feedback system and restoration method | |
CN112861712B (en) | Agricultural pest and disease monitoring method based on artificial intelligence and multi-temporal remote sensing | |
CN111832920A (en) | Disaster emergency decision method, device, equipment and readable storage medium | |
Schindler et al. | Method for national mapping spatial extent of southern beech forest using temporal spectral signatures | |
CN112613648A (en) | Method for training drought monitoring model, drought monitoring method and equipment | |
Scalisi et al. | Detecting, mapping and digitising canopy geometry, fruit number and peel colour in pear trees with different architecture | |
KR101976090B1 (en) | Monitoring method and device of cultural assets using digital image analysis | |
CN115115955A (en) | Pine wood nematode disease tree monitoring method and device based on unmanned aerial vehicle hyperspectral remote sensing | |
Wang et al. | Extraction of Jujube Planting Areas in Sentinel-2 Image Based on NDVI Threshold—A case study of Ruoqiang County | |
Laneve et al. | Dragon 4-Satellite Based Analysis of Diseases on Permanent and Row Crops in Italy and China | |
Shahin et al. | Southern Wyoming Ecological Forecasting: Monitoring Cheatgrass in Southern Wyoming and Northern Colorado to Inform Management Efforts Post-Mullen Fire | |
de Lima Pessoa et al. | Landscape dynamics and its impacts in the Atlantic Forest area of Southeastern Brazil | |
Lock | The essentiality of essential biodiversity variables: implementation in nature conservation policies |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant |