CN106056044B - A kind of high spectrum image emissivity spectra extracting method based on end member abundance inverting - Google Patents

A kind of high spectrum image emissivity spectra extracting method based on end member abundance inverting Download PDF

Info

Publication number
CN106056044B
CN106056044B CN201610340061.1A CN201610340061A CN106056044B CN 106056044 B CN106056044 B CN 106056044B CN 201610340061 A CN201610340061 A CN 201610340061A CN 106056044 B CN106056044 B CN 106056044B
Authority
CN
China
Prior art keywords
end member
high spectrum
image
pixel
curve
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
Application number
CN201610340061.1A
Other languages
Chinese (zh)
Other versions
CN106056044A (en
Inventor
刘畅
李军伟
王广平
何茜
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.)
Beijing Institute of Environmental Features
Original Assignee
Beijing Institute of Environmental Features
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 Beijing Institute of Environmental Features filed Critical Beijing Institute of Environmental Features
Priority to CN201610340061.1A priority Critical patent/CN106056044B/en
Publication of CN106056044A publication Critical patent/CN106056044A/en
Application granted granted Critical
Publication of CN106056044B publication Critical patent/CN106056044B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/08—Feature extraction
    • G06F2218/10—Feature extraction by analysing the shape of a waveform, e.g. extracting parameters relating to peaks

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Theoretical Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Signal Processing (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Radiation Pyrometers (AREA)

Abstract

The present invention provides a kind of high spectrum image emissivity spectra extracting method based on end member abundance inverting, Decomposition of Mixed Pixels operation is carried out to whole picture infrared high spectrum image first, obtain Abundances corresponding to the pure end member and each end member in image, and then obtain the spoke luminance spectral curve of each pixel point in image, then to interesting target spoke luminance spectral march line integral within the scope of the temperature threshold of setting, the absolute black body temperature of infrared high spectrum emissivity curve simulation is estimated according to integral area, and then acquires emissivity spectra curve.The present invention is when carrying out the extraction of interesting target spoke brightness curve, consider the influence of mixed pixel, the true spoke brightness curve of target has been obtained by Endmember extraction and abundance inverting, spoke brightness curve extraction accuracy has been improved, so that the extraction accuracy of emissivity curve is improved.

Description

A kind of high spectrum image emissivity spectra extracting method based on end member abundance inverting
Technical field
The invention belongs to infrared high spectrum target properties and target identification technology field, and in particular to one kind is rich based on end member Spend the high spectrum image emissivity spectra extracting method of inverting.
Background technique
Infrared high spectrum remote sensing is the new technology that recent decades grow up.In infrared high spectrum remote sensing, primary work Work is inverting temperature and emissivity.Emissivity is the physical quantity for characterizing object to ambient radiation ability size, object emission Rate spectrum is related with its ingredient and structure, therefore, in many applications, can be identified by the shape of emissivity spectra curve Object.And the extraction of emissivity spectra feature has to the separation (TES) that temperature and emissivity are carried out to infrared high spectrum image.
Temperature and emissivity separation method generally refer to seek temperature and emissivity using a phase of infrared high spectrum Wave spectrum.The key problem of this method is: in the case where emissivity is totally unknown, the observation of N number of wave band has N number of data, still Also there is N number of unknown number accordingly, along with target temperature is unknown, just has N+1 unknown number, this is just a underdetermined equation always Problem.In order to solve this underdetermined equation, it is necessary to introduce additional condition.This additional condition is usually to backs Certain priori knowledge of spectral shape also just determines different algorithm types according to the difference of Prior Knowledge Constraints condition.Temperature The solution procedure of degree and emissivity separation algorithm is exactly the process of emissivity spectra feature extraction.Common method has: envelope Method；Reference channel method (REF)；Emissivity normalization method (NEM)；α residual basis (ADE) etc..
Ground object emission spectrum signal acquired in infrared high spectrum imaging instrument is recorded as unit of pixel.Due to The limitation of sensor spatial resolution and the complexity of background, so that the generally existing mixed pixel of high spectrum image.Such as The application studies such as fruit classified using mixed pixel as pure pixel, target detection and identification, as a result have very big error.This Just infrared high spectrum Decomposition of Mixed Pixels problem is made to become an important and key link of infrared high Spectrum data processing.It is mixed Closing pixel analysis to need the work that carries out is exactly that Decomposition of Mixed Pixels for end member and is found out the ratio shared by it inside pixel Example, makes image analysis enter sub-pixed mapping rank.
Traditional emissivity extraction algorithm is when carrying out the extraction of interesting target spoke brightness curve, usually using in the picture The method of straight access point carries out.It is at this moment extracted interested due to mixed pixel generally existing in infrared high spectrum image Target spoke brightness curve is probably a mixed pixel, if using the mixed pixel spoke luminance spectral as the true of the target Real spectrum continues subsequent transmission rate spectrum extraction operation, as a result in will necessarily introduce certain error.
Summary of the invention
It is an object of the present invention to: in view of the above-mentioned drawbacks of the prior art, proposing a kind of based on end member abundance inverting High spectrum image emissivity spectra extracting method solves carrying out introducing when temperature is separated with emissivity to infrared high spectrum image The problem of mixed pixel, has obtained the true spoke brightness curve of target by Endmember extraction and abundance inverting, and it is bent to improve spoke brightness Line drawing precision, so that the extraction accuracy of emissivity curve is improved.
A kind of high spectrum image emissivity spectra extracting method based on end member abundance inverting provided by the invention is improved Place is that method includes the following steps:
1) to whole picture infrared high spectrum image carry out Decomposition of Mixed Pixels, obtain infrared high spectrum image end member and each end The corresponding Abundances of member；
2) interesting target is determined；
3) according to the pure end member and the Abundances, the interesting target is extracted, obtains infrared high spectrum image In each pixel point spoke luminance spectral curve；
4) set temperature threshold range；
5) to the spoke luminance spectral curve march line integral of the interesting target；
6) the absolute black body temperature of infrared high spectrum emissivity curve simulation is estimated according to integral area；
7) emissivity spectra curve is obtained, and is normalized.
Preferably, step 1) Decomposition of Mixed Pixels includes Endmember extraction step and abundance inversion step, in which:
Endmember extraction step is extracted using the infrared high spectrum end member extraction method based on extended mathematical morphology；
Abundance inversion step carries out abundance inverting using linear solution mixing method.
More preferably, Endmember extraction step includes:
1. reading in infrared high spectrum original image, the least structure element K is utilizedmin, to target EO-1 hyperion in structural element Each pixel of image carries out expansion based on mass center in neighborhood and corrosion calculates, find most pure pixel d (x, y, w) and The pixel e (x, y, w) of most serious is mixed, and calculates corresponding MEI value at this time；
MEI indicates morphology eccentricity index, is defined as:
MEI (x, y, w)=dist (d (x, y, w), e (x, y, w)), w=1,2 ..., n
Wherein, dist indicates the distance between two o'clock, and d (x, y, w) indicates pixel most pure in image, e (x, y, w) table Show the pixel of mixing most serious；
2. the step of being continuously increased the size of structural element, repeating step 1., until reaching max architecture element Kmax；Instead MEI value is calculated again and is noted down, and the corresponding each MEI value of identical pixel is averaged, final MEI image is obtained；
3. carrying out threshold decision, the gray value of obtained MEI image all the points is taken into the average threshold value as setting, Pixel in MEI image greater than this threshold value will be marked as Pure pixel；
4. carrying out image segmentation to the end member image finally obtained and region growing completing automatically selecting for end member；
5. carrying out linear spectral separation using linear solution mixing method obtains end member abundance figure.
More preferably, the expression formula of the spoke luminance spectral curve of step 3) each pixel point is calculated are as follows:
Wherein, ejIndicate each end member in image, αijIndicating each end member proportion in the picture, E indicates end member matrix, αiIndicate abundance matrix.
More preferably, the expression formula of step 6) the absolute black body temperature is calculated are as follows:
Wherein: T1,T2For the initial temperature threshold of setting；L1,L2Respectively T1,T2Corresponding black matrix spoke brightness；LVFor this Curve of spectrum integral.
More preferably, the expression formula of step 7) the emissivity spectra curve is calculated are as follows:
Bυ,T=2hc2υ3(echυ/kT-1)-1
Wherein: c is the light velocity；H is planck constant, h=6.6262 × 10-34J·S；K is Boltzmann constant, k= 1.3806×10-23J/K。
More preferably, the expression formula that emissivity spectra curve is normalized in step 7) are as follows:
The present invention is solved by establishing a kind of high spectrum image emissivity spectra extraction algorithm based on end member abundance inverting Is carried out to infrared high spectrum image, in progress interesting target spoke the problem of introducing mixed pixel when temperature is separated with emissivity When brightness curve extracts, it is contemplated that it is bright to have obtained the true spoke of target by Endmember extraction and abundance inverting for the influence of mixed pixel It writes music line, improves spoke brightness curve extraction accuracy, so that the extraction accuracy of emissivity curve is improved, for high-precision Emissivity spectra provide strong support.
Detailed description of the invention
Fig. 1 is the stream of the high spectrum image emissivity spectra extraction algorithm based on end member abundance inverting of the embodiment of the present invention Cheng Tu；
Fig. 2 is the infrared high spectrum end member extraction method flow chart based on extended mathematical morphology of the embodiment of the present invention.
Specific embodiment
To make the objectives, technical solutions, and advantages of the present invention more comprehensible, referring to the drawings to the present invention into one Step is described in detail.However, it is necessary to illustrate, many details listed in specification are used for the purpose of making reader to of the invention One or more aspects have a thorough explanation, these sides of the invention also may be implemented even without these specific details Face.
Due to the limitation of sensor spatial resolution and the complexity of background, so that high spectrum image is generally existing Mixed pixel.If classified using mixed pixel as pure pixel, the application studies such as target detection and identification, as a result have very Big error.
A kind of high spectrum image emissivity spectra extracting method based on end member abundance inverting proposed by the present invention, flow chart As shown in Figure 1, the specific steps are as follows:
1) Decomposition of Mixed Pixels, including Endmember extraction step and abundance inverting step are carried out to whole picture infrared high spectrum image Suddenly, the end member (i.e. Pure pixel) and the corresponding Abundances of each end member of infrared high spectrum image are obtained then；
It is specific in the present invention that end member is carried out using the infrared high spectrum end member extraction method based on extended mathematical morphology It extracts, then carries out abundance inverting using linear solution mixing method, as shown in Figure 2, the specific steps are as follows:
1. reading in infrared high spectrum original image, the least structure element K is utilizedmin, to target EO-1 hyperion in structural element Each pixel of image carries out expansion and erosion operation based on mass center in neighborhood, find most pure pixel d (x, y, w) and The pixel e (x, y, w) of most serious is mixed, and calculates corresponding MEI value at this time.Wherein, MEI indicates morphology eccentricity index, Is defined as:
MEI (x, y, w)=dist (d (x, y, w), e (x, y, w)), w=1,2 ..., n
Wherein, dist indicates the distance between two o'clock, and d (x, y, w) indicates pixel most pure in image, e (x, y, w) table Show the pixel of mixing most serious.
2. the step of being continuously increased the size of structural element, repeating step 1, until reaching max architecture element Kmax.Simultaneously MEI value is calculated repeatedly and is noted down, and the corresponding each MEI value of identical pixel is averaged, final MEI image is obtained.
3. carrying out threshold decision, the gray value of obtained MEI image all the points is taken into the average threshold value as setting, Pixel in MEI image greater than this threshold value will be marked as end member.
4. carrying out image segmentation to finally obtained end member image and region growing being completed end member and automatically selected.
5. carrying out linear spectral separation using linear solution mixing method obtains end member abundance figure.
2) interesting target is determined；
3) according to the end member and the Abundances, the interesting target is extracted, is obtained every in infrared high spectrum image The spoke luminance spectral curve of 1 pixel point.The expression formula of spoke luminance spectral curve are as follows:
Wherein, ejIndicate each end member in image, αijIndicating each end member proportion in the picture, E indicates end member matrix, αiIndicate abundance matrix.
4) set temperature threshold range；
5) to the spoke luminance spectral curve march line integral of the interesting target；
6) the absolute black body temperature of infrared high spectrum emissivity curve simulation, expression formula are estimated according to integral area Are as follows:
Wherein: T1,T2For the initial temperature threshold of setting；L1,L2Respectively T1,T2Corresponding black matrix spoke brightness；LVFor this Curve of spectrum integral.
7) emissivity spectra curve is obtained, and is normalized.Wherein:
Using Planck law, the simulation absolute black body temperature spoke brightness of the point changes with wave number (ν) on acquisition image Spectral emissivity curve:
Bυ,T=2hc2υ3(echυ/kT-1)-1
In above-mentioned expression formula: c is the light velocity；H is planck constant, h=6.6262 × 10-34J·S；K is that Boltzmann is normal Number, k=1.3806 × 10-23J/K。
The expression formula that emissivity spectra curve is normalized are as follows:
The above is only a preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art For member, without departing from the principle of the present invention, it can also make several improvements and retouch, these improvements and modifications are also answered It is considered as protection scope of the present invention.

Claims (5)

1. a kind of high spectrum image emissivity spectra extracting method based on end member abundance inverting, which is characterized in that the method Include the following steps:
1) Decomposition of Mixed Pixels is carried out to whole picture infrared high spectrum image, obtains the end member and each end member pair of infrared high spectrum image The Abundances answered；
2) interesting target is determined；
3) according to pure end member and the Abundances, the interesting target is extracted, obtains each picture in infrared high spectrum image The spoke luminance spectral curve of member point；
4) set temperature threshold range；
5) to the spoke luminance spectral curve march line integral of the interesting target within the scope of the temperature threshold of setting；
6) the absolute black body temperature of infrared high spectrum emissivity curve simulation is estimated according to integral area；
7) emissivity spectra curve is obtained, and is normalized.
2. high spectrum image emissivity spectra extracting method as described in claim 1, which is characterized in that step 1) mixed pixel Decompose includes Endmember extraction step and abundance inversion step, in which:
Endmember extraction step is extracted using the infrared high spectrum end member extraction method based on extended mathematical morphology；
Abundance inversion step carries out abundance inverting using linear solution mixing method.
3. high spectrum image emissivity spectra extracting method as claimed in claim 2, which is characterized in that the Endmember extraction step Suddenly include:
1. reading in infrared high spectrum original image, the least structure element K is utilizedmin, to target high spectrum image in structural element Each pixel expansion based on mass center is carried out in neighborhood and corrosion calculates, find most pure pixel d (x, y, w) and mixing The pixel e (x, y, w) of most serious, and calculate corresponding MEI value at this time；
MEI indicates morphology eccentricity index, is defined as:
MEI (x, y, w)=dist (d (x, y, w), e (x, y, w)), w=1,2 ..., n
Wherein, dist indicates the distance between two o'clock, and d (x, y, w) indicates that pixel most pure in image, e (x, y, w) indicate mixed Close the pixel of most serious；
2. the step of being continuously increased the size of structural element, repeating step 1., until reaching max architecture element Kmax；It counts repeatedly It calculates MEI value and notes down, the corresponding each MEI value of identical pixel is averaged, final MEI image is obtained；
3. carrying out threshold decision, the gray value of obtained MEI image all the points is taken into the average threshold value as setting, MEI figure The pixel for being greater than this given threshold as in will be marked as Pure pixel；
4. carrying out image segmentation to the end member image finally obtained and region growing completing automatically selecting for end member；
5. carrying out linear spectral separation using linear solution mixing method obtains end member abundance figure.
4. high spectrum image emissivity spectra extracting method as described in claim 1, which is characterized in that it is described to calculate step 3) The expression formula of the spoke luminance spectral curve of each pixel point are as follows:
Wherein, ejIndicate each end member in image, αijIndicate each end member proportion in the picture, E indicates end member matrix, αiTable Show abundance matrix.
5. high spectrum image emissivity spectra extracting method as described in claim 1, which is characterized in that it is described to calculate step 6) The expression formula of absolute black body temperature are as follows:
Wherein: T1,T2For the initial temperature threshold of setting；L1,L2Respectively T1,T2Corresponding black matrix spoke brightness；LVFor spectrum song Line integral.
CN201610340061.1A 2016-05-19 2016-05-19 A kind of high spectrum image emissivity spectra extracting method based on end member abundance inverting Active CN106056044B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201610340061.1A CN106056044B (en) 2016-05-19 2016-05-19 A kind of high spectrum image emissivity spectra extracting method based on end member abundance inverting

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201610340061.1A CN106056044B (en) 2016-05-19 2016-05-19 A kind of high spectrum image emissivity spectra extracting method based on end member abundance inverting

Publications (2)

Publication Number Publication Date
CN106056044A CN106056044A (en) 2016-10-26
CN106056044B true CN106056044B (en) 2019-03-29

Family

ID=57177292

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201610340061.1A Active CN106056044B (en) 2016-05-19 2016-05-19 A kind of high spectrum image emissivity spectra extracting method based on end member abundance inverting

Country Status (1)

Country Link
CN (1) CN106056044B (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106595873B (en) * 2017-01-03 2019-03-29 哈尔滨工业大学 Sub-pixed mapping temperature retrieval method based on LONG WAVE INFRARED surface air radiation and visible light wave range linear mixed model
CN107274387B (en) * 2017-05-19 2019-09-06 西安电子科技大学 The end member extraction method of target in hyperspectral remotely sensed image based on Evolutionary multiobjective optimization
CN109696406B (en) * 2018-11-29 2020-10-02 北京航空航天大学 Moon table hyperspectral image shadow region unmixing method based on composite end member
CN110428454B (en) * 2019-08-13 2022-02-25 电子科技大学中山学院 Hyperspectral unmixing method and device, electronic equipment and storage medium

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101887012A (en) * 2010-06-28 2010-11-17 中国国土资源航空物探遥感中心 Spectral reflectance peak decomposition based quantitative inversion method of hyperspectral remote sensing mineral content
CN102194099A (en) * 2010-02-17 2011-09-21 波音公司 Advanced background estimation technique and circuit for a hyper-spectral target detection method
CN103325096A (en) * 2013-06-25 2013-09-25 中国科学院遥感与数字地球研究所 Method for reconstructing wide hyperspectral image based on fusion of multispectral/hyperspectral images

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8417748B2 (en) * 2007-09-28 2013-04-09 University Of Maryland At Baltimore County Maximum simplex volume criterion-based endmember extraction algorithms

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102194099A (en) * 2010-02-17 2011-09-21 波音公司 Advanced background estimation technique and circuit for a hyper-spectral target detection method
CN101887012A (en) * 2010-06-28 2010-11-17 中国国土资源航空物探遥感中心 Spectral reflectance peak decomposition based quantitative inversion method of hyperspectral remote sensing mineral content
CN103325096A (en) * 2013-06-25 2013-09-25 中国科学院遥感与数字地球研究所 Method for reconstructing wide hyperspectral image based on fusion of multispectral/hyperspectral images

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
基于能量平衡与端元合成技术的地表热辐射场景动态模拟;赵利民，等.;《中国科学》;20121231;第42卷(第11期);第1330-1338页
高光谱遥感岩矿端元提取与分析方法研究;刘汉湖，等.;《岩石矿物学杂志》;20130331;第32卷(第2期);第213-221页

Also Published As

Publication number Publication date
CN106056044A (en) 2016-10-26

Similar Documents

Publication Publication Date Title
CN106056044B (en) A kind of high spectrum image emissivity spectra extracting method based on end member abundance inverting
CN109949278A (en) Hyperspectral abnormity detection method based on confrontation autoencoder network
CN107301624B (en) Convolutional neural network defogging method based on region division and dense fog pretreatment
CN109447008A (en) Population analysis method based on attention mechanism and deformable convolutional neural networks
Zhang et al. Statistical properties of solar active regions obtained from an automatic detection system and the computational biases
CN111007039B (en) Automatic extraction method and system for sub-pixel level water body of medium-low resolution remote sensing image
CN107239781B (en) Hyperspectral reflectivity reconstruction method based on RGB image
CN109308715A (en) A kind of optical imagery method for registering combined based on point feature and line feature
Etezadifar et al. A new sample consensus based on sparse coding for improved matching of SIFT features on remote sensing images
CN106407978B (en) Method for detecting salient object in unconstrained video by combining similarity degree
CN115330876B (en) Target template graph matching and positioning method based on twin network and central position estimation
US20080310735A1 (en) Anomalous change detection in imagery
CN107644203B (en) Feature point detection method for shape adaptive classification
TWI792696B (en) Methods and apparatuses of contrastive learning for color constancy
Zhang et al. A novel multitemporal cloud and cloud shadow detection method using the integrated cloud Z-scores model
CN110276782B (en) Hyperspectral target tracking method combining spatial spectral features and related filtering
CN106846377A (en) A kind of target tracking algorism extracted based on color attribute and active features
CN114117908A (en) High-precision ASI sea ice density inversion algorithm for data correction based on CGAN
CN106204596B (en) Panchromatic waveband remote sensing image cloud detection method based on Gaussian fitting function and fuzzy mixed estimation
CN108596906A (en) It is a kind of to refer to screen image quality evaluating method entirely based on sparse locality preserving projections
CN109726679B (en) Remote sensing classification error spatial distribution mapping method
CN107273793A (en) A kind of feature extracting method for recognition of face
CN115620132A (en) Unsupervised comparative learning ice lake extraction method
Zeng et al. U-net-based multispectral image generation from an rgb image
Zhuang et al. Harbor water area extraction from pan-sharpened remotely sensed images based on the definition circle model

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