CN107133954A - Spectrum angle matching process based on mathematical morphology - Google Patents
Spectrum angle matching process based on mathematical morphology Download PDFInfo
- Publication number
- CN107133954A CN107133954A CN201710229752.9A CN201710229752A CN107133954A CN 107133954 A CN107133954 A CN 107133954A CN 201710229752 A CN201710229752 A CN 201710229752A CN 107133954 A CN107133954 A CN 107133954A
- Authority
- CN
- China
- Prior art keywords
- pixel
- mrow
- spectrum
- mei
- image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/155—Segmentation; Edge detection involving morphological operators
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/187—Segmentation; Edge detection involving region growing; involving region merging; involving connected component labelling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/13—Satellite images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10032—Satellite or aerial image; Remote sensing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Astronomy & Astrophysics (AREA)
- Remote Sensing (AREA)
- Multimedia (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
Disclose the spectrum angle matching process based on mathematical morphology.Mathematical morphology Endmember extraction algorithm and spectrum angle matching algorithm are combined by the present invention, are carried out Endmember extraction using mathematical morphology expansion and erosion operation, have been effectively combined the information of object spectrum and spatial coherence;The detection identification of interesting target is carried out using spectrum angle matching algorithm, overcome that spectrum angle matching algorithm reacts spectral signature sensitive in itself, nicety of grading can very low shortcoming in mixed pixel generally existing, the influence of background and noise is restrained effectively, so as to reduce the false alarm rate of target detection.The present invention can solve the problem that, in the case where target and background are unknown, high spectrum image interesting target detects the problem of recognizing, practicality is good.
Description
Technical field
The present invention relates to EO-1 hyperion target property and target identification technology field, more particularly to the light based on mathematical morphology
Spectral corner degree matching process.
Background technology
The background of related to the present invention is illustrated below, but these explanations might not constitute the existing of the present invention
Technology.
The rise of high-spectrum remote-sensing is one of maximum achievement of 1980s remote sensing technology, and EO-1 hyperion target detection
Technology is one of mostly important application direction of high-spectrum remote-sensing.Traditional algorithm of target detection is usually assuming that data are obeyed
Detective operators are constructed on the basis of certain statistics or geometrical model, and the statistical parameter in operator, example are estimated according to prior information
Such as adaptive cosine estimation (Adaptive cosine estimator, ACE), Orthogonal subspace projection (Orthogonal
Subspace projector, OSP) algorithm etc..But in actual applications, the prior information of target is difficult to obtain.
Target detection in the case of unknown object, Unknown Background can realize that one kind is direct by two ways
Anomaly target detection is carried out according to the distribution of information content, mainly has abnormality detection (RXD) algorithm, equalization target detection (UTD) to calculate
Method etc..Another is the non-supervisory target inspection for utilizing the Endmember extraction technology in Decomposition of Mixed Pixels to obtain target and background information
Survey.Under normal circumstances, because the limitation by remote sensing images spatial resolution and atural object complicated variety are influenceed, some senses are emerging
Interesting target exists in the form of mixed pixel mostly in the picture.Therefore, the high spectrum image based on Endmember extraction algorithm is studied
Target detection technique is just particularly important.
Traditional high spectrum image target detection technique based on Endmember extraction algorithm is from data spectral information and spy
The angle for levying spatial analysis sets out what is handled, have ignored the spatial coherence existed between pixel.In order to more accurately
The analysis of high-spectrum remote sensing is carried out, it is very necessary to consider the spectral information and spatial information of high-spectral data offer
's.
The content of the invention
It is an object of the invention to propose the spectrum angle matching process based on mathematical morphology, can solve the problem that target with
In the case that background is unknown the problem of high spectrum image interesting target detection identification.
Spectrum angle matching process of the invention based on mathematical morphology, including:
S1, expansion and corrosion of the progress of each pixel based on barycenter to target high spectrum image in structural element are transported
Calculate, obtain the corresponding MEI values of each pixel, obtain MEI images;
Each pixel in S2, traversal MEI images, end member is labeled as by the pixel that MEI values are more than default MEI threshold values,
Obtain end member image;
S3, image segmentation and region growing are carried out to the end member image, obtain endmember spectra image;
Each end member in S4, traversal endmember spectra image, the endmember spectra is obtained using spectrum angle matching algorithm
Each spectrum angle between pixel spectrum and current endmember spectra in image, if the spectrum angle is 0, by current picture
It is first to be classified as a class with current end member.
Preferably, in step S1, each picture in two or more structural elements respectively to target high spectrum image
Member carries out expansion and erosion operation based on barycenter, obtains the MEI values of each pixel;
For any one pixel, the average value of the MEI values obtained using different structure element is regard as pixel correspondence
MEI values;
Wherein, different structure element is of different sizes.
Preferably, in step S1, first in the least structure element KminIn each pixel of target high spectrum image is carried out
Expansion and erosion operation based on barycenter, obtain the MEI values of each pixel;
Then the size of structural element is increased successively, to each pixel of target high spectrum image in each structural element
The expansion based on barycenter and erosion operation are carried out, the MEI values of each pixel are obtained, until reaching max architecture element Kmax;
For any one pixel, the average value of the MEI values obtained using different structure element is regard as pixel correspondence
MEI values;
Wherein, the least structure element Kmin, max architecture element KmaxAnd between the least structure element KminWith maximum knot
Constitutive element KmaxBetween structural element pre-set.
Preferably, step S1 includes:For each structural element:
In structural element, the dilation operation based on barycenter is carried out to each pixel of target high spectrum image, is up to
The maximum pixel of the distance of structural element barycenter is used as most Pure pixel;
In structural element, the erosion operation based on barycenter is carried out to each pixel of target high spectrum image, is up to
The minimum pixel of the distance of structural element barycenter is used as mixing most serious pixel;
Most Pure pixel is regard as structural element current location pair with mixing the spectrum angular distance between most serious pixel
The MEI values for the pixel answered.
Preferably for any one structural element, its barycenter is:
In formula, K representative structure elements;M is the quantity of pixel in structural element K;cKFor structural element K barycenter;f(s,
T, w) it is any one pixel in structural element K, (s, t, w) is pixel f (s, t, w) coordinate.
Preferably, further comprise before step S2:Using the average value of the MEI values of each pixel in MEI images as pre-
If MEI threshold values.
Preferably, in step S4 using equation below determine each pixel spectrum in the endmember spectra image with currently
Spectrum angle between endmember spectra:
In formula, n is wave band number;X is pixel spectrum, and Y is endmember spectra;θ (X, Y) be pixel spectrum X and endmember spectra Y it
Between spectrum angle, codomain is 0~pi/2, when θ (X, Y) is 0, represents that pixel spectrum X is identical with endmember spectra, when θ=
During pi/2, represent that pixel spectrum X and endmember spectra are entirely different.
The present invention by mathematical morphology Endmember extraction algorithm and spectrum angle automatching (Spectral angle mapping,
SAM) algorithm is combined, and is carried out Endmember extraction using mathematical morphology expansion and erosion operation, has been effectively combined object spectrum
With the information of spatial coherence;The detection identification of interesting target is carried out using spectrum angle matching algorithm, spectral modeling is overcome
Spend matching algorithm spectral signature is reacted in itself it is sensitive, in mixed pixel generally existing nicety of grading can very low shortcoming, have
The influence of background and noise is inhibited to effect, so as to reduce the false alarm rate of target detection.The present invention can solve the problem that target with
In the case that background is unknown, the problem of detection of high spectrum image interesting target is recognized, practicality is good.
Brief description of the drawings
By the embodiment part of offer referring to the drawings, the features and advantages of the present invention will become more
It is readily appreciated that, in the accompanying drawings:
Fig. 1 is the schematic flow sheet for showing the spectrum angle matching process of the invention based on mathematical morphology.
Embodiment
The illustrative embodiments to the present invention are described in detail with reference to the accompanying drawings.Illustrative embodiments are retouched
State merely for the sake of demonstration purpose, and be definitely not to the present invention and its application or the limitation of usage.
As shown in figure 1, the spectrum angle matching process of the invention based on mathematical morphology includes:
S1, expansion and corrosion of the progress of each pixel based on barycenter to target high spectrum image in structural element are transported
Calculate, obtain the corresponding MEI values of each pixel, obtain MEI images;
Each pixel in S2, traversal MEI images, end member is labeled as by the pixel that MEI values are more than default MEI threshold values,
Obtain end member image;
S3, image segmentation and region growing are carried out to the end member image, obtain endmember spectra image;
Each end member in S4, traversal endmember spectra image, the endmember spectra is obtained using spectrum angle matching algorithm
Each spectrum angle between pixel spectrum and current endmember spectra in image, if the spectrum angle is 0, by current picture
It is first to be classified as a class with current end member.
The present invention using mathematical morphology expansion and erosion operation carry out Endmember extraction, be effectively combined object spectrum and
The information of spatial coherence;The detection identification of interesting target is carried out using spectrum angle matching algorithm, spectrum angle is overcome
Matching algorithm spectral signature is reacted in itself it is sensitive, in mixed pixel generally existing nicety of grading can very low shortcoming, effectively
Ground inhibits the influence of background and noise, so as to reduce the false alarm rate of target detection.
A structural element can be used only in those skilled in the art, and such as 3 × 3 structural element is expanded and rotten
Lose computing.It can also be expanded and erosion operation using the structural element of two or more sizes.In certain embodiments,
In order to reduce in influence of the selection to operation result due to structural element, step S1 as far as possible, in two or more structural elements
The expansion based on barycenter and erosion operation are carried out to each pixel of target high spectrum image respectively in element, each pixel is obtained
MEI values;It is for any one pixel, the average value of the MEI values obtained using different structure element is corresponding as the pixel
MEI values;Wherein, different structure element is of different sizes, for example, using 3 × 3 structural element, 5 × 5 structural element and 7 ×
7 structural element.The quantity of structural element is more, and the result of Endmember extraction is more accurate.
For the ease of being expanded and erosion operation automatically, the least structure element K can be presetmin, max architecture element
KmaxAnd between the least structure element KminWith max architecture element KmaxBetween structural element, in step sl, exist first
The least structure element KminIn the expansion based on barycenter and erosion operation are carried out to each pixel of target high spectrum image, obtain
The MEI values of each pixel;
Then the size of structural element is increased successively, to each pixel of target high spectrum image in each structural element
The expansion based on barycenter and erosion operation are carried out, the MEI values of each pixel are obtained, until reaching max architecture element Kmax;
For any one pixel, the average value of the MEI values obtained using different structure element is regard as pixel correspondence
MEI values.
In structural element K, its barycenter can be defined as:
In formula, K representative structure elements;M is the quantity of pixel in structural element K;cKFor structural element K barycenter;f(s,
T, w) it is any one pixel in structural element K, (s, t, w) is pixel f (s, t, w) coordinate.
MEI is morphology eccentricity index, represents the purity of pixel in structural element.MEI indexes are pure in structural element
The light spent between highest pixel (being obtained by expansive working) and mixability highest pixel (being obtained by etching operation)
Angular distance is composed, is defined as follows:
MEI (x, y, w)=dist (d (x, y, w), e (x, y, w))
In order to obtain the MEI values of each pixel of target high spectrum image in structural element, in certain embodiments, step
S1 includes:For each structural element:
In structural element, the dilation operation based on barycenter is carried out to each pixel of target high spectrum image, is up to
The maximum pixel of the distance of structural element barycenter is used as most Pure pixel;
In structural element, the erosion operation based on barycenter is carried out to each pixel of target high spectrum image, is up to
The minimum pixel of the distance of structural element barycenter is used as mixing most serious pixel;
Most Pure pixel is regard as structural element current location pair with mixing the spectrum angular distance between most serious pixel
The MEI values for the pixel answered.
Barycenter c is arrived for some pixel f (x, y, w) in structural element KKRange formula be defined as:
D (f (x, y, w), K)=dist (f (x, y, w), cK)
Wherein, dist represents the spectrum angular distance between two vectors.
Expansion and erosion operation can respectively be represented with following formula:
Wherein, what arg_Max and arg_Min referred to making obtaining barycenter reaches minimum and maximum pixel vector apart from D,
Difference counter structure element K moderate purity highest pixels and mixability highest pixel.
The bigger pixel of MEI values in the MEI characterization images purity information of original high spectrum image, image, its correspondence is former
Pixel in beginning image is more likely to become end member;The smaller pixel of MEI values in image, its correspondence original image in pixel into
It is smaller for the possibility of end member.The present invention is obtained after endmember spectra image, is completed with spectrum angle matching algorithm to feeling emerging
The detection and identification of interesting target.In order to reduce the amount of calculation of spectrum angle matching algorithm as far as possible, the present invention is traveled through in step s 2
Each pixel in MEI images, is labeled as end member by the pixel that MEI values are more than default MEI threshold values, obtains end member image.
Those skilled in the art can select suitable MEI threshold values, in certain embodiments, step S2 according to actual conditions
Further comprise before:It regard the average value of the MEI values of each pixel in MEI images as default MEI threshold values.So determine
MEI threshold values can not only exclude the interference of the unlikely pixel as end member well, greatly reduce follow-up spectral modeling degree
The amount of calculation of amount of calculation with algorithm.In addition, the method for this determination MEI threshold values is simple and quick.And it is easy to high according to target
The actual conditions applicability of spectrum picture it is adjusted.
Spectrum angle matching algorithm determines two by calculating the angle between a pixel spectrum and a reference spectra
Similitude between person.Reference spectra used of the invention is the endmember spectra extracted using end member extraction method from image.It is existing
Have in the spectrum angle matching algorithm of technology, the angle between reference spectra and pixel spectrum suffers from this vectorial body length
Influence, i.e., can not exclude the multiplier interference of reference spectra and pixel spectral vector.Based on this, in some embodiments of the present invention
In, determined in step S4 in the endmember spectra image using equation below each between pixel spectrum and current endmember spectra
Spectrum angle:
In formula, n is wave band number;X is pixel spectrum, and Y is endmember spectra;θ (X, Y) be pixel spectrum X and endmember spectra Y it
Between spectrum angle, codomain is 0~pi/2, when θ (X, Y) is 0, represents that pixel spectrum X is identical with endmember spectra, when θ=
During pi/2, represent that pixel spectrum X and endmember spectra are entirely different.
It is assumed that spectrum is by after multiplier interference, pixel spectral vector is changed into aX, and reference spectra vector is changed into bY, a and b in fact
Number, then the spectral modeling after multiplier interference is calculated as follows:
It can be seen that, the spectrum between pixel spectrum and endmember spectra is determined using the formula in above preferred embodiment of the present invention
Angle, the spectral modeling calculated is not influenceed by this vectorial body length, to pixel spectrum and the multiplier interference of endmember spectra vector
With consistency.This consistency is very useful in EO-1 hyperion is matched and is recognized, Atmospheric Compensation is calculated first in high-spectrum remote-sensing
The finiteness of method brings multiplier interference, and secondly the gradient of topographical surface has an impact to illumination, can also cause multiplier interference, these
Multiplier interference can cause the change of object spectrum amplitude, and spectrum angle matching algorithm to spectral line when matching, not by
The influence of length is measured, therefore, above-mentioned spectrum angle matching algorithm of the invention can effectively weaken Atmospheric Compensation and the atural object gradient
Influence to spectral line similarity measure.
The codomain of spectrum angle between pixel spectrum and endmember spectra is 0~pi/2, between pixel spectrum and endmember spectra
Spectrum angle it is smaller, show the pixel spectrum and endmember spectra closer to when θ (X, Y) is 0, representing pixel spectrum X and end
First spectrum is identical, when θ=pi/2, represents that pixel spectrum X and endmember spectra are entirely different.Find out between endmember spectra
Spectrum angle be 0 pixel spectrum, these pixels and the end member are classified as same class, you can reach and detect interesting target
Effect.In some embodiments, it is also possible to carry out EO-1 hyperion target detection according to the following equation:
Wherein, T (X) represents any one pixel spectrum XiWith endmember spectra YjBetween spectrum angle, m is endmember spectra
Pixel number in image, n is the end member number in endmember spectra image, and arg min are minimum angles value.Counted first during detection
Pixel spectral vector X in nomogram pictureiWith reference spectra vector YjBetween angle, then find out the every kind of end member of correspondence and cause minimum
Angle is 0 image picture elements, these pixels and the end member is classified as into same class, you can reach the effect for detecting interesting target
Really.
Compared with prior art, the present invention mutually ties mathematical morphology Endmember extraction algorithm with spectrum angle matching algorithm
Close, can solve the problem that in the case where target and background are unknown, the problem of detection of high spectrum image interesting target is recognized.
Although with reference to illustrative embodiments, invention has been described, but it is to be understood that the present invention does not limit to
The embodiment that Yu Wenzhong is described in detail and shown, in the case of without departing from claims limited range, this
Art personnel can make various changes to the illustrative embodiments.
Claims (7)
1. the spectrum angle matching process based on mathematical morphology, it is characterised in that including:
S1, expansion and erosion operation of the progress of each pixel based on barycenter in structural element to target high spectrum image, are obtained
The corresponding MEI values of each pixel are taken, MEI images are obtained;
Each pixel in S2, traversal MEI images, is labeled as end member by the pixel that MEI values are more than default MEI threshold values, obtains
End member image;
S3, image segmentation and region growing are carried out to the end member image, obtain endmember spectra image;
Each end member in S4, traversal endmember spectra image, the endmember spectra image is obtained using spectrum angle matching algorithm
In each spectrum angle between pixel spectrum and current endmember spectra, if the spectrum angle is 0, by current pixel with
Current end member is classified as a class.
2. the method as described in claim 1, it is characterised in that in step S1, in two or more structural elements respectively
Expansion based on barycenter and erosion operation are carried out to each pixel of target high spectrum image, the MEI values of each pixel are obtained;
For any one pixel, the average value of the MEI values obtained using different structure element is regard as the corresponding MEI of the pixel
Value;
Wherein, different structure element is of different sizes.
3. method as claimed in claim 2, it is characterised in that in step S1, first in the least structure element KminIn to target
Each pixel of high spectrum image carries out expansion and erosion operation based on barycenter, obtains the MEI values of each pixel;
Then increase the size of structural element successively, each pixel of target high spectrum image is carried out in each structural element
Expansion and erosion operation based on barycenter, obtain the MEI values of each pixel, until reaching max architecture element Kmax;
For any one pixel, the average value of the MEI values obtained using different structure element is regard as the corresponding MEI of the pixel
Value;
Wherein, the least structure element Kmin, max architecture element KmaxAnd between the least structure element KminWith max architecture member
Plain KmaxBetween structural element pre-set.
4. method as claimed in claim 3, it is characterised in that step S1 includes:For each structural element:
In structural element, the dilation operation based on barycenter is carried out to each pixel of target high spectrum image, structure is up to
The maximum pixel of the distance of element barycenter is used as most Pure pixel;
In structural element, the erosion operation based on barycenter is carried out to each pixel of target high spectrum image, structure is up to
The minimum pixel of the distance of element barycenter is used as mixing most serious pixel;
Spectrum angular distance between most Pure pixel and mixing most serious pixel is corresponding as structural element current location
The MEI values of pixel.
5. the method as described in claim 1, it is characterised in that for any one structural element, its barycenter is:
<mrow>
<msub>
<mi>c</mi>
<mi>K</mi>
</msub>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mi>M</mi>
</mfrac>
<munder>
<mo>&Sigma;</mo>
<mi>s</mi>
</munder>
<munder>
<mo>&Sigma;</mo>
<mi>t</mi>
</munder>
<mi>f</mi>
<mrow>
<mo>(</mo>
<mi>s</mi>
<mo>,</mo>
<mi>t</mi>
<mo>,</mo>
<mi>w</mi>
<mo>)</mo>
</mrow>
<mo>,</mo>
<mo>&ForAll;</mo>
<mrow>
<mo>(</mo>
<mi>s</mi>
<mo>,</mo>
<mi>t</mi>
<mo>,</mo>
<mi>w</mi>
<mo>)</mo>
</mrow>
<mo>&Element;</mo>
<mi>K</mi>
</mrow>
In formula, K representative structure elements;M is the quantity of pixel in structural element K;cKFor structural element K barycenter;F (s, t, w) is
Any one pixel in structural element K, (s, t, w) is pixel f (s, t, w) coordinate.
6. method as claimed in claim 4, it is characterised in that further comprise before step S2:By each picture in MEI images
The average value of the MEI values of member is used as default MEI threshold values.
7. the method as described in claim 1, it is characterised in that the endmember spectra figure is determined using equation below in step S4
Each spectrum angle between pixel spectrum and current endmember spectra as in:
<mrow>
<mi>&theta;</mi>
<mrow>
<mo>(</mo>
<mi>X</mi>
<mo>,</mo>
<mi>Y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mi>arccos</mi>
<mo>&lsqb;</mo>
<mrow>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>n</mi>
</munderover>
<msub>
<mi>x</mi>
<mi>i</mi>
</msub>
<msub>
<mi>y</mi>
<mi>i</mi>
</msub>
<mo>/</mo>
<mrow>
<mo>&lsqb;</mo>
<mrow>
<msup>
<mrow>
<mo>(</mo>
<mrow>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>n</mi>
</munderover>
<msubsup>
<mi>x</mi>
<mi>i</mi>
<mn>2</mn>
</msubsup>
</mrow>
<mo>)</mo>
</mrow>
<mrow>
<mn>1</mn>
<mo>/</mo>
<mn>2</mn>
</mrow>
</msup>
<msup>
<mrow>
<mo>(</mo>
<mrow>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>n</mi>
</munderover>
<msubsup>
<mi>y</mi>
<mi>i</mi>
<mn>2</mn>
</msubsup>
</mrow>
<mo>)</mo>
</mrow>
<mrow>
<mn>1</mn>
<mo>/</mo>
<mn>2</mn>
</mrow>
</msup>
</mrow>
<mo>&rsqb;</mo>
</mrow>
</mrow>
<mo>&rsqb;</mo>
</mrow>
In formula, n is wave band number;X is pixel spectrum, and Y is endmember spectra;θ (X, Y) is between pixel spectrum X and endmember spectra Y
Spectrum angle, codomain is 0~pi/2, when θ (X, Y) is 0, represents that pixel spectrum X is identical with endmember spectra, when θ=pi/2
When, represent that pixel spectrum X and endmember spectra are entirely different.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710229752.9A CN107133954A (en) | 2017-04-10 | 2017-04-10 | Spectrum angle matching process based on mathematical morphology |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710229752.9A CN107133954A (en) | 2017-04-10 | 2017-04-10 | Spectrum angle matching process based on mathematical morphology |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107133954A true CN107133954A (en) | 2017-09-05 |
Family
ID=59715536
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710229752.9A Pending CN107133954A (en) | 2017-04-10 | 2017-04-10 | Spectrum angle matching process based on mathematical morphology |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107133954A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112613413A (en) * | 2020-12-25 | 2021-04-06 | 平安国际智慧城市科技股份有限公司 | Perishable garbage classification quality determination method and device and computer readable storage medium |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101504315A (en) * | 2009-02-23 | 2009-08-12 | 北京航空航天大学 | Expansion morphology and orthogonal subspace projection combined end member automatic extraction method |
CN103258330A (en) * | 2013-05-24 | 2013-08-21 | 大连海事大学 | Method for estimating abundance of hyperspectral image end member |
CN106447688A (en) * | 2016-03-31 | 2017-02-22 | 大连海事大学 | Method for effectively segmenting hyperspectral oil-spill image |
-
2017
- 2017-04-10 CN CN201710229752.9A patent/CN107133954A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101504315A (en) * | 2009-02-23 | 2009-08-12 | 北京航空航天大学 | Expansion morphology and orthogonal subspace projection combined end member automatic extraction method |
CN103258330A (en) * | 2013-05-24 | 2013-08-21 | 大连海事大学 | Method for estimating abundance of hyperspectral image end member |
CN106447688A (en) * | 2016-03-31 | 2017-02-22 | 大连海事大学 | Method for effectively segmenting hyperspectral oil-spill image |
Non-Patent Citations (2)
Title |
---|
刘畅 等: "基于扩展数学形态学的高光谱亚像元目标检测", 《红外与激光工程》 * |
孟强强 等: "基于端元提取的高光谱图像亚像元目标异常检测算法", 《科学技术与工程》 * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112613413A (en) * | 2020-12-25 | 2021-04-06 | 平安国际智慧城市科技股份有限公司 | Perishable garbage classification quality determination method and device and computer readable storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN101539629B (en) | Remote sensing image change detection method based on multi-feature evidence integration and structure similarity | |
CN106650829A (en) | Picture similarity calculation method | |
CN109949340A (en) | Target scale adaptive tracking method based on OpenCV | |
CN104200236A (en) | Quick target detection method based on DPM (deformable part model) | |
CN106548462A (en) | Non-linear SAR image geometric correction method based on thin-plate spline interpolation | |
CN103839257B (en) | Method for detecting changes of SAR images of generalized Gaussian K&I | |
CN101398934A (en) | Method and system for tracking objects in images | |
CN104517276A (en) | Checker corner detection method | |
CN105934757B (en) | A kind of method and apparatus of the key point for detecting the first image and the incorrect incidence relation between the key point of the second image | |
CN108492298A (en) | Based on the multispectral image change detecting method for generating confrontation network | |
CN108492313A (en) | A kind of dimension self-adaption visual target tracking method based on middle intelligence similarity measure | |
CN108765476A (en) | Polarized image registration method | |
CN107341824B (en) | Comprehensive evaluation index generation method for image registration | |
CN104933738A (en) | Visual saliency map generation method based on local structure detection and contrast | |
CN103559725B (en) | A kind of wireless sensor node optimum choice method of following the tracks of towards vision | |
CN103679720A (en) | Fast image registration method based on wavelet decomposition and Harris corner detection | |
CN106570886A (en) | Target tracking method based on super-resolution reconstruction | |
CN104318559A (en) | Quick feature point detecting method for video image matching | |
CN110659637A (en) | Electric energy meter number and label automatic identification method combining deep neural network and SIFT features | |
CN109255785A (en) | A kind of bearing open defect detection system | |
CN106709943A (en) | Point cloud registration method based on optimal transmission | |
CN103729462B (en) | A kind of pedestrian retrieval method blocked based on rarefaction representation process | |
CN107133954A (en) | Spectrum angle matching process based on mathematical morphology | |
CN107273803A (en) | Cloud layer image detecting method | |
CN111798516B (en) | Method for detecting running state quantity and analyzing errors of bridge crane equipment |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170905 |