CN105005789B - A kind of remote sensing images terrain classification method of view-based access control model vocabulary - Google Patents
A kind of remote sensing images terrain classification method of view-based access control model vocabulary Download PDFInfo
- Publication number
- CN105005789B CN105005789B CN201510379234.6A CN201510379234A CN105005789B CN 105005789 B CN105005789 B CN 105005789B CN 201510379234 A CN201510379234 A CN 201510379234A CN 105005789 B CN105005789 B CN 105005789B
- Authority
- CN
- China
- Prior art keywords
- remote sensing
- slice map
- word
- sensing images
- preliminary
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of remote sensing images terrain classification method of view-based access control model vocabulary, comprise the following steps：All remote sensing images are divided into training set and test set first, every width remote sensing images are cut with fixed size and obtain preliminary slice map, extraction includes the preliminary slice map of target；Pass through Gaussian Blur and sampling generation multilayer Gaussian spatial pyramid for the preliminary slice map comprising target；SIFT feature extraction and LBP feature extractions are carried out to each layer of image；The remote sensing word of all remote sensing images in training set and test set is clustered, obtains multiple cluster centres, all cluster centres composition remote sensing dictionary；Different radii value is set, frequency histogram is established to the remote sensing word in each preliminary slice map different radii value：It is trained for the frequency histogram of remote sensing word in training set using SVMs RBF SVM, terrain classification then is carried out to remote sensing images in test set using the RBF SVM after training.
Description
Technical field
The invention belongs to technical field of remote sensing image processing, is related to a kind of terrain classification based on high-resolution remote sensing image
Method.
Background technology
With developing rapidly for remote sensing satellite, the resolution ratio of the satellite-remote-sensing image of acquisition also more and more higher, thus trigger
The also more and more higher of the processing requirement for remote sensing images, wherein, for the object Classification and Identification of High Resolution Remote Sensing Satellites
It is an important research direction, especially city scientific allocation, agricultural planting distribution planning and military sensitive target point
Class extraction is significant.The high-resolution of High Resolution Remote Sensing Satellites also implies that details is more rich simultaneously, and background is to mesh
Target influence is also bigger, and this just brings challenge for the classification of image.
In remote sensing image classification field, mainly include two class sorting techniques：One kind is non-supervisory method, by using
The thought of cluster realizes, including the methods of k-means, isodata.But as remote sensing images resolution ratio more and more higher has
Inter-class variance is reduced, variance within clusters increase, overlapping between class so as to cause, and the different spectrum of jljl and the phenomenon of same object different images is occurred, is led
Cause classification error；Another kind of is to have measure of supervision, and the sorting technique generally used now, using the think of of study and training
Think, including BP neural network model, genetic model and SVMs etc., the method for supervision possess certain increasing amount adjustment,
But when remote sensing images resolution ratio constantly increases, selection for training sample and for every class sample training speed all
It is a challenge, the selection of especially training sample is the process of an artificial selection, brings bigger time cost.
The development of high-resolution remote sensing image means that the species of feature selecting increases simultaneously, with color, texture, shape etc.
The relation with sample high layer information set up based on low-level image feature, it may appear that the shortcomings that generalization is poor, and relevance grade is not high.
In order to overcome the wide gap of low-level image feature and the high-level semantic information of sample, be thus born based on sample intermediate features with
Derivation model between the high-level semantic information of sample.Representative of BOV (vision bag of words) models as intermediate features, in OBIA
There are some to be in progress in terms of (object-based remote-sensing image analysis).It is but traditional
BOV model methods itself are defective, it is impossible to the good full detail using image, such as spatial level information etc., and
BOV categories of model are mainly used in the Classification and Identification of natural forms, are not utilized very well for remote sensing images terrain classification.
How to be ground with the thought of bag of words to solve remote sensing images terrain classification into remote sensing images field is important
Study carefully one of direction.But species are varied due to remote sensing images, single feature bag of words method and single remote sensing list
Word frequency histogram is difficult to represent all remote sensing images species.
The content of the invention
In view of this, the invention provides a kind of remote sensing images terrain classification method of view-based access control model vocabulary, solve distant
The problem of feeling image terrain classification accuracy rate deficiency.
In order to achieve the above object, technical scheme comprises the following steps：
Step 1: all remote sensing images are divided into training set and test set, for every width remote sensing images, with fixed big
Small cut to it obtains preliminary slice map, and extraction includes the preliminary slice map of target.
It is Step 2: more by Gaussian Blur and sampling generation for the preliminary slice map comprising target extracted in step 1
This spatial pyramid of floor height.
Step 3: carrying out feature extraction to each layer of image in Gaussian spatial pyramid, feature extraction includes office
Portion's feature is SIFT feature extraction and the i.e. LBP feature extractions of image texture characteristic.
When carrying out SIFT feature extraction, one layer of gaussian filtering is only carried out, is then divided into image by way of sliding window
16 × 16 subdivision slice map, SIFT feature vector extraction is carried out according to the graded of each subdivision slice map and closed
In the SIFT feature vector of 128 dimensions of the subdivision slice map；According to the radius and sampled point of setting in each subdivision slice map
Number determines the LBP characteristic vectors on the subdivision slice map.
Then each SIFT feature vector sum LBP combination of eigenvectors corresponding to subdivision slice map forms a remote sensing word；
Thus the remote sensing word on the remote sensing images is generated.
Wherein top Gaussian spatial pyramid meets correspondingly to obtain remote sensing word number at least in its preliminary slice map
For the 1/2 of the quantity of remote sensing dictionary.
The mode of following steps four~five is taken to be handled for the preliminary slice map in each layer of gaussian pyramid：
Step 4: being clustered to the remote sensing word of all remote sensing images in training set and test set, multiple clusters are obtained
Center, all cluster centres composition remote sensing dictionary.
In cluster process, dimensionality reduction is carried out using PCA PCAs to remote sensing word, retained to covariance contribution most
Big dimension, and set data loss rate threshold value disratio so that data loss rate is no more than disratio.
Step 5: setting different radii value, frequency is established to the remote sensing word in each preliminary slice map different radii value
Rate histogram：
The abscissa of the histogram is each remote sensing word, and ordinate is remote sensing word tentatively cutting in current radius value
The frequency occurred in piece figure, the computational methods of the frequency are：Frequency values are initially 0, calculate the preliminary section in current radius value
Euclidean distance between remote sensing word two-by-two in figure, if the Euclidean distance between current remote sensing word A to another remote sensing word B is
Frequency values increase by 1 of other all remote sensing words to the minimum value, then A of B Euclidean distance.
Step 6: be trained for the frequency histogram of remote sensing word in training set using SVMs RBF-SVM,
Then terrain classification is carried out to remote sensing images in test set using the RBF-SVM after training.
If the number of plies of the gaussian pyramid belonging to remote sensing word in the bottom, is distributed in its corresponding SVMs point
The maximum weight of class device, more up, the weights of corresponding grader are smaller for the number of plies.
Further, when image to be divided into 16*16 subdivision slice map by way of sliding window, the step-length of sliding window is arranged to
Between 1 to 8.
Further, the Selection of kernel function histogram intersection in RBF-SVM SVMs
kernel。
Further, K-means clustering methods are used in step 2, during cluster.
Beneficial effect：
1st, remote sensing images collection is divided into test set training set two parts by this method first, then to all remote sensing images
Section carries out feature point extraction, obtains the word of every width slice map, the characteristic point for training set is clustered to obtain word afterwards
Allusion quotation, the word structure frequency histogram for being finally directed to dictionary and test set are trained, and use the word histogram of test set
Tested, more accurately remote sensing image atural object can be classified, classification results are more accurate.
2nd, the method in this method when carrying out Gaussian Blur and sampling to slice map using multilayer gaussian pyramid, therefore
The spatial information of remote sensing images can be better profited from, so that classification is more accurate.
3rd, this method it is traditional due to SIFT feature vector only include 16*16 slice maps in direction and gradient information,
The information of point of safes i.e. in slice map, do not possess the texture feature information of whole slice map, therefore on the basis of SIFT feature
The uniform LBP features of invariable rotary are added, the texture feature information not possessed in SIFT feature is compensate for, can express in all directions
All information in remote sensing images.
4th, in this method when generating frequency histogram, multiple different radiuses are chosen, while ensure that selected radius should
When cause it is internal include enough remote sensing word quantity, to the remote sensing word in pyramid in each layer of slice map different radii and
Remote sensing dictionary generates frequency histogram, so can be on the basis of meeting the needs of generating remote sensing word frequencies histogram, together
Spatial information in Shi Liyong remote sensing images at center to surrounding, so as to preferably express the target in remote sensing images,
Improve the classification degree of accuracy.
5th, this method finally adds decision-making device to obtain the result of decision using multi-categorizer, for each layer in Gauss sky pyramid
Remote sensing word frequencies histogram classified, distribute weights to different grader, it is believed that gaussian pyramid bottom
Information maximum weight, more up information weights are smaller, the generic of remote sensing slice map obtained by decision-making, by Gauss each time
Pyramid classification results distribute weights, do not trust the classification results of a certain tomographic image merely, pass through different levels gaussian pyramid
Classification results determine target in remote sensing images jointly, improve the classification degree of accuracy.
Brief description of the drawings
Fig. 1 is the implementing procedure figure of the present invention.
Fig. 2 is the product process figure of remote sensing word of the present invention.
Fig. 3 is the product process figure of remote sensing word frequencies histogram of the present invention.
Fig. 4 is the design of grader of the present invention.
Embodiment
The present invention will now be described in detail with reference to the accompanying drawings and examples.
Step 1: all remote sensing images are divided into training set and test set, for every width remote sensing images, with fixed big
Small cut to it obtains preliminary slice map, and extraction includes the preliminary slice map of target；
It is Step 2: more by Gaussian Blur and sampling generation for a kind of preliminary slice map comprising target of extraction of step
This spatial pyramid of floor height, the pyramidal number of plies of Gaussian spatial determine according to the size of preliminary slice map；
Accompanying drawing 2 is the product process figure of remote sensing word, includes the generation of Gaussian spatial pyramid and multi-feature extraction two parts.
First against remote sensing slice map by Gaussian Blur and sampling generation multilayer Gaussian spatial pyramid, difference is directed to
The slice map of size, can generate the Gaussian spatial pyramid of the different numbers of plies, and top pyramid is required to meet at least
The 1/2 of remote sensing dictionary quantity is obtained with the remote sensing word frequencies histogram after composition, is 16 groups of characteristic values in the present embodiment, this
Sample can ensure that generated frequency histogram is significant.
The Gaussian spatial pyramid of multilayer can better profit from the spatial information of remote sensing images, so that classification is more
Accurately.
Secondly, multiple features (SIFT feature and LBP features) are carried out for every piece image in Gaussian spatial pyramid to carry
Take.In order to keep the uniformity of each image frequency histogram quantity, SIFT feature is changed, only carrying out a floor height, this is filtered
Ripple, image is divided into subdivision slice map by way of sliding window, the slice map for being 16*16 in the present embodiment, the step-length of sliding window can
It is actually needed with basis and is arranged to 1 to 8,128 dimensions is obtained according to the graded of each 16*16 subdivision slice map
SIFT feature vector.It is consistent to be so directed to the SIFT feature points of each sub-picture, the statistics with histogram after being easy to.
Direction and the gradient information in 16*16 slice maps, i.e. point of safes in slice map are only included due to SIFT feature vector
Information, do not possess the texture feature information of whole slice map, therefore it is uniform to add on the basis of SIFT feature invariable rotary
LBP features can express all information in remote sensing images in all directions to make up the sign for image.In each 16*16
Slice map in LBP features are determined according to radius and sampling number (being manually set), it is special as a result of the LBP of invariable rotary
Sign so that the dimension of representative image textural characteristics is less.
With rotating uniformly constant LBP features to combine it is exactly represent diagram picture distant by the SIFT feature of each image
Feel word (the corresponding remote sensing word of each 16*16 slice map), the remote sensing word quantity from the generation of every width slice map is one
Cause.
Step 2: being clustered based on remote sensing word all in training set and test set, multiple cluster centres, institute are obtained
Some cluster centres just constitute remote sensing dictionary.
Because cluster is an offline process, with increasing for cluster data, the overall partition clustering side of ordinary meaning
Method can produce the computing cost of geometry multiple.Therefore the thought of dimensionality reduction cluster handles extensive and ultra-large remote sensing word
Allusion quotation data.
Herein, dimensionality reduction is carried out using PCA PCAs to remote sensing word, retains the dimension maximum to covariance contribution
Degree, it is however noted that, dimensionality reduction number needs to ensure not lose useful data largely, therefore, defines data loss rate
Disratio, ensure that remaining data retains most of useful information.
For the method for cluster using the balanced K-means clustering methods of time efficiency and space efficiency, K-means clusters can
To directly obtain cluster centre, and cluster centre is the average value of all properties in class, can be very good to represent center in class.
The number of cluster centre need to be determined according to the number of classification, it is necessary to the classification number classified it is more, it is necessary to cluster centre just
It is more, also imply that the word number in remote sensing dictionary is more, could meet the needs of multi-class targets classification accuracy.
Step 3: setting different radii value, frequency is established to the remote sensing word in each preliminary slice map different radii value
Rate histogram：
The abscissa of the histogram is each remote sensing word, and ordinate is remote sensing word tentatively cutting in current radius value
The frequency occurred in piece figure, the computational methods of the frequency are：Frequency values are initially 0, calculate the preliminary section in current radius value
Euclidean distance between remote sensing word two-by-two in figure, if the Euclidean distance between current remote sensing word A to another remote sensing word B is
Frequency values increase by 1 of other all remote sensing words to the minimum value, then A of B Euclidean distance.
Accompanying drawing 3 is the product process figure of remote sensing word frequencies histogram of the present invention, herein in order to effectively using in picture
Spatial positional information, be directed to the spatial pyramid of different levels, choose suitable radius, in pyramid each layer section
Remote sensing word and remote sensing dictionary generation frequency histogram in figure different radii, by the remote sensing word all on this slice map
Frequency histogram combines, as the remote sensing word frequencies histogram for representing this preliminary slice map.
When choosing suitable radius, it should meet that its inside includes enough remote sensing word quantity, it is distant so as to meet to generate
Feel the demand of word frequencies histogram.The benefit for selecting multiple radiuses be utilize remote sensing images at center to the space of surrounding
Information, so as to preferably express the target in remote sensing images, improve the classification degree of accuracy.
Step 4: the remote sensing frequency histogram of the remote sensing word frequencies histogram and test set for training set uses RBF-
SVM SVMs is trained, and Selection of kernel function has the histogram intersection of good behaviour in statistics field
kernel。
Accompanying drawing 4 is the composition of grader, the result of decision is obtained using multi-categorizer plus decision-making device, for Gauss sky pyramid
In each layer of remote sensing word frequencies histogram classified, distribute weights to different graders, it is believed that Gauss gold word
The information maximum weight of bottom of towe layer, more up information weights are smaller, and the generic of remote sensing slice map is obtained by decision-making.By to every
One time gaussian pyramid classification results distribute weights, do not trust the classification results of a certain tomographic image merely, pass through different levels height
This pyramidal classification results determines the target in remote sensing images jointly, improves the classification degree of accuracy.
Since then, the terrain classification of remote sensing images is just completed.
To sum up, presently preferred embodiments of the present invention is these are only, is not intended to limit the scope of the present invention.It is all
Within the spirit and principles in the present invention, any modification, equivalent substitution and improvements made etc., the protection of the present invention should be included in
Within the scope of.
Claims (4)
1. a kind of remote sensing images terrain classification method of view-based access control model vocabulary, it is characterised in that comprise the following steps：
Step 1: all remote sensing images are divided into training set and test set, for every width remote sensing images, with fixed size pair
It, which cuts, obtains preliminary slice map, and extraction includes the preliminary slice map of target；
Step 2: more floor heights are generated by Gaussian Blur and sampling for the preliminary slice map comprising target extracted in step 1
This spatial pyramid；
Step 3: feature extraction, the feature extraction bag are carried out to each layer of image in the Gaussian spatial pyramid
Include local feature i.e. SIFT feature extraction and the i.e. LBP feature extractions of image texture characteristic；
When carrying out SIFT feature extraction, one layer of gaussian filtering is only carried out, image is then divided into 16 by way of sliding window ×
16 subdivision slice map, SIFT feature vector extraction is carried out according to the graded of each subdivision slice map and obtained on this
Segment the SIFT feature vector of 128 dimensions of slice map；Radius and sampling number according to setting in each subdivision slice map is true
The fixed LBP characteristic vectors on the subdivision slice map；
Then each SIFT feature vector sum LBP combination of eigenvectors corresponding to subdivision slice map forms a remote sensing word；Thus
Generate the remote sensing word on the remote sensing images；
Wherein top Gaussian spatial pyramid meets correspondingly to obtain remote sensing word number in its preliminary slice map at least distant
Feel the 1/2 of the quantity of dictionary；
The mode of following steps four~five is taken to be handled for the preliminary slice map in each layer of gaussian pyramid：
Step 4: being clustered to the remote sensing word of all remote sensing images in training set and test set, multiple cluster centres are obtained,
All cluster centre composition remote sensing dictionaries；
In cluster process, dimensionality reduction is carried out using PCA PCAs to remote sensing word, retained to covariance contribution maximum
Dimension, and set data loss rate threshold value disratio so that data loss rate is no more than disratio；
Step 5: setting different radii value, it is straight to establish frequency to the remote sensing word in each preliminary slice map different radii value
Fang Tu：
The abscissa of the histogram is each remote sensing word, and ordinate is preliminary slice map of the remote sensing word in current radius value
The frequency of middle appearance, the computational methods of the frequency are：Frequency values are initially 0, calculate in the preliminary slice map in current radius value
Euclidean distance between remote sensing word two-by-two, if the Euclidean distance between current remote sensing word A to another remote sensing word B is all
Frequency values increase by 1 of the remote sensing word to the minimum value, then A of B Euclidean distance；
Step 6: it is trained for the frequency histogram of remote sensing word in training set using SVMs RBF-SVM, then
Terrain classification is carried out to remote sensing images in test set using the RBF-SVM after training；
If the number of plies of the gaussian pyramid belonging to remote sensing word in the bottom, distributes grader in its corresponding SVMs
Maximum weight, more up, the weights of corresponding grader are smaller for the number of plies.
2. the remote sensing images terrain classification method of a kind of view-based access control model vocabulary as claimed in claim 1, it is characterised in that described
When image to be divided into 16*16 subdivision slice map by way of sliding window, the step-length of sliding window is arranged between 1 to 8.
3. the remote sensing images terrain classification method of a kind of view-based access control model vocabulary as claimed in claim 1, it is characterised in that described
Selection of kernel function histogram intersection kernel in RBF-SVM SVMs.
A kind of 4. remote sensing images terrain classification method of view-based access control model vocabulary as claimed in claim 1, it is characterised in that step
K-means clustering methods are used in two, during cluster.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510379234.6A CN105005789B (en) | 2015-07-01 | 2015-07-01 | A kind of remote sensing images terrain classification method of view-based access control model vocabulary |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510379234.6A CN105005789B (en) | 2015-07-01 | 2015-07-01 | A kind of remote sensing images terrain classification method of view-based access control model vocabulary |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105005789A CN105005789A (en) | 2015-10-28 |
| CN105005789B true CN105005789B (en) | 2018-04-03 |
Family
ID=54378454
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510379234.6A Active CN105005789B (en) | 2015-07-01 | 2015-07-01 | A kind of remote sensing images terrain classification method of view-based access control model vocabulary |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105005789B (en) |
Families Citing this family (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105654122B (en) * | 2015-12-28 | 2018-11-16 | 江南大学 | Based on the matched spatial pyramid object identification method of kernel function |
| US11423651B2 (en) * | 2016-02-09 | 2022-08-23 | Hrl Laboratories, Llc | System and method for the fusion of bottom-up whole-image features and top-down enttiy classification for accurate image/video scene classification |
| CN106778494A (en) * | 2016-11-21 | 2017-05-31 | 河海大学 | A kind of target in hyperspectral remotely sensed image feature extracting method based on SIFT LPP |
| CN108427955B (en) * | 2017-10-27 | 2022-02-01 | 平安科技（深圳）有限公司 | Electronic device, chaotic sample sorting method, and computer-readable storage medium |
| CN108230240B (en) * | 2017-12-31 | 2020-07-31 | 厦门大学 | Method for obtaining position and posture in image city range based on deep learning |
| CN109325501B (en) * | 2018-08-14 | 2021-12-03 | 王斌 | Guitar backboard image-based material identification method and device and readable storage medium |
| CN111860615A (en) * | 2020-06-30 | 2020-10-30 | 河海大学 | Remote sensing water body target extraction method based on feature dictionary fusion |
| CN111563937B (en) * | 2020-07-14 | 2020-10-30 | 成都四方伟业软件股份有限公司 | Picture color extraction method and device |
| CN113012049B (en) * | 2021-04-15 | 2022-08-02 | 山东新一代信息产业技术研究院有限公司 | Remote sensing data privacy protection method based on GAN network |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101777125A (en) * | 2010-02-03 | 2010-07-14 | 武汉大学 | Method for supervising and classifying complex class of high-resolution remote sensing image |
| CN102208038A (en) * | 2011-06-27 | 2011-10-05 | 清华大学 | Image classification method based on visual dictionary |
| CN102622607A (en) * | 2012-02-24 | 2012-08-01 | 河海大学 | Remote sensing image classification method based on multi-feature fusion |
-
2015
- 2015-07-01 CN CN201510379234.6A patent/CN105005789B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101777125A (en) * | 2010-02-03 | 2010-07-14 | 武汉大学 | Method for supervising and classifying complex class of high-resolution remote sensing image |
| CN102208038A (en) * | 2011-06-27 | 2011-10-05 | 清华大学 | Image classification method based on visual dictionary |
| CN102622607A (en) * | 2012-02-24 | 2012-08-01 | 河海大学 | Remote sensing image classification method based on multi-feature fusion |
Non-Patent Citations (4)
| Title |
|---|
| Bag of spatial visual words model for scene classification;Wang Yuxin等;《Computer Science》;20111231;第265-268页 * |
| 一种基于多级空间视觉词典集体的图像分类方法;罗会兰等;《电子学报》;20150430;第684-692页 * |
| 基于BoW模型的图像分类方法研究;王莹;《万方数据》;20130426;第1-68页 * |
| 基于视觉词袋模型的遥感图像分类方法;周宇谷等;《重庆理工大学学报（自然科学）》;20150531;第71-77页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105005789A (en) | 2015-10-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105005789B (en) | A kind of remote sensing images terrain classification method of view-based access control model vocabulary | |
| CN105184309B (en) | Classification of Polarimetric SAR Image based on CNN and SVM | |
| CN105138970B (en) | Classification of Polarimetric SAR Image method based on spatial information | |
| CN103955702B (en) | SAR image terrain classification method based on depth RBF network | |
| CN103489005B (en) | A kind of Classification of High Resolution Satellite Images method based on multiple Classifiers Combination | |
| CN106355011B (en) | Geochemical data element sequence structure analysis method and device | |
| CN109255790A (en) | A kind of automatic image marking method of Weakly supervised semantic segmentation | |
| CN103971123B (en) | Hyperspectral image classification method based on linear regression Fisher discrimination dictionary learning (LRFDDL) | |
| CN107194433A (en) | A kind of Radar range profile's target identification method based on depth autoencoder network | |
| CN106650830A (en) | Deep model and shallow model decision fusion-based pulmonary nodule CT image automatic classification method | |
| CN105069478B (en) | High-spectrum remote-sensing terrain classification method based on super-pixel tensor sparse coding | |
| CN105631892B (en) | It is a kind of that detection method is damaged based on the aviation image building of shade and textural characteristics | |
| Liu et al. | Multimorphological superpixel model for hyperspectral image classification | |
| CN102982338B (en) | Classification of Polarimetric SAR Image method based on spectral clustering | |
| CN103839261A (en) | SAR image segmentation method based on decomposition evolution multi-objective optimization and FCM | |
| CN105427309A (en) | Multiscale hierarchical processing method for extracting object-oriented high-spatial resolution remote sensing information | |
| CN102005034A (en) | Remote sensing image segmentation method based on region clustering | |
| CN108256436A (en) | A kind of radar HRRP target identification methods based on joint classification | |
| CN111339947A (en) | Method and system for extracting remote sensing image fuzzy boundary ground object, storage medium and equipment | |
| CN107368852A (en) | A kind of Classification of Polarimetric SAR Image method based on non-down sampling contourlet DCGAN | |
| CN107067405A (en) | Based on the preferred Remote Sensing Image Segmentation of yardstick | |
| CN107292336A (en) | A kind of Classification of Polarimetric SAR Image method based on DCGAN | |
| CN105138966B (en) | Classification of Polarimetric SAR Image method based on fast density peak value cluster | |
| CN110060273A (en) | Remote sensing image landslide plotting method based on deep neural network | |
| CN104156945A (en) | Method for segmenting gray scale image based on multi-objective particle swarm optimization algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |