CN113221972A - Unbalanced hyperspectral data classification method based on weighted depth random forest - Google Patents
Unbalanced hyperspectral data classification method based on weighted depth random forest Download PDFInfo
- Publication number
- CN113221972A CN113221972A CN202110452286.7A CN202110452286A CN113221972A CN 113221972 A CN113221972 A CN 113221972A CN 202110452286 A CN202110452286 A CN 202110452286A CN 113221972 A CN113221972 A CN 113221972A
- Authority
- CN
- China
- Prior art keywords
- classification
- layer
- random forest
- hyperspectral data
- samples
- 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.)
- Granted
Links
- 238000007637 random forest analysis Methods 0.000 title claims abstract description 54
- 238000000034 method Methods 0.000 title claims abstract description 32
- 238000012549 training Methods 0.000 claims abstract description 17
- 230000015572 biosynthetic process Effects 0.000 claims abstract description 8
- 238000003786 synthesis reaction Methods 0.000 claims abstract description 8
- 238000005070 sampling Methods 0.000 claims description 4
- 230000002194 synthesizing effect Effects 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 7
- 238000004088 simulation Methods 0.000 description 7
- 238000012360 testing method Methods 0.000 description 6
- 240000008042 Zea mays Species 0.000 description 4
- 235000005824 Zea mays ssp. parviglumis Nutrition 0.000 description 4
- 235000002017 Zea mays subsp mays Nutrition 0.000 description 4
- 235000005822 corn Nutrition 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 238000012706 support-vector machine Methods 0.000 description 4
- 244000068988 Glycine max Species 0.000 description 3
- 235000010469 Glycine max Nutrition 0.000 description 3
- 238000013527 convolutional neural network Methods 0.000 description 3
- 230000003595 spectral effect Effects 0.000 description 3
- 238000001228 spectrum Methods 0.000 description 3
- 238000003971 tillage Methods 0.000 description 3
- 244000098338 Triticum aestivum Species 0.000 description 2
- 238000013145 classification model Methods 0.000 description 2
- 238000003066 decision tree Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 244000025254 Cannabis sativa Species 0.000 description 1
- 238000007476 Maximum Likelihood Methods 0.000 description 1
- 240000004658 Medicago sativa Species 0.000 description 1
- 235000017587 Medicago sativa ssp. sativa Nutrition 0.000 description 1
- 235000021307 Triticum Nutrition 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000003333 near-infrared imaging Methods 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 239000002689 soil Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A40/00—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production
- Y02A40/10—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production in agriculture
Abstract
The invention discloses an unbalanced hyperspectral data classification method based on a weighted depth random forest in the technical field of remote sensing, which comprises the following steps of firstly, converting an original unbalanced hyperspectral data set into a balanced hyperspectral data set through artificial synthesis oversampling; then, constructing a weighted depth random forest classification model, and training the model layer by layer; and finally, continuously updating the sample weight by using the average classification probability of all samples output by each layer of classification module, so that the next layer of classification module focuses more on a few types of ground object samples, and the classification precision of the original unbalanced hyperspectral data set is improved.
Description
Technical Field
The invention belongs to the technical field of remote sensing, and particularly relates to an unbalanced hyperspectral data classification method based on weighted depth random forests, which is used for solving the problem of low recognition precision of few types of ground objects in unbalanced hyperspectral data.
Background
The hyperspectral remote sensing technology originated in the early 80 s of the 20 th century and was developed on the basis of multispectral remote sensing data. The hyperspectral remote sensing can acquire an approximately continuous spectrum curve in the electromagnetic spectrum ranges of visible light, near infrared, short wave infrared, intermediate infrared and the like through an imaging spectrometer, and organically fuses the spatial information representing the geometric position relation of the ground objects and the spectrum information representing the attribute characteristics of the ground objects together, so that the extraction of the detail information of the ground objects becomes possible. With the improvement of the spectral resolution of the novel imaging spectrometer, people continuously deepen the understanding of spectral attribute characteristics of related ground objects, and a plurality of ground object characteristics hidden in a narrow spectral range are gradually discovered by people, so that the factors greatly accelerate the development of the remote sensing technology, and the hyperspectral remote sensing becomes one of important research directions in the technical field of the remote sensing in the 21 st century.
The hyperspectral image classification is an important research topic, and the core of the hyperspectral image classification is to assign a class label to a pixel. The category distribution, i.e. the proportion of each category in the sample, plays an extremely important role in classification studies. Some traditional classification methods, such as maximum likelihood classification, support vector machines, artificial neural networks, and the like, achieve satisfactory results in balancing hyperspectral data. However, in the real world, class imbalance is a fundamental problem of hyperspectral data. For the hyperspectral data with complicated ground object distribution and unbalanced class, the recognition accuracy of a few classes of ground objects is low, and the requirement of practical application is often difficult to meet. Therefore, the improvement of the classification precision of a few classes in the hyperspectral data is very important for the development of the future hyperspectral technology.
At present, two ways exist to solve the problem of unbalanced hyperspectral image class. The first is to balance the class sample distribution by using data sampling, wherein artificially synthesized oversampling is a sampling method most widely used at present, which balances the class distribution of a data set by synthesizing samples of a few classes of ground objects, but which is prone to generate data noise. And the other method is to design a new classifier to improve the recognition accuracy of the few types of ground objects. The random forest algorithm is a common hyperspectral data classification method, can well tolerate abnormal values and noise, and has parallelism and expandability when processing high-dimensional data, but the random forest has poor capability of processing unbalanced data and needs to be improved to meet the requirement on the classification precision of the unbalanced hyperspectral data.
Disclosure of Invention
Aiming at the defects in the prior art, the invention aims to provide an unbalanced hyperspectral data classification method based on a weighted deep random forest, which combines the deep random forest and artificial synthesis oversampling and improves the overall classification precision and training speed of unbalanced hyperspectral data; the sample weight is utilized to relieve data noise generated by an artificial synthesis oversampling method, and the sample weight is continuously updated to enable the classification model to pay more attention to a few types of samples during classification, so that the overall classification precision of unbalanced hyperspectral data is improved.
In order to achieve the purpose, the invention adopts the following technical scheme to realize the purpose:
the unbalanced hyperspectral data classification method based on the weighted deep random forest comprises the following steps:
step 1, acquiring an original unbalanced hyperspectral data set, and dividing the original unbalanced hyperspectral data set into a majority of ground object samples and a minority of ground object samples;
step 2, performing artificial synthesis oversampling on the original unbalanced hyperspectral data set to obtain a balanced hyperspectral data set;
step 3, constructing a weighted depth random forest classification model, wherein the model comprises a plurality of layers of classification modules, and each layer of classification module comprises random forest classifiers with equal number;
step 4, training the weighted depth random forest classification model layer by layer to obtain the average value of the overall classification precision of a plurality of random forest classifiers in each layer of classification module and the average value of the classification probability of all samples, and updating the sample weight parameter of each layer according to the average value; and when the average value of the overall classification precision of a certain layer of classification module is less than or equal to that of the previous layer, obtaining the classification result of the weighted depth random forest classification model on the original unbalanced hyperspectral data set.
Compared with the prior art, the invention has the following advantages:
firstly, random forests and artificially synthesized oversampling are combined, so that unbalanced hyperspectral data can obtain higher overall classification precision and minority classification precision;
secondly, the invention utilizes the sample weight to relieve the accessory noise generated when a new sample is synthesized by an artificial synthesis oversampling method, and enables the classification model to pay more attention to a few types of samples by continuously updating the sample weight, thereby improving the overall classification precision of the unbalanced hyperspectral data.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings described below are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
FIG. 1 is a flow chart of the present invention;
FIG. 2 is a schematic diagram of artificially synthesized oversampling according to the present invention;
FIG. 3 is a schematic diagram of a weighted depth random forest classification model according to the present invention;
FIG. 4 is a graph of IndianPines hyperspectral image data in simulation of the present invention, wherein (a) is a gray image and (b) is a real terrain map;
FIG. 5 is a diagram of classification results obtained by different methods according to the present invention, wherein (a) is a diagram of the classification results obtained by a support vector machine, (b) is a diagram of the classification results obtained by a random forest, (c) is a diagram of the classification results obtained by a convolutional neural network, and (d) is a diagram of the classification results obtained by the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Referring to fig. 1 and 2, an embodiment of the present invention provides an unbalanced hyperspectral data method based on a weighted depth random forest; the method comprises the following steps:
step 1, obtaining an original unbalanced hyperspectral data setWherein x isiRepresenting samples in the set S, yiDenotes xiSample label of (1), and yiE {1, 2.. eta., C }, wherein C is the number of ground object classes of the original unbalanced hyperspectral data, i e {1, 2.. eta., M }, and M represents the sum of the number of samples in the set S. The category with the largest number of samples is a majority of categories of samples, and the remaining categories are minority of categories of samples, wherein the samples refer to pixels of an image shot by a satellite.
Step 2, carrying out artificial synthesis oversampling on the original unbalanced hyperspectral data to obtain a balanced hyperspectral data set, and specifically comprising the following substeps:
2.1, calculating the sampling rate n of each few types of ground object samples:
wherein, num (y)max) And num (y) represents the number of samples of the minority class of feature and the number of samples of the majority class of feature, respectively, and Round represents rounding;
2.2, randomly selecting a sample x of a few classes of terrainpRoot of Chinese characterFinding x according to Euclidean distance rhopK nearest neighbor samples in a few kinds of ground objects, wherein k is more than or equal to 1, and 1 sample from the k nearest neighbor samples is randomly selected as an auxiliary sample for synthesizing a new sample and is marked as xjThe Euclidean distance formula is as follows:
wherein, ypDenotes xpSample label of (a), yjDenotes xjThe sample label of (1). Calculating to obtain a new sample x according to an interpolation formulanew:
xnew=xp+rand(0,1)×(xp-xj)
Wherein rand (0,1) represents a random number within the interval (0, 1);
2.3, repeating the step 2.2 for n times to obtain n new samples;
2.4, merging the n new samples into the unbalanced hyperspectral data sets S to obtain balanced hyperspectral data setsWherein x isqRepresenting samples in the set S', yqDenotes xqQ ∈ {1, 2., Q }, Q denoting the sum of the number of samples contained in the set S'.
4. The method for classifying unbalanced hyperspectral data based on weighted depth random forests as recited in claim 1, wherein in step 3, the weighted depth random forest classification model consists of L layers of classification modules, each layer of classification module comprises T random forest classifiers, wherein L >1, and L is an integer.
And 4, training the weighted depth random forest classification model layer by layer to obtain the overall classification precision average value of the T random forest classifiers in each layer of classification module and the classification probability average value of all samples, updating the sample weight parameter of each layer of classification module according to the overall classification precision average value, and obtaining the classification result of the weighted depth random forest classification model on the original unbalanced hyperspectral data set when the overall classification precision average value of a certain layer of classification module is less than or equal to the previous layer. The specific training process is as follows:
4.1, setting the initial weight parameter of each sample in the balanced hyperspectral data set S' as 1;
4.2, respectively inputting the set S 'into T random forest classifiers of the first-layer classification module for training to obtain the total classification precision OA of the set S' output by each random forest classifiertAnd the classification probability G of each sampleq,tAnd calculating the total classification accuracy OA output by the T random forest classifierstAverage value of (2)
And the classification probability G of each sample output by the T random forest classifiersq,tAverage value of (2)
4.3, calculating the sample weight W of each sample in the set S ″qThe calculation formula is as follows:
4.4, respectively inputting the set S 'into T random forest classifiers of a second-layer classification module for training to obtain an overall classification precision average value of the second-layer classification module and a classification probability average value of all samples, combining the obtained classification probability average values of all samples of the second-layer classification module into the set S' to obtain a set S ', taking the sample weight of each sample in the set S' as an initial weight parameter of each sample in the second-layer classification module, and so on until the overall classification precision average value of the current-layer classification module is less than or equal to the previous layer, stopping training, taking the current layer as the last layer, and taking the obtained category of each sample as a classification result of the weighted depth random forest classification model on the original unbalanced hyperspectral data set.
After the training is finished, the hyperspectral data are obtained and used as test data to be tested, the test process is the same as the training process, and the class of each test sample can be obtained by inputting the test data into the trained model.
Simulation experiment:
the technical effects of the invention are further explained by simulation experiments as follows:
simulation conditions are as follows: the computer hardware environment of the simulation experiment is Intel (R) core (TM) i5-10200H CPU @2.40 GHz; the software environment of the simulation experiment of the invention is Window 10 and python 3.7.
Referring to fig. 4, data are presented using the airborne visible/near infrared imaging spectrometer AVIRIS of the NASA jet propulsion laboratory in the united states space agency, NASA, acquired in north of indiana nasus at 6 months 1992 as hyperspectral images, where (a) is a gray image and (b) is a real terrain map with an image size of 145X145, with 220 bands, there are 16 terrain categories, and in specific implementations, 30% of samples were selected for training in each category and the rest were used for testing, as shown in table 1:
TABLE 1 number of samples of 16 terrain categories taken by satellite
Categories | Category name | Number of training samples | Number of test samples |
1 | Alfalfa | 14 | 32 |
2 | Corn slight tillage | 428 | 1000 |
3 | Corn (corn) | 249 | 581 |
4 | Pasturing land | 71 | 166 |
5 | No tillage of corn | 144 | 339 |
6 | Woodlands | 219 | 511 |
7 | Reaping pasturing land | 8 | 20 |
8 | Dry grass | 143 | 335 |
9 | Oat | 6 | 14 |
10 | Slightly ploughing soybean | 291 | 681 |
11 | No tillage of soybean | 736 | 1719 |
12 | Cultivated soybean | 177 | 416 |
13 | Wheat (Triticum aestivum L.) | 61 | 144 |
14 | (Forest) | 379 | 886 |
15 | Mixed soil room | 115 | 271 |
16 | Building tree | 28 | 65 |
Referring to fig. 4(a), 4(b) and 5, simulation content and analysis: the method is used for classifying the IndianPines data with the existing three methods (a support vector machine, random forests and a convolutional neural network), in the invention, a weighted depth random forest classification model is constructed to totally comprise 5 layers, wherein the number of decision trees of each random forest is set to be 29; setting a kernel function of a support vector machine method as a Gaussian kernel function; the number of decision trees of the random forest method is set to be 20; the convolutional neural network is set to be 5 layers, namely an input layer, a convolutional layer, a maximum pooling layer, a full-link layer and an output layer. As shown in table 2:
TABLE 2 comparison of classification accuracy of the present invention with that of the prior art
The comparison of the data shows that: according to the method, the artificial synthesis oversampling technology is combined with the weighted depth random forest, and the continuously updated sample weight is utilized to improve the recognition capability of a few types of ground objects in the unbalanced hyperspectral data, so that the classification precision of the unbalanced hyperspectral data is improved. Compared with the three methods in the prior art, the method has the advantages that the overall classification precision of the hyperspectral data and the classification precision of a few types of ground objects are remarkably improved.
The above description is only for the specific embodiments of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present invention, and all the changes or substitutions should be covered within the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the appended claims.
Claims (5)
1. The unbalanced hyperspectral data classification method based on the weighted deep random forest is characterized by comprising the following steps of:
step 1, acquiring an original unbalanced hyperspectral data set, and dividing the original unbalanced hyperspectral data set into a majority of ground object samples and a minority of ground object samples;
step 2, performing artificial synthesis oversampling on the original unbalanced hyperspectral data set to obtain a balanced hyperspectral data set;
step 3, constructing a weighted depth random forest classification model, wherein the model comprises a plurality of layers of classification modules, and each layer of classification module comprises random forest classifiers with equal number;
step 4, training the weighted depth random forest classification model layer by layer to obtain the average value of the overall classification precision of a plurality of random forest classifiers in each layer of classification module and the average value of the classification probability of all samples, and updating the sample weight parameter of each layer of classification module according to the average value; and when the average value of the overall classification precision of a certain layer of classification module is less than or equal to that of the previous layer, obtaining the classification result of the weighted depth random forest classification model on the original unbalanced hyperspectral data set.
2. The method for classifying unbalanced hyperspectral data based on weighted deep random forest as claimed in claim 1, wherein in step 1, an original unbalanced hyperspectral dataset isWherein x isiRepresenting in set SSample, yiDenotes xiSample label of (1), and yiE {1, 2.. eta., C }, wherein C is the number of ground object classes of the original unbalanced hyperspectral data, i e {1, 2.. eta., M }, and M represents the sum of the number of samples in the set S.
3. The method for classifying unbalanced hyperspectral data based on weighted deep random forest as recited in claim 1, wherein the specific steps of artificially synthesizing oversampling are as follows:
2.1, calculating the sampling rate n of each few types of ground object samples:
wherein, num (y)max) And num (y) represents the number of samples of the minority class of feature and the number of samples of the majority class of feature, respectively, and Round represents rounding;
2.2, randomly selecting a sample x of a few classes of terrainpFinding x from Euclidean distancepAnd randomly selecting 1 sample from the k nearest neighbor samples as auxiliary samples, wherein k is more than or equal to 1, and is marked as xjCalculating to obtain a new sample x according to an interpolation formulanew:
xnew=xp+rand(0,1)×(xp-xj)
Wherein rand (0,1) represents a random number within the interval (0, 1);
2.3, repeating the step 2.2 for n times to obtain n new samples;
4. The method for classifying unbalanced hyperspectral data based on weighted depth random forests as recited in claim 1, wherein in step 3, the weighted depth random forest classification model consists of L layers of classification modules, each layer of classification module comprises T random forest classifiers, wherein L >1, and L is an integer.
5. The unbalanced hyperspectral data classification method based on the weighted deep random forest as recited in claim 1, wherein the specific training process in the step 4 is as follows:
4.1, setting the initial weight parameter of each sample in the balanced hyperspectral data set S' as 1;
4.2, respectively inputting the set S 'into T random forest classifiers of the first-layer classification module for training to obtain the total classification precision OA of the set S' output by each random forest classifiertAnd the classification probability G of each sampleq,tAnd calculating the total classification accuracy OA output by the T random forest classifierstAverage value of (2)
And the classification probability G of each sample output by the T random forest classifiersq,tAverage value of (2)
4.3, calculating the sample weight W of each sample in the set S ″qThe calculation formula is as follows:
4.4, respectively inputting the set S 'into T random forest classifiers of a second-layer classification module for training to obtain an overall classification precision average value of the second-layer classification module and a classification probability average value of all samples, combining the obtained classification probability average values of all samples of the second-layer classification module into the set S' to obtain a set S ', taking the sample weight of each sample in the set S' as an initial weight parameter of each sample in the second-layer classification module, and so on until the overall classification precision average value of the current-layer classification module is smaller than or equal to the previous layer, stopping training, taking the current layer as the last layer, and taking the category of each sample output by the last-layer classification module as a classification result of the weighted depth random forest classification model on the original unbalanced hyperspectral data set.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110452286.7A CN113221972B (en) | 2021-04-26 | 2021-04-26 | Unbalanced hyperspectral data classification method based on weighted depth random forest |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110452286.7A CN113221972B (en) | 2021-04-26 | 2021-04-26 | Unbalanced hyperspectral data classification method based on weighted depth random forest |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113221972A true CN113221972A (en) | 2021-08-06 |
CN113221972B CN113221972B (en) | 2024-02-13 |
Family
ID=77088994
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110452286.7A Active CN113221972B (en) | 2021-04-26 | 2021-04-26 | Unbalanced hyperspectral data classification method based on weighted depth random forest |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113221972B (en) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
AU2020100709A4 (en) * | 2020-05-05 | 2020-06-11 | Bao, Yuhang Mr | A method of prediction model based on random forest algorithm |
CN111680615A (en) * | 2020-06-04 | 2020-09-18 | 西安电子科技大学 | Multi-class unbalanced remote sensing land cover image classification method based on integration interval |
CN112131931A (en) * | 2020-08-07 | 2020-12-25 | 中国地质大学(武汉) | Depth forest hyperspectral image classification method and system based on attention mechanism |
WO2021022970A1 (en) * | 2019-08-05 | 2021-02-11 | 青岛理工大学 | Multi-layer random forest-based part recognition method and system |
CN112633337A (en) * | 2020-12-14 | 2021-04-09 | 哈尔滨理工大学 | Unbalanced data processing method based on clustering and boundary points |
-
2021
- 2021-04-26 CN CN202110452286.7A patent/CN113221972B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2021022970A1 (en) * | 2019-08-05 | 2021-02-11 | 青岛理工大学 | Multi-layer random forest-based part recognition method and system |
AU2020100709A4 (en) * | 2020-05-05 | 2020-06-11 | Bao, Yuhang Mr | A method of prediction model based on random forest algorithm |
CN111680615A (en) * | 2020-06-04 | 2020-09-18 | 西安电子科技大学 | Multi-class unbalanced remote sensing land cover image classification method based on integration interval |
CN112131931A (en) * | 2020-08-07 | 2020-12-25 | 中国地质大学(武汉) | Depth forest hyperspectral image classification method and system based on attention mechanism |
CN112633337A (en) * | 2020-12-14 | 2021-04-09 | 哈尔滨理工大学 | Unbalanced data processing method based on clustering and boundary points |
Non-Patent Citations (2)
Title |
---|
刘学;张素伟;: "基于二次随机森林的不平衡数据分类算法", 软件, no. 07 * |
武复宇;王雪;丁建伟;杜培军;谭琨;: "高光谱遥感影像多级联森林深度网络分类算法", 遥感学报, no. 04 * |
Also Published As
Publication number | Publication date |
---|---|
CN113221972B (en) | 2024-02-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110516596B (en) | Octave convolution-based spatial spectrum attention hyperspectral image classification method | |
CN112766199B (en) | Hyperspectral image classification method based on self-adaptive multi-scale feature extraction model | |
CN105608474B (en) | Region adaptivity plant extraction method based on high resolution image | |
CN111062885B (en) | Mark detection model training and mark detection method based on multi-stage transfer learning | |
CN110287869A (en) | High-resolution remote sensing image Crop classification method based on deep learning | |
CN110222215B (en) | Crop pest detection method based on F-SSD-IV3 | |
CN109446986B (en) | Effective feature extraction and tree species identification method for tree laser point cloud | |
CN105184314B (en) | Wrapper formula EO-1 hyperion band selection methods based on pixel cluster | |
CN111898688A (en) | Airborne LiDAR data tree species classification method based on three-dimensional deep learning | |
Tong et al. | An improved multiobjective discrete particle swarm optimization for hyperspectral endmember extraction | |
CN110728197B (en) | Single-tree-level tree species identification method based on deep learning | |
CN109753996B (en) | Hyperspectral image classification method based on three-dimensional lightweight depth network | |
CN107145831B (en) | Hyperspectral image classification method based on vector probability diffusion and Markov random field | |
CN112766161B (en) | Hyperspectral target detection method based on integrated constraint multi-example learning | |
CN109190511B (en) | Hyperspectral classification method based on local and structural constraint low-rank representation | |
CN114266961A (en) | Method for integrating, learning and classifying marsh vegetation stacks by integrating hyperspectral and multiband fully-polarized SAR images | |
Zafari et al. | A multiscale random forest kernel for land cover classification | |
CN110008912B (en) | Social platform matching method and system based on plant identification | |
CN112949738B (en) | Multi-class unbalanced hyperspectral image classification method based on EECNN algorithm | |
CN112329818B (en) | Hyperspectral image non-supervision classification method based on graph convolution network embedded characterization | |
CN107358625B (en) | SAR image change detection method based on SPP Net and region-of-interest detection | |
Liu et al. | Endmember bundle extraction based on improved multi-objective particle swarm optimization | |
CN108509835A (en) | PolSAR image terrain classification methods based on DFIC super-pixel | |
CN110399887B (en) | Representative color extraction method based on visual saliency and histogram statistical technology | |
CN113221972B (en) | Unbalanced hyperspectral data classification method based on weighted depth random forest |
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 | ||
GR01 | Patent grant |