CN112329703A - Construction method of deep convolutional neural network suitable for identifying remote sensing images of pine wilt disease - Google Patents
Construction method of deep convolutional neural network suitable for identifying remote sensing images of pine wilt disease Download PDFInfo
- Publication number
- CN112329703A CN112329703A CN202011311592.0A CN202011311592A CN112329703A CN 112329703 A CN112329703 A CN 112329703A CN 202011311592 A CN202011311592 A CN 202011311592A CN 112329703 A CN112329703 A CN 112329703A
- Authority
- CN
- China
- Prior art keywords
- neural network
- remote sensing
- image
- convolutional neural
- squeezenet
- 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.)
- Pending
Links
- 201000010099 disease Diseases 0.000 title claims abstract description 35
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 title claims abstract description 35
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 22
- 241000018646 Pinus brutia Species 0.000 title claims abstract description 17
- 235000008331 Pinus X rigitaeda Nutrition 0.000 title claims abstract description 16
- 235000011613 Pinus brutia Nutrition 0.000 title claims abstract description 16
- 238000010276 construction Methods 0.000 title claims abstract description 6
- 238000012549 training Methods 0.000 claims abstract description 34
- 238000000034 method Methods 0.000 claims abstract description 19
- 241000243771 Bursaphelenchus xylophilus Species 0.000 claims abstract description 13
- 238000013526 transfer learning Methods 0.000 claims abstract description 4
- 238000013528 artificial neural network Methods 0.000 claims description 16
- 230000000694 effects Effects 0.000 claims description 10
- 230000006872 improvement Effects 0.000 claims description 8
- 238000005457 optimization Methods 0.000 claims description 7
- 238000007781 pre-processing Methods 0.000 claims description 6
- 238000012216 screening Methods 0.000 claims description 6
- 230000004913 activation Effects 0.000 claims description 3
- 230000006870 function Effects 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 230000002194 synthesizing effect Effects 0.000 claims description 3
- 238000012795 verification Methods 0.000 claims description 3
- 241000607479 Yersinia pestis Species 0.000 abstract description 5
- 230000008569 process Effects 0.000 abstract description 5
- 238000012544 monitoring process Methods 0.000 abstract description 3
- 238000013135 deep learning Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 238000013508 migration Methods 0.000 description 4
- 230000005012 migration Effects 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- 230000000007 visual effect Effects 0.000 description 4
- 241000238631 Hexapoda Species 0.000 description 3
- 238000001514 detection method Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 238000003745 diagnosis Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 238000011835 investigation Methods 0.000 description 2
- 238000003672 processing method Methods 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 238000001228 spectrum Methods 0.000 description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000001066 destructive effect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/13—Satellite images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/56—Extraction of image or video features relating to colour
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/194—Terrestrial scenes using hyperspectral data, i.e. more or other wavelengths than RGB
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Multimedia (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Mathematical Physics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Astronomy & Astrophysics (AREA)
- Remote Sensing (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a construction method of a deep convolutional neural network suitable for identifying remote sensing images of pine wilt disease, which comprises the following steps: the method comprises the steps of firstly, constructing a pine wood nematode remote sensing image sample data set, secondly, selecting a Squeezenet convolutional neural network for transfer learning, thirdly, optimizing training parameters of the Squeezenet deep convolutional neural network, and fourthly, optimizing a deep convolutional neural network structure. The method is simple in process, convenient to practical operate and high in practicability, provides a new technical means for monitoring the pine wood nematode disease, and is suitable for the fields of pest and disease monitoring, remote sensing image classification and the like.
Description
One, the technical field
The invention relates to the field of disease and pest remote sensing data analysis and processing and deep learning, in particular to a method for constructing a deep convolutional neural network suitable for identifying remote sensing images of pine wood nematode disease.
Second, background Art
Pine wood nematode disease is one of the most dangerous forest biological disasters in China, and is a destructive disease for pine species. The disease is spread and spread rapidly since 1982, so that a large number of pine trees die, pine forest resources, natural landscapes and ecological environments in China are seriously damaged, and serious economic and ecological losses are caused.
At present, the forest region of the pine wood nematode disease is mainly monitored by manual on-site sampling detection and a remote sensing technology, the manual on-site sampling detection is time-consuming and labor-consuming, the efficiency is too low, the accuracy of the traditional remote sensing image processing method is not high, and deep learning is taken as a novel technology and is increasingly widely researched and applied in the fields of image classification, identification, detection and the like. The pine wilt disease is monitored by a deep learning method, a large amount of manpower and material resources can be saved, and no good method exists for constructing a deep convolution neural network suitable for identifying the remote sensing image of the pine wilt disease at present.
Third, the invention
Technical problem to be solved
The technical problem to be solved by the invention is as follows: how to construct a deep convolutional neural network suitable for identifying the remote sensing image of the pine wilt disease, so that the deep learning technology can be better applied to accurately identify the pine wilt disease disaster area.
(II) technical scheme
In order to solve the technical problem, the invention provides a method for constructing a deep convolutional neural network suitable for identifying remote sensing images of the pine wilt disease, which comprises the following steps of:
1. acquiring a 1A-level remote sensing image product from domestic high-resolution satellites GF-1 and GF-2, and preprocessing the remote sensing image.
2. According to field investigation data, the position and the occurrence time of the disease-sensitive area are determined, the characteristics of image spectra, textures and the like of the areas in corresponding time periods are compared, and the visual interpretation characteristics of the disaster-affected area of the pine wood nematode disease are determined by combining the diagnosis characteristics of the pine wood nematode disease.
3. And according to the visual interpretation result, approximately determining the selection range of the sample to cut the sample, outputting an image in a Tiff/GeoTiff format, and realizing the true value marking of the sample in a file folder and image file filing management mode.
4. Dividing the cut sample into a training-verifying data set and a testing data set, wherein the training-verifying data set is used as input data of a training convolutional neural network, accounts for about 85% of the sample amount of the data set, comprises 80% of training data and 20% of testing data, and is respectively only used for parameter updating training and real-time classification effect evaluation of the convolutional neural network; the latter is only used as input data of the convolutional neural network after training, accounts for about 15% of the sample size of the data set, is independent of the training process, and is used for evaluating the processing capacity and generalization performance of the model.
5. Selecting a SqueezeNet pre-training model for training, and setting common hyper-parameters (batch size is 64, learning rate is 0.001, and epoch is 20) for transfer learning.
6. And optimizing the batch quantity and the learning rate of the Squeezenet deep convolution neural network training parameters.
7. And performing structure optimization of the SqueezeNet deep convolutional neural network, including the improvement of adding bypass connection and module replacement on two macro structures, then performing adjustment optimization on the aspects of replacing an activation function type, introducing a batch normalization layer, introducing a Dropout layer, reducing a network structure and the like on the basis of the improvement of each macro structure, determining an optimal improved model under the improvement of each macro structure, comparing the optimal models after the improvement of the two macro structures with an original model, and finding out the deep convolutional neural network which is most suitable for identifying the remote sensing image of the pine wood nematode disease.
Preferably, the process of preprocessing the remote sensing image in step 1 is as follows: firstly, preliminarily screening GF images, and screening out images which are slightly interfered by cloud and fog and have clear ground objects; then, utilizing ENVI software to respectively carry out orthorectification on the multispectral and panchromatic wave bands of each image according to RPC information of the multispectral and panchromatic wave bands and combining with global 900m resolution DEM data; then, fusing the corrected multispectral image and the panchromatic band image to generate an image with high spatial resolution and multispectral information; and finally, synthesizing an image base map used for marking sample selection by respectively taking the red-green ratio vegetation index (RGRI ═ R/G), the near infrared (Band4) and the blue light Band (Band1) as input bands of R, G, B channels.
Preferably, in the samples described in step 3, the sample categories are classified into 5 categories, which are: pine wilt disease causing areas, healthy forest lands, agricultural land, water and construction land.
Preferably, the optimization of the training parameters and structure of the squeezet deep convolution neural network in the steps 6 and 7 is performed, and4 indexes, such as model training duration, classification accuracy of a verification sample set, convergence speed of a model, stability after model convergence and the like, are mainly used as comparison variables for measuring the network training effect.
Preferably, in step 6, the batch size and the learning rate of the training parameters of the deep convolutional neural network are optimized, the batch size is sequentially set to 32, 64, 128 and 256 which are commonly used in the past research and application to compare the migration training effect of the suitable model under the condition of 4 batch sizes, the learning rate parameter is set to a constant learning rate series and a variable learning rate, the constant learning rate series comprises 1e-4, 5e-4, 1e-3, 3e-3 and 5e-3, the initial value of the variable learning rate is 1e-3, the reduction coefficient is 0.5, and the training is changed once every 5 generations to compare the migration training effect of the suitable model under the condition of different learning rates.
(III) advantageous effects of the invention
Compared with the traditional disease and insect pest remote sensing image processing method, the disease and insect pest image recognition accuracy is higher, the defects of the traditional method in the aspects of processing mass remote sensing data in efficiency and precision can be well overcome, favorable technical support is provided for monitoring, prevention and control of the pine wood nematode disease, and meanwhile, a method reference is provided for research on other disease and insect pests and extraction and analysis of relevant information of the remote sensing image by using a deep learning technology.
Description of the drawings
FIG. 1 is a flow chart of the method of the present invention.
Fifth, detailed description of the invention
The following description will further describe the embodiments of the present invention with reference to the drawings. It should be noted that the description of the embodiments is provided to help understanding of the present invention, but the present invention is not limited thereto. In addition, the technical features involved in the embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.
1. Acquiring a grade-1A remote sensing image product from domestic high-resolution satellites GF-1 and GF-2, and preprocessing the remote sensing image, wherein the preprocessing process comprises the following steps: firstly, preliminarily screening GF images, and screening out images which are slightly interfered by cloud and fog and have clear ground objects; then, utilizing ENVI software to respectively carry out orthorectification on the multispectral and panchromatic wave bands of each image according to RPC information of the multispectral and panchromatic wave bands and combining with global 900m resolution DEM data; then, fusing the corrected multispectral image and the panchromatic band image to generate an image with high spatial resolution and multispectral information; and finally, synthesizing an image base map used for marking sample selection by respectively taking the red-green ratio vegetation index (RGRI ═ R/G), the near infrared (Band4) and the blue light Band (Band1) as input bands of R, G, B channels.
2. The position and the occurrence time of the disease-sensitive area are determined based on field investigation data, the characteristics of image spectra, textures and the like of the areas in corresponding time periods are compared, and the visual interpretation characteristics of the disaster-affected area of the pine wood nematode disease are determined by combining the diagnosis characteristics of the pine wood nematode disease.
3. And determining the selection range of the sample according to the visual interpretation result, cutting the sample, outputting an image in a Tiff/GeoTiff format, and realizing true value marking of the sample in a file folder and image file filing management mode.
Wherein, the sample classification is divided into 5 types, which are respectively: pine wilt disease causing areas, healthy forest lands, agricultural land, water and construction land.
4. Dividing the cut sample into a training-verifying data set and a testing data set, wherein the training-verifying data set is used as input data of a training convolutional neural network, accounts for about 85% of the sample amount of the data set, comprises 80% of training data and 20% of testing data, and is respectively only used for parameter updating training and real-time classification effect evaluation of the convolutional neural network; the latter is only used as input data of the convolutional neural network after training, accounts for about 15% of the sample size of the data set, is independent of the training process, and is used for evaluating the processing capacity and generalization performance of the model.
5. Selecting a SqueezeNet pre-training model for training, and setting common hyper-parameters (batch size is 64, learning rate is 0.001, and epoch is 20) for transfer learning.
6. Optimizing the batch quantity and the learning rate of the training parameters of the SqueezeNet deep convolution neural network, sequentially setting the batch quantity to 32, 64, 128 and 256 which are commonly used in the past research and application, and determining the optimum batch quantity of the SqueezeNet deep convolution neural network by comparing the migration training effect of the suitable model under the condition of 4 batch quantities. And setting the learning rate parameters as a constant learning rate series and a variable learning rate, wherein the constant learning rate series comprises 1e-4, 5e-4, 1e-3, 3e-3 and 5e-3, the initial value of the variable learning rate is 1e-3, the reduction coefficient is 0.5, the training is changed once every 5 generations, and the optimum learning rate of the Squeezet deep convolution neural network is determined by comparing the migration training effects of suitable models under different learning rates.
4 indexes, such as model training duration, classification accuracy of the verification sample set, convergence speed of the model, stability after the model is converged, and the like are used as comparison variables for measuring the network training effect.
7. Performing structure optimization of an SqueezeNet deep convolution neural network, firstly improving the deep convolution neural network in the aspects of increasing bypass connection and module replacement, then performing adjustment optimization in the aspects of replacing activation function types (ReLu, Leaky ReLu, Elu and Tanh), introducing a batch normalization layer, introducing a Drapout layer, reducing a network structure and the like on the basis of improvement of each macro structure, determining an optimal improved model under improvement of each macro structure, comparing the improved models of the two macro structures with an original model, and finding out the deep convolution neural network most suitable for identifying remote sensing images of pine wood nematode diseases.
Claims (5)
1. A construction method of a deep convolution neural network suitable for identifying remote sensing images of pine wood nematode disease is characterized by comprising the following steps: firstly, acquiring a high-resolution remote sensing image, preprocessing the remote sensing image, visually interpreting, cutting the image and manually marking; secondly, selecting a SqueezeNet deep convolution neural network for transfer learning; then, optimizing the training parameters of the SqueezeNet deep convolution neural network; and finally, optimizing the structure of the Squeezenet deep convolution neural network.
2. The method for constructing the deep convolutional neural network suitable for identifying the remote sensing image of the pine wilt disease according to claim 1, which is characterized by comprising the following steps of: preprocessing the remote sensing image, firstly, preliminarily screening the GF image, and screening out the images which are less interfered by cloud and fog and clear in various ground objects; and then performing orthorectification on the multispectral and panchromatic bands of each image according to RPC information of the multispectral and panchromatic bands and combining with global 900m resolution DEM data by utilizing ENVI software. And then, fusing the corrected multispectral image and the panchromatic band image to generate an image with high spatial resolution and multispectral information. And finally, synthesizing an image base map used for marking sample selection by respectively taking the red-green ratio vegetation index (RGRI ═ R/G), the near infrared (Band4) and the blue light Band (Band1) as input bands of R, G, B channels.
3. The method for constructing the deep convolutional neural network suitable for identifying the remote sensing image of the pine wilt disease according to claim 1, which is characterized by comprising the following steps of: optimizing the training parameters and structure of the SqueezeNet deep convolution neural network, and taking 4 indexes of model training duration, classification accuracy of a verification sample set, convergence speed of the model, stability of the model after convergence and the like as comparison variables for measuring the network training effect.
4. The method for constructing the deep convolutional neural network suitable for identifying the remote sensing image of the pine wilt disease according to claim 1, which is characterized by comprising the following steps of: the method mainly comprises the step of optimizing training parameters of the Squeezenet deep convolution neural network, wherein the optimal batch quantity and learning rate are researched.
5. The method for constructing the deep convolutional neural network suitable for identifying the remote sensing image of the pine wilt disease according to claim 1, which is characterized by comprising the following steps of: the optimization of the SqueezeNet deep convolution neural network structure comprises the improvement of adding bypass connection and module replacement on two macrostructures, and then adjustment optimization of replacing an activation function type, introducing a batch normalization layer, introducing a Dropout layer, reducing the network structure and the like is carried out on the basis of each macrostructure improvement.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011311592.0A CN112329703A (en) | 2020-11-20 | 2020-11-20 | Construction method of deep convolutional neural network suitable for identifying remote sensing images of pine wilt disease |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011311592.0A CN112329703A (en) | 2020-11-20 | 2020-11-20 | Construction method of deep convolutional neural network suitable for identifying remote sensing images of pine wilt disease |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112329703A true CN112329703A (en) | 2021-02-05 |
Family
ID=74321010
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011311592.0A Pending CN112329703A (en) | 2020-11-20 | 2020-11-20 | Construction method of deep convolutional neural network suitable for identifying remote sensing images of pine wilt disease |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112329703A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112861812A (en) * | 2021-03-25 | 2021-05-28 | 东北林业大学 | Pine wood nematode disaster prediction and forecast method and device based on deep learning |
| CN113033520A (en) * | 2021-05-25 | 2021-06-25 | 华中农业大学 | Tree nematode disease wood identification method and system based on deep learning |
| CN114140428A (en) * | 2021-11-30 | 2022-03-04 | 东北林业大学 | Method and system for detecting and identifying larch caterpillars based on YOLOv5 |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108596248A (en) * | 2018-04-23 | 2018-09-28 | 上海海洋大学 | A kind of classification of remote-sensing images model based on improvement depth convolutional neural networks |
| CN109948563A (en) * | 2019-03-22 | 2019-06-28 | 华南农业大学 | A kind of withered tree detection localization method of the pine nematode based on deep learning |
| CN111951285A (en) * | 2020-08-12 | 2020-11-17 | 湖南神帆科技有限公司 | Optical remote sensing image woodland classification method based on cascade deep convolutional neural network |
-
2020
- 2020-11-20 CN CN202011311592.0A patent/CN112329703A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108596248A (en) * | 2018-04-23 | 2018-09-28 | 上海海洋大学 | A kind of classification of remote-sensing images model based on improvement depth convolutional neural networks |
| CN109948563A (en) * | 2019-03-22 | 2019-06-28 | 华南农业大学 | A kind of withered tree detection localization method of the pine nematode based on deep learning |
| CN111951285A (en) * | 2020-08-12 | 2020-11-17 | 湖南神帆科技有限公司 | Optical remote sensing image woodland classification method based on cascade deep convolutional neural network |
Non-Patent Citations (7)
| Title |
|---|
| VARSHA BHOLE等: "Analysis of Convolutional Neural Network Using Pre-Trained Squeezenet Model for Classification of Thermal Fruit Images" * |
| 吕明站;朱子魁;: "基于无人机和人工智能的异常林木快速识别技术研究" * |
| 孟祥锐: "基于深度学习的淡水湿地遥感精细分类研究" * |
| 张瑞瑞;夏浪;陈立平;谢春春;陈梅香;王维佳;: "基于U-Net网络和无人机影像的松材线虫病变色木识别" * |
| 徐信罗等: "基于Faster R-CNN 的松材线虫病受害木识别与定位" * |
| 李坤伦;魏泽发;宋焕生;: "基于SqueezeNet卷积神经网络的车辆颜色识别" * |
| 董艺威等: "基于SqueezeNet的轻量化卷积神经网络SlimNet" * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112861812A (en) * | 2021-03-25 | 2021-05-28 | 东北林业大学 | Pine wood nematode disaster prediction and forecast method and device based on deep learning |
| CN113033520A (en) * | 2021-05-25 | 2021-06-25 | 华中农业大学 | Tree nematode disease wood identification method and system based on deep learning |
| CN113033520B (en) * | 2021-05-25 | 2021-08-13 | 华中农业大学 | Tree nematode disease wood identification method and system based on deep learning |
| CN114140428A (en) * | 2021-11-30 | 2022-03-04 | 东北林业大学 | Method and system for detecting and identifying larch caterpillars based on YOLOv5 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110321963B (en) | Hyperspectral image classification method based on fusion of multi-scale and multi-dimensional space spectrum features | |
| CN112329703A (en) | Construction method of deep convolutional neural network suitable for identifying remote sensing images of pine wilt disease | |
| CN112287892B (en) | Arbor biomass measurement and calculation method based on unmanned aerial vehicle hyperspectral and machine learning algorithm | |
| CN104408478A (en) | Hyperspectral image classification method based on hierarchical sparse discriminant feature learning | |
| CN111738066B (en) | Grid late rice sheath blight disease habitat evaluation method integrating multisource remote sensing information | |
| CN112102312A (en) | Moso bamboo forest remote sensing identification method based on satellite image and phenological difference containing red edge wave band | |
| CN111126511B (en) | Vegetation index fusion-based LAI quantitative model establishment method | |
| CN113657158B (en) | Google EARTH ENGINE-based large-scale soybean planting area extraction algorithm | |
| CN116883853B (en) | Crop space-time information remote sensing classification method based on transfer learning | |
| CN113221765A (en) | Vegetation phenological period extraction method based on digital camera image effective pixels | |
| CN112906537A (en) | Crop identification method and system based on convolutional neural network | |
| CN110751035A (en) | Seed corn production identification method and device | |
| CN117058522A (en) | Remote sensing earth surface ecological change detection method integrating spectral neural network and leaf area index | |
| Bhuyar et al. | Crop classification with multi-temporal satellite image data | |
| Fu et al. | Cotton aphid infestation monitoring using Sentinel-2 MSI imagery coupled with derivative of ratio spectroscopy and random forest algorithm | |
| CN111832480B (en) | Remote sensing identification method for rape planting area based on spectral characteristics | |
| Lemenkova et al. | Time series analysis of Landsat images for monitoring flooded areas in the inner Niger Delta, Mali | |
| Wang et al. | Poplar seedling varieties and drought stress classification based on multi-source, time-series data and deep learning | |
| CN112528726A (en) | Aphis gossypii insect pest monitoring method and system based on spectral imaging and deep learning | |
| CN115063610B (en) | Soybean planting area identification method based on Sentinel-1 and 2 images | |
| CN116486282A (en) | Digital elevation model manufacturing method and system based on deep learning, electronic equipment and storage medium | |
| CN114373140A (en) | Intelligent identification method for pine wood nematode disease occurrence area | |
| CN113705454A (en) | Method for extracting forest land containing infrared spectrum remote sensing image | |
| CN113191177A (en) | Pine wood nematode remote sensing image sample labeling method for deep learning | |
| CN114220028B (en) | Rare tree species identification method based on unmanned aerial vehicle hyperspectral image and machine learning algorithm model research and development |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20210205 |