CN107748875A - A kind of earthquake building recognition method based on multidate radar image texture feature - Google Patents

A kind of earthquake building recognition method based on multidate radar image texture feature Download PDF

Info

Publication number
CN107748875A
CN107748875A CN201711072305.3A CN201711072305A CN107748875A CN 107748875 A CN107748875 A CN 107748875A CN 201711072305 A CN201711072305 A CN 201711072305A CN 107748875 A CN107748875 A CN 107748875A
Authority
CN
China
Prior art keywords
mrow
shake
sar image
image
feature
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
Application number
CN201711072305.3A
Other languages
Chinese (zh)
Inventor
李强
张景发
龚丽霞
薛腾飞
蒋洪波
罗毅
焦其松
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Institute of Crustal Dynamics of China Earthquake Administration
Original Assignee
Institute of Crustal Dynamics of China Earthquake Administration
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Institute of Crustal Dynamics of China Earthquake Administration filed Critical Institute of Crustal Dynamics of China Earthquake Administration
Priority to CN201711072305.3A priority Critical patent/CN107748875A/en
Publication of CN107748875A publication Critical patent/CN107748875A/en
Pending legal-status Critical Current

Links

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
    • G06V20/176—Urban or other man-made structures

Abstract

The invention belongs to SAR image identification technology field, more particularly to a kind of earthquake building recognition method based on multidate radar image texture feature.Earthquake building recognition method based on multidate radar image texture feature, it comprises the following steps：A. SAR image I before shake and SAR image I' after shake are established into corresponding relation；B. main texture component calculates；C. feature correlation is calculated and analyzed；D. relevance threshold statistics and classification.Present invention fusion correlation analysis and textural characteristics principal component component, realize the automation extraction of earthquake building, extraction accuracy meets the needs of earthquake emergency is with assessing, and extraction result can be the information foundation that the distribution of earthquake emergency rescue strength, crop loss rate, post-disaster reconstruction etc. provide science.

Description

A kind of earthquake building recognition method based on multidate radar image texture feature
Technical field
It is more particularly to a kind of to be based on multidate radar image texture feature the invention belongs to SAR image identification technology field Earthquake building recognition method.
Background technology
Existing multidate SAR image change detection is based primarily upon the intensity image of SAR image, different phase intensity Coherent speckle noise intrinsic easily in by SAR image and earth's surface object variations are influenceed between image, cause to change testing result not The change of true atural object can be reflected completely.However, the intact building distribution rule in cities and towns is orderly, generally in shapes such as wall, ground Into substantial amounts of dihedral angle structure, the high point of brightness is shown as in SAR image, textural characteristics are obvious.In SAR image forming process In, dihedral angle structure formed corner reflector influenceed by incidence angle and coherent speckle noise it is less, textural characteristics performance it is relatively steady It is fixed so that the change detection based on textural characteristics is possibly realized.Simultaneously in the environment of earthquake region, earthquake building it is more prominent be Spatial texture feature rather than isolated pixel grey scale information, this also causes texture analysis to turn into the important side of earthquake image interpretation Method, the existing method based on textural characteristics change detection are typically the textural characteristics using single channel, and unrealized multiple features melt Close with effectively utilizing, cause the loss of potential available information.
The content of the invention
It is an object of the invention to：Easily influenceed for intensity image by speckle noise, feature changes, it is special using single texture Sign easily causes the problem of available information is lost, there is provided a kind of earthquake building based on multidate radar image texture feature is known Other method.
The technical scheme is that：Earthquake building recognition method based on multidate radar image texture feature, it Comprise the following steps：
A. SAR image I before shake and SAR image I' after shake are established into corresponding relation；
B. main texture component calculates；
Main texture component calculates the principal component component for referring to that acquisition multi-texturing feature is analyzed by principal component analysis；SAR Image has abundant textural characteristics, and textural characteristics can describe the spatial distribution state and degree of roughness of earth's surface, is to differentiate atural object Important symbol；In analysis of texture method, statistical analysis method is one of the most frequently used method, and this method is certain by setting The statistical analysis of window size unit obtains textural characteristics, and the characteristic parameter of acquisition mainly describes the random and empty of texture local mode Between statistical nature, so as to for expressing the uniformity in region and interregional diversity；
B1. it is special to obtain SAR image I and two or more textures in SAR image I' after shake before shaking for statistical analysis Levy parameter；
B2. by two or more textural characteristics parameter combinations in SAR image I before shake, multi-texturing feature is calculated Principal component；During principal component is obtained, it is not necessary to parameter setting and optimization demand, information in the principal component obtained by Level is reduced, and the first factor is contained comprising the former information above amount of textural characteristics parameter 90%, therefore after Selective principal component analysis The first factor as shake before SAR image I the first factor；
SAR image I' the first factor after similarly being shaken；
C. feature correlation is calculated and analyzed；
Setup algorithm window calculation obtains the dependency graph of SAR image I and SAR image I' the first factors after shake before shaking, The value of each pixel represents correlation coefficient r in dependency graph；Built in the big I reflection BEFORE AND AFTER EARTHQUAKE image of coefficient correlation The degree that object area changes；
In formula：M and n is SAR image I before shake and SAR image I' calculation window sizes after shake, takes odd number；Yls, XlsRespectively For the characteristic value of the pixels corresponding with SAR image I' the first factors after shake of SAR image I before shake；SAR before shaking is represented respectively Image I and SAR image I' the first factors after shake provincial characteristics average value；
Correlation coefficient r takes absolute value, and span is [0,1], and r absolute value is closer to 1, the linear relationship of two images Closer, i.e., similarity degree is higher, and building destruction degree is smaller after illustrating shake；Closer to 0, the linear relationship of two images is got over Difference, i.e. otherness are bigger, illustrate that the destructiveness of building after shaking is more serious；
D. relevance threshold statistics and classification；
D1. the building sample point of different destructiveness is randomly selected in the optical remote sensing image obtained after shake, according to The probability distribution of samples points, count the size of the coefficient correlation of sample point position in dependency graph；
D2. the destructiveness of building sample point in being interpreted based on optical image, the sample point phase relation obtained to statistics Number size is classified according to destructiveness, so as to obtain different Earthquake hazard degree building coefficient correlation classification thresholds scopes.
Further, after obtaining coefficient correlation classification thresholds scope using step D2, by optical remote sensing image after shake first The dependency graph picture of component is divided into three classes, and then obtains the spatial distribution of different Earthquake hazard degree buildings.
Beneficial effect：Present invention fusion correlation analysis and textural characteristics principal component component, realize earthquake building oneself Dynamicization is extracted, and extraction accuracy meets the needs of earthquake emergency is with assessing, compared to traditional SAR image earthquake building recognition side Method, this method accuracy of identification increase significantly, after extraction result can be for the distribution of earthquake emergency rescue strength, crop loss rate, calamity Reconstruction etc. provides the information foundation of science.
Brief description of the drawings
Fig. 1 is the techniqueflow chart of the present invention；
Fig. 2 is earthquake building recognition result figure in the embodiment of the present invention 2.
Embodiment
Following examples are used to illustrate the present invention, but are not limited to the scope of the present invention.
Embodiment 1
Referring to accompanying drawing 1, the earthquake building recognition method based on multidate radar image texture feature, it includes following step Suddenly：
A. SAR image I before shake and SAR image I' after shake are established into corresponding relation；In this example, using based on SIFT feature Image matching method, establish shake before SAR image I with shake after SAR image I' corresponding relation；
B. main texture component calculates；
Main texture component calculates the principal component component for referring to that acquisition multi-texturing feature is analyzed by principal component analysis；SAR Image has abundant textural characteristics, and textural characteristics can describe the spatial distribution state and degree of roughness of earth's surface, is to differentiate atural object Important symbol；In analysis of texture method, statistical analysis method is one of the most frequently used method, and this method is certain by setting The statistical analysis of window size unit obtains textural characteristics, and the characteristic parameter of acquisition mainly describes the random and empty of texture local mode Between statistical nature, so as to for expressing the uniformity in region and interregional diversity；
B1. it is special to obtain SAR image I and two or more textures in SAR image I' after shake before shaking for statistical analysis Levy parameter；
Analysis of texture method based on gray level co-occurrence matrixes is one of method classical in statistical analysis technique, this example It is middle using this method come calculate obtain textural characteristics parameter, textural characteristics parameter include average, variance, contrast, entropy, homogeneity, Distinctiveness ratio, correlation and angular second moment；
B2. by two or more textural characteristics parameter combinations in SAR image I before shake, multi-texturing feature is calculated Principal component；
In this example, the principal component of multi-texturing feature is calculated using principal component analytical method, principal component analysis is a kind of letter Single non-parametric method, linear projection conversion can be carried out to primitive image features, eliminate figure on the premise of maximum fault information is retained As the correlation between internal each passage, to suppress noise, prominent main information, its central idea is to retain as much as possible The dimension of data set is reduced while data set variance, increases the required characteristic information of information identification to a certain extent and carries High information recognition efficiency；During principal component is obtained, it is not necessary to which parameter setting is with optimization demand, in the principal component obtained Information is reduced step by step, and the first factor is contained comprising the former information above amount of textural characteristics parameter 90%, therefore selects principal component The first factor of the first factor as SAR image I before shake after analysis；
SAR image I' the first factor after similarly being shaken；
C. feature correlation is calculated and analyzed；
Setup algorithm window calculation obtains the dependency graph of SAR image I and SAR image I' the first factors after shake before shaking, The value of each pixel represents correlation coefficient r in dependency graph；Built in the big I reflection BEFORE AND AFTER EARTHQUAKE image of coefficient correlation The degree that object area changes；
In formula：M and n is SAR image I before shake and SAR image I' calculation window sizes after shake, takes the value of odd number, m and n It is relevant with image resolution, typically take 3,5 or 7；
Yls, XlsThe characteristic value of SAR image I pixels corresponding with SAR image I' the first factors after shake before respectively shaking；The provincial characteristics average value of SAR image I and SAR image I' the first factors after shake before shaking are represented respectively；
Correlation coefficient r takes absolute value, and span is [0,1], and r absolute value is closer to 1, the linear relationship of two images Closer, i.e., similarity degree is higher, and building destruction degree is smaller after illustrating shake；Closer to 0, the linear relationship of two images is got over Difference, i.e. otherness are bigger, illustrate that the destructiveness of building after shaking is more serious；
D. relevance threshold statistics and classification；
D1. visual interpretation is carried out to the optical remote sensing image obtained after shake, difference is randomly selected in optical remote sensing image The building sample point of destructiveness, according to the probability distribution of samples points, count the phase relation of the sample point position in dependency graph Several sizes；
D2. the destructiveness of building sample point in being interpreted based on optical image, the sample point phase relation obtained to statistics Number size is classified according to destructiveness, so as to obtain different Earthquake hazard degree building coefficient correlation classification thresholds scopes.
Further, after obtaining coefficient correlation classification thresholds scope using step D2, by optical remote sensing image after shake first The dependency graph picture of component is divided into three classes, i.e., substantially intact, moderate damage and damage, and then obtains different Earthquake hazard degree buildings Spatial distribution.
Embodiment 2
Earthquake building recognition is carried out using the method described in embodiment 1.
The data that this example uses are as shown in table 1 for single polarization ALOS-2 satellite datas, data basic condition.
The data basic condition of table 1
During analysis of texture, the quantization of gray level co-occurrence matrixes and selected direction, step-length, window size and image Grade is about, it is necessary to select the parameter of generation gray level co-occurrence matrixes according to specific image texture feature.Texture analysis be Realized in remote sensing image processing software ENVI.
After obtaining 8 textural characteristics, principal component analysis is carried out, principal component analysis is in remote sensing image processing software Realized in ENVI.Because of follow-up principal component component, to cover information content almost nil, therefore sets 4 principal component components of generation, becomes Change rear each band class information amount statistics and be shown in Table 2.The first factor contains 96.39% characteristic information, in first principal component component The main information of earthquake damage characteristics can be characterized by covering in image, therefore, select the first factor to participate in analysis in this example.
Each principal component band class information amount statistical form after the principal component transform of table 1
Multidate SAR image feature correlation is calculated and analyzed；The calculation window that correlation analysis uses is set as 3, Afterwards, the dependency graph picture of acquisition is taken absolute value in ENVI and its value is set as in the range of [0,1].
In this example, according to investigation image, 1102 sample pixels (including substantially intact building 427 is randomly selected Individual pixel, 398 pixels of moderate damage building, 277 pixels of damage building), count different damage degree building samples The eigenvalue threshold scope of this corresponding image pixel, the threshold value distribution of statistics are as shown in table 3.It is distributed afterwards according to threshold value Scope, dependency graph picture is divided into three classes, classification results are as shown in Figure 2
The class method difference Earthquake hazard degree building classification thresholds of table 3 three
Embodiment 3
On the basis of the on-site inspection result issued by the technical policy comprehensive study of Japanese traffic province territory, count respectively not Number is distributed with earthquake building pixel, builds confusion matrix, calculates the overall accuracy (OA) of building recognition.Computational methods are such as Under：OA=TP/Total, wherein TP represent the pixel summation correctly classified, total pixel number of Total presentation class images.
(multidate SAR image difference operation, utilizes identical building to this method with intensity image differential technique change detection Sample carrys out statistical threshold and obtains different damage degree building classification thresholds), intensity image correlation change detecting method is (when more Phase SAR image intensity dependence is analyzed, and different damage degree buildings are obtained come statistical threshold using identical building sample Classification thresholds), (single texture template image correlation analysis, utilization are identical for single textural characteristics correlation change detecting method Building sample carry out statistical threshold and obtain different damage degree building classification thresholds) compare, overall extraction accuracy carries respectively It is high by 7.7%, 5.7% and 8.9%.
Although above with general explanation and specific embodiment, the present invention is described in detail, at this On the basis of invention, it can be made some modifications or improvements, this will be apparent to those skilled in the art.Therefore, These modifications or improvements without departing from theon the basis of the spirit of the present invention, belong to the scope of protection of present invention.

Claims (8)

  1. A kind of 1. earthquake building recognition method based on multidate radar image texture feature, it is characterised in that：The identification Method comprises the following steps：
    A. SAR image I before shake and SAR image I' after shake are established into corresponding relation；
    B. main texture component calculates；
    B1. statistical analysis obtains SAR image I and two or more textural characteristics ginseng in SAR image I' after shake before shaking Number；
    B2. by two or more textural characteristics parameter combinations in SAR image I before shake, the master of calculating multi-texturing feature Component, first master point of principal component of the selection comprising the former information above amount of textural characteristics parameter 90% as SAR image I before shake Amount；
    SAR image I' the first factor after similarly being shaken；
    C. feature correlation is calculated and analyzed；
    Setup algorithm window calculation obtains the dependency graph of SAR image I and SAR image I' the first factors after shake before shaking, related The value of each pixel represents correlation coefficient r in property figure；
    <mrow> <mi>r</mi> <mo>=</mo> <mfrac> <mrow> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>i</mi> <mo>=</mo> <mn>0</mn> </mrow> <mrow> <mi>m</mi> <mo>,</mo> <mi>n</mi> </mrow> </munderover> <mrow> <mo>(</mo> <msub> <mi>X</mi> <mrow> <mi>l</mi> <mo>,</mo> <mi>s</mi> </mrow> </msub> <mo>-</mo> <mover> <mi>X</mi> <mo>&amp;OverBar;</mo> </mover> <mo>)</mo> </mrow> <mrow> <mo>(</mo> <msub> <mi>Y</mi> <mrow> <mi>l</mi> <mo>,</mo> <mi>s</mi> </mrow> </msub> <mo>-</mo> <mover> <mi>Y</mi> <mo>&amp;OverBar;</mo> </mover> <mo>)</mo> </mrow> </mrow> <mrow> <msqrt> <mrow> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>i</mi> <mo>=</mo> <mn>0</mn> </mrow> <mrow> <mi>m</mi> <mo>,</mo> <mi>n</mi> </mrow> </munderover> <msup> <mrow> <mo>(</mo> <msub> <mi>X</mi> <mrow> <mi>l</mi> <mo>,</mo> <mi>s</mi> </mrow> </msub> <mo>-</mo> <mover> <mi>X</mi> <mo>&amp;OverBar;</mo> </mover> <mo>)</mo> </mrow> <mn>2</mn> </msup> </mrow> </msqrt> <mo>&amp;CenterDot;</mo> <msqrt> <mrow> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>i</mi> <mo>=</mo> <mn>0</mn> </mrow> <mrow> <mi>m</mi> <mo>,</mo> <mi>n</mi> </mrow> </munderover> <msup> <mrow> <mo>(</mo> <msub> <mi>Y</mi> <mrow> <mi>l</mi> <mo>,</mo> <mi>s</mi> </mrow> </msub> <mo>-</mo> <mover> <mi>Y</mi> <mo>&amp;OverBar;</mo> </mover> <mo>)</mo> </mrow> <mn>2</mn> </msup> </mrow> </msqrt> </mrow> </mfrac> </mrow>
    In formula：M and n is SAR image I before shake and SAR image I' calculation window sizes after shake, takes odd number；Yls, XlsBefore respectively shaking The characteristic value of SAR image I pixels corresponding with SAR image I' the first factors after shake；SAR image I before shaking is represented respectively With the provincial characteristics average value of SAR image I' the first factors after shake；
    Correlation coefficient r takes absolute value, and span is [0,1], and for r absolute value closer to 1, the linear relationship of two images is closeer Cut, i.e., similarity degree is higher, and building destruction degree is smaller after illustrating shake；Closer 0, the linear relationship of two images is poorer, I.e. otherness is bigger, illustrates that the destructiveness of building after shaking is more serious；
    D. relevance threshold statistics and classification；
    D1. the building sample point of different destructiveness is randomly selected in the optical remote sensing image obtained after shake, according to sample Point distribution, count the size of the coefficient correlation of sample point position in dependency graph；
    D2. the destructiveness of building sample point in being interpreted based on optical image, the sample Point correlation coefficient obtained to statistics are big It is small to be classified according to destructiveness, so as to obtain different Earthquake hazard degree building coefficient correlation classification thresholds scopes.
  2. 2. the earthquake building recognition method based on multidate radar image texture feature as claimed in claim 1, its feature It is：After obtaining coefficient correlation classification thresholds scope using step D2, by the correlation of the component of optical remote sensing image after shake first Image is divided into three classes, and then obtains the spatial distribution of different Earthquake hazard degree buildings.
  3. 3. the earthquake building recognition method based on multidate radar image texture feature as claimed in claim 1, its feature It is：In step A, using the image matching method based on SIFT feature, SAR image I and SAR image I' after shake before shake is established Corresponding relation.
  4. 4. the earthquake building recognition method based on multidate radar image texture feature as claimed in claim 1, its feature It is：In step B1, textural characteristics parameter is obtained using the analysis of texture method based on gray level co-occurrence matrixes.
  5. 5. the earthquake building recognition method based on multidate radar image texture feature as claimed in claim 1, its feature It is：Textural characteristics parameter in step B1 includes：Average, variance, contrast, entropy, homogeneity, distinctiveness ratio, correlation and angle two Rank square.
  6. 6. the earthquake building recognition method based on multidate radar image texture feature as claimed in claim 1, its feature It is：In step B2, the principal component of multi-texturing feature is calculated using principal component analytical method.
  7. 7. the earthquake building recognition method based on multidate radar image texture feature as claimed in claim 1, its feature It is：In step D1, by carrying out visual interpretation, sample drawn point to the optical remote sensing image obtained after shake.
  8. 8. the earthquake building recognition method based on multidate radar image texture feature as claimed in claim 1, its feature It is：In step C, m and n value are 3,5 or 7.
CN201711072305.3A 2017-11-03 2017-11-03 A kind of earthquake building recognition method based on multidate radar image texture feature Pending CN107748875A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201711072305.3A CN107748875A (en) 2017-11-03 2017-11-03 A kind of earthquake building recognition method based on multidate radar image texture feature

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201711072305.3A CN107748875A (en) 2017-11-03 2017-11-03 A kind of earthquake building recognition method based on multidate radar image texture feature

Publications (1)

Publication Number Publication Date
CN107748875A true CN107748875A (en) 2018-03-02

Family

ID=61253573

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201711072305.3A Pending CN107748875A (en) 2017-11-03 2017-11-03 A kind of earthquake building recognition method based on multidate radar image texture feature

Country Status (1)

Country Link
CN (1) CN107748875A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109543616A (en) * 2018-11-23 2019-03-29 北京师范大学 Damage assessment method, apparatus, equipment and the medium of target material object after a kind of shake
CN111608410A (en) * 2020-04-29 2020-09-01 湖南南派古建园林工程有限公司 In-situ repairing method and system for historic building timber structure

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102004915A (en) * 2010-10-19 2011-04-06 中国科学院遥感应用研究所 Earthquake-damaged construction remote sensing quick extraction technique based on textural feature space
CN103761740A (en) * 2014-01-23 2014-04-30 武汉大学 Construction damage assessment method based on single post-earthquake POLSAR image
CN104268879A (en) * 2014-09-28 2015-01-07 民政部国家减灾中心 Physical building quantity damage evaluation method based on remote sensing multi-spectral images
JP2016507735A (en) * 2012-12-28 2016-03-10 ユニバーシティー オブ ソウル インダストリー コーポレーション ファンデーション Ion distortion correction method and apparatus for satellite radar interference degree
CN105701481A (en) * 2016-02-26 2016-06-22 民政部国家减灾中心 Collapsed building extraction method

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102004915A (en) * 2010-10-19 2011-04-06 中国科学院遥感应用研究所 Earthquake-damaged construction remote sensing quick extraction technique based on textural feature space
JP2016507735A (en) * 2012-12-28 2016-03-10 ユニバーシティー オブ ソウル インダストリー コーポレーション ファンデーション Ion distortion correction method and apparatus for satellite radar interference degree
CN103761740A (en) * 2014-01-23 2014-04-30 武汉大学 Construction damage assessment method based on single post-earthquake POLSAR image
CN104268879A (en) * 2014-09-28 2015-01-07 民政部国家减灾中心 Physical building quantity damage evaluation method based on remote sensing multi-spectral images
CN105701481A (en) * 2016-02-26 2016-06-22 民政部国家减灾中心 Collapsed building extraction method

Non-Patent Citations (7)

* Cited by examiner, † Cited by third party
Title
张晓斌 等: "《工程地质与水文地质 第2版》", 31 July 2016 *
李强 等: "SAR图像纹理特征相关变化检测的震害建筑物提取", 《遥感学报》 *
窦爱霞 等: "SAR图像在震害变化检测中的应用", 《万方数据知识服务平台》 *
薛腾飞 等: "基于相关变化检测与面向对象分类技术的多源遥感图像震害信息提取", 《地震学报》 *
赫晓慧 等: "《遥感基础导论》", 31 May 2016 *
陈文凯: "面向震害评估的遥感应用技术研究", 《中国优秀硕士学位论文全文数据库基础科学辑》 *
靳国旺: "《雷达摄影测量》", 30 April 2015 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109543616A (en) * 2018-11-23 2019-03-29 北京师范大学 Damage assessment method, apparatus, equipment and the medium of target material object after a kind of shake
CN111608410A (en) * 2020-04-29 2020-09-01 湖南南派古建园林工程有限公司 In-situ repairing method and system for historic building timber structure

Similar Documents

Publication Publication Date Title
CN110136170B (en) Remote sensing image building change detection method based on convolutional neural network
CN112668494A (en) Small sample change detection method based on multi-scale feature extraction
CN109934154B (en) Remote sensing image change detection method and detection device
CN106446936B (en) Empty spectrum joint data based on convolutional neural networks turn the Hyperspectral data classification method of waveform diagram
Shi et al. Building collapse assessment by the use of postearthquake Chinese VHR airborne SAR
CN103226826B (en) Based on the method for detecting change of remote sensing image of local entropy visual attention model
CN112287983B (en) Remote sensing image target extraction system and method based on deep learning
Ming et al. Modified ALV for selecting the optimal spatial resolution and its scale effect on image classification accuracy
CN109726649B (en) Remote sensing image cloud detection method and system and electronic equipment
Rathje et al. Damage patterns from satellite images of the 2003 Bam, Iran, earthquake
Ghasemi et al. Urban classification using preserved information of high dimensional textural features of Sentinel-1 images in Tabriz, Iran
CN107748875A (en) A kind of earthquake building recognition method based on multidate radar image texture feature
Chen et al. Scene segmentation of remotely sensed images with data augmentation using U-net++
CN106971402B (en) SAR image change detection method based on optical assistance
JP6509887B2 (en) Image change detection
CN102938148A (en) High-spectrum image texture analysis method based on V-GLCM (Gray Level Co-occurrence Matrix)
CN112966656A (en) Data processing method and device
CN116363516A (en) Remote sensing image change detection method based on edge auxiliary self-adaption
CN109934103A (en) Method based on obvious object in dark channel prior and region covariance detection image
Yuval et al. Assessment of storm impact on coral reef structural complexity
CN104616302A (en) Real-time object identification method
CN114283336A (en) Anchor-frame-free remote sensing image small target detection method based on mixed attention
Yang et al. Improving Semantic Segmentation Performance by Jointly Using High Resolution Remote Sensing Image and Ndsm
Xue et al. A parallel fusion method of remote sensing image based on NSCT
CN104615988A (en) Picture identification method

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