CN105678229B - A kind of Hyperspectral imaging search method - Google Patents
A kind of Hyperspectral imaging search method Download PDFInfo
- Publication number
- CN105678229B CN105678229B CN201511020652.2A CN201511020652A CN105678229B CN 105678229 B CN105678229 B CN 105678229B CN 201511020652 A CN201511020652 A CN 201511020652A CN 105678229 B CN105678229 B CN 105678229B
- Authority
- CN
- China
- Prior art keywords
- curve
- spectrum
- point
- retrieved
- value
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
A kind of Hyperspectral imaging search method, a plurality of curve of spectrum that the curve of spectrum and needs including obtaining atural object to be retrieved are retrieved, and M SARI characteristic point is extracted respectively in the position of every curve of spectrum wave crest and trough；According to M on every curve of spectrum fixed SARI characteristic points, by calculating the N number of total point obtained on every curve of spectrum；According to N number of total point on every curve of spectrum, the match point of the curve of spectrum to be retrieved and every curve of spectrum for needing to retrieve is obtained by the method that wave band compares；Calculate square of the difference of the spectral value of the match point, and the similarity of square the calculating curve of spectrum to be retrieved and every curve of spectrum for needing to retrieve according to the difference of the spectral value of match point；The normalized value of the calculated result of similarity is obtained, and according to Sequential output search result from big to small.The curve of spectrum can effectively be simplified, and effectively improve the retrieval precision of the curve of spectrum.
Description
Technical field
The invention belongs to technical field of remote sensing image processing more particularly to a kind of Hyperspectral imaging search methods.
Background technique
Hyperspectral imaging can provide more information content and higher spectral resolution than common image, to Objects recognition
It is of great significance, but the magnanimity for also bringing data volume at the same time increases, and how efficiently to realize video search, is bloom
Research hotspot in modal data application.For Hyperspectral imaging, the curve of spectrum is that expression image spectral signature is most common
Mode, therefore the feature of the curve of spectrum is extracted, carrying out video search using similarity measurements figureofmerit is a kind of efficiently feasible side
Method.
Curve of spectrum feature extraction is also widely used in the fields such as crop identification, chemical analysis.And in similarity evaluation side
Face, common similarity measurements figureofmerit have Euclidean distance, related coefficient, spectral information divergence (Spectral Information
Divergence, SID), various squares, spectrum angle automatching (Spectral Angle Match, SAM) etc..However utilizing this
When a little index of similarity are retrieved, usually all wave band datas all participate in operation, for hundreds and thousands of a wave bands,
Calculation amount is very big, and recall precision is relatively low.
A kind of feasible solution is to carry out curve to spectral vector to simplify.The simplification algorithm of comparative maturity has at present
Douglas-Peucker (DP) method is hung down away from method, Li-Openshaw method etc., and wherein DP algorithm is the most commonly used, the basic principle is that:
Straight line is virtually connected to the first and last two o'clock of curve, all the points and find maximum range value to the distance of straight line on calculated curve
Dmax is compared with threshold value D, if dmax < D, the intermediate point on this curve is all cast out, and otherwise retains the corresponding point of dmax,
And using this point as boundary, curve is divided into two parts, operation above then is repeated to this two parts, until there is no distance big
Until the point of threshold value.
Although DP algorithm can effectively save the morphological feature of curve on the whole, to be realized using recursive algorithm,
It is more complicated, while in the upper also following two defect of application:
(1) specified threshold is needed, and the selection of the effect and threshold value of curve simplification has very big relationship, when threshold value increases
When, the point of reservation will be reduced, it is possible to be distorted；Otherwise the point retained will increase, and the purpose of simplified curve is not achieved.In order to
Reach set compression effectiveness, needs to be constantly trying to threshold value, larger workload.
(2) when a plurality of curve carries out while simplifying, to guarantee that all curves can effectively be simplified, retain enough
Point needs the threshold value different to different curve settings, and workload is bigger, and the points that every curve retains have very big difference,
This problem is especially prominent in the simplification of bloom spectral curve and comparison.
Summary of the invention
Present invention seek to address that computationally intensive when Hyperspectral imaging is retrieved in the prior art, the simplified process of curve needs to specify
The low technical problem of threshold value, recall precision provides a kind of bloom that retrieval calculation amount is small, high without specified threshold, recall precision
Compose method for retrieving image.
The embodiment of the present invention provides a kind of Hyperspectral imaging search method, comprising: the spectrum for obtaining atural object to be retrieved is bent
The a plurality of curve of spectrum that line and needs are retrieved, and X SARI is extracted respectively in the position of every curve of spectrum wave crest and trough
Characteristic point；
According to X on every curve of spectrum fixed SARI characteristic points, by calculating the N obtained on every curve of spectrum
A total point, N number of total point on every curve of spectrum include X described on the curve fixed SARI characteristic points；
According to N number of total point on every curve of spectrum, the curve of spectrum to be retrieved is obtained by the method that wave band compares and is needed
The match point of every curve of spectrum being retrieved；
Square of the difference of the spectral value of the match point is calculated, and according to square meter of the difference of the spectral value of the match point
Calculate the similarity of the curve of spectrum to be retrieved and every curve of spectrum for needing to retrieve；
The normalized value of the calculated result of similarity is obtained, and according to Sequential output search result from big to small；
Wherein, the SARI is spectral absorption reflection index, and the X is the first preset value, and the N is the second preset value,
And N > the X.
Further, described according to X on every curve of spectrum fixed SARI characteristic points, every is obtained by calculating
N number of specific steps always put on the curve of spectrum are as follows:
Step 1: sequentially straight line connects the first point of every curve of spectrum, first point in X point, second in X point
A point ..., X point and tail point in X point, formed X+1 line segment；
Step 2: calculating all the points on the corresponding curve of spectrum of each line segment to the distance of each line segment, and compare institute
Obtained all distance values retain that maximum point of distance value, and form X+1 point on the corresponding curve of spectrum；
Step 3: step 1 and step 2 are repeated, until light described in road according to the X+1 point formed on the curve of spectrum
The points retained on spectral curve are N.
Further, by calculate on each curve selection retained apart from maximum point when, if multiple points away from
From being equidistant to corresponding line segment, then retain one of them apart from maximum point by following steps:
Each point is obtained to the distance value of the corresponding line segment and the length value of corresponding line segment；
It calculates and the distance value of each point to the corresponding line segment is worth ratio with the corresponding line segment length
Size, and obtain the smallest point of ratio；
Retain the point on the smallest curve of spectrum of the ratio.
Further, according to N number of total point on every curve of spectrum, spectrum to be retrieved is obtained by the method that wave band compares
The step of match point of curve and every curve of spectrum for needing to retrieve are as follows:
The wave where the curve of spectrum to be retrieved and the point for needing to be retained on the every curve of spectrum retrieved is calculated by wave band
Section, if the current signature point on each wave band, on the same wave band, the two points match.
Further, square of the difference of the spectral value for calculating the match point, and according to the spectrum of the match point
The step of similarity for every curve of spectrum that square the calculating curve of spectrum to be retrieved and the needs of the difference of value are retrieved are as follows:
The similarity of the curve of spectrum to be retrieved and every curve of spectrum for needing to retrieve
Wherein, C is the number of match point, RA_iFor i-th point on the curve of spectrum to be retrieved of spectral value, RB_iTo need to examine
I-th point of spectral value on the rope curve of spectrum.
Further, the spectral absorption reflection indexWherein M is that corresponding spectrum is bent
SARI characteristic point on line, ρMFor the reflectivity of M point, ρS1、ρS2Respectively indicate the left shoulder point of M point on the curve of spectrum, right shoulder point it is anti-
Rate is penetrated, d is asymmetry parameter.
Further, the asymmetry parameterW=λS2-λS1；
Wherein, λS1、λS2And λMRespectively indicate the wavelength of S1, S2 and M point.
Further, a plurality of curve of spectrum that the curve of spectrum for obtaining atural object to be retrieved and needs are retrieved it is specific
Method are as follows: directly extract and obtain from the high spectrum image of corresponding atural object.
In above technical scheme, using X on the curve of spectrum fixed SARI characteristic points, every light is obtained by calculating
N number of total point on spectral curve, and according to N number of total point, it obtains the curve of spectrum to be retrieved by the method that wave band compares and needs to examine
The match point of every curve of spectrum of rope, so by the difference for the spectral value for calculating the match point square and obtain to be retrieved
The similarity of the curve of spectrum and every curve of spectrum for needing to retrieve accurately is describing spectrum based on spectral absorption reflection index
On the basis of the physical features of curve, it can effectively simplify the curve of spectrum, and effectively improve the retrieval precision of the curve of spectrum.
Detailed description of the invention
Fig. 1 is SARI Absorption Characteristics point valley point of the present invention and two shoulder schematic diagrames；
Fig. 2 is SARI Absorption Characteristics point peak dot and two shoulder schematic diagrames of the present invention；
Fig. 3 is the SARI characteristic point schematic diagram that the present invention extracts on the curve of spectrum in forest land；
Fig. 4 is the SARI characteristic point schematic diagram that the present invention extracts on the curve of spectrum of wheat；
Fig. 5 is the simplification Dependence Results schematic diagram in the forest land that algorithm according to the present invention obtains；
Fig. 6 is the simplification Dependence Results schematic diagram for the wheat that algorithm according to the present invention obtains；
Fig. 7 be the soybean of the Indian remote sensing trial zone in the Indiana, USA northwestward of in June, 1992 shooting, wheat and
The curve of spectrum schematic diagram of three kinds of forest land atural object；
Fig. 8 is the Hyperspectral imaging search method workflow schematic diagram of an embodiment of the present invention.
Specific embodiment
In order to which the technical problems, technical solutions and beneficial effects solved by the present invention is more clearly understood, below in conjunction with
Accompanying drawings and embodiments, the present invention will be described in further detail.It should be appreciated that specific embodiment described herein is only used
To explain the present invention, it is not intended to limit the present invention.
In conjunction with shown in Fig. 1 and Fig. 2, the curve of spectrum of different atural objects absorbs Wave crest and wave trough shape, position, width, depth
And the attributes such as symmetry are also different, can use spectral absorption reflection index SARI accordingly to describe the identification of curve of spectrum spy
Sign, it inherently expresses the variation characteristic that object spectrum absorbs reflection coefficient.
The curve of spectrum can be extracted directly from the high spectrum image of corresponding atural object and be obtained, spectral absorption reflectance signature
Two shoulders S1 and S2 can be absorbed by spectral absorption valley point or peak point M and curve of spectrum polishing wax to form, H indicates spectrum in figure
Depth is absorbed, is to absorb valley point M to form at a distance from " non-absorbing baseline " with two shoulders, enables ρS1、ρS2And ρMRespectively indicate suction
Left shoulder point S1 is received, right shoulder point S2 is absorbed and absorbs the reflectivity of point M, λS1、λS2And λMIt respectively indicates left shoulder point S1, absorb right shoulder point
S2 and the wavelength for absorbing point M, then absorption bands width is W=λS2-λS1, asymmetry parameter isAbsorb point of shoulder
Reflection differences are as follows: △ ρS=ρS2-ρS1, then " non-absorbing baseline " equation are as follows:
W·ρ-△ρSλ=ρS1-△ρS·λS1
Above formula expresses the spectral contribution and spectrum behavior of no spectral absorption characteristics atural object.Therefore spectral absorption reflection index
SARI may be defined as the spectral value of absorption or reflection position and the ratio of corresponding baseline value:
As shown in figure 8, including the following steps: the embodiment of the invention provides a kind of Hyperspectral imaging search method
The a plurality of curve of spectrum that the curve of spectrum and needs for obtaining atural object to be retrieved are retrieved, and in every curve of spectrum wave
X SARI characteristic point is extracted respectively in the position of peak and trough；
In conjunction with shown in Fig. 3 and Fig. 4, Fig. 3 and Fig. 4 are respectively the spectrum of forest land and wheat song in AVIRIS Hyperspectral imaging
Line is extracted 10 SARI characteristic points respectively on wave crest and trough, and X value here is 10.The value of X is according to need in practice
It is manually set, the maximum value of setting cannot be greater than a numerical value of wave crest and trough on the curve of spectrum.
According to X on every curve of spectrum fixed SARI characteristic points, by calculating the N obtained on every curve of spectrum
A total point, N number of total point on every curve of spectrum include X described on the curve fixed SARI characteristic points；
According to N number of total point on every curve of spectrum, the curve of spectrum to be retrieved is obtained by the method that wave band compares and is needed
The match point of every curve of spectrum being retrieved；
Square of the difference of the spectral value of the match point is calculated, and according to square meter of the difference of the spectral value of the match point
Calculate the similarity of the curve of spectrum to be retrieved and every curve of spectrum for needing to retrieve；
The normalized value of the calculated result of similarity is obtained, and according to Sequential output search result from big to small；
Wherein, the X is the first preset value, and the N is the second preset value, and the N > X.
The present invention is directed to the defect of existing DP algorithm, proposes a kind of new innovatory algorithm, is not necessarily to given threshold
Reach and not only retain enough points, but also the purpose of a plurality of tracing pattern can be kept simultaneously.Its basic step is as follows:
(1) the first point S and last point E on the curve of spectrum are connected, all the points and are found to the distance of line segment SE on calculated curve
The point O of maximum distance, which is retained.
(2) using point O as boundary, connect SO and OE, curve is divided into two parts, on calculated curve two parts point to correspondence line segment
The distance of SO and OE, wherein maximum distance point P is retained for selection.
(3) again using O and P as boundary, curve is divided into three parts SO, OP, PE, repeats the operation of (2), and so on, every time
Only retain one apart from maximum point, until the points of reservation reach set requirement.
The basis of design of the points N finally retained: the points of reservation are more, and calculation amount is with regard to greatly a bit；The points of reservation are few, meter
Calculation amount is with regard to a little bit smaller.But can not be too small, otherwise precision will be very poor, so the points specifically retained are needed according to practical feelings
Condition is manually set.
(4) when selection is apart from maximum point, if two points or multiple points are equidistant, calculate these points to pair
The distance of line segment and the ratio of line segment length are answered, the point for selecting ratio small is retained.
As long as can be seen that specify the number of point for needing to retain, inventive algorithm can Complete Convergence, to guarantee
The effect that curve simplifies.Simultaneously because without threshold value, regardless of the form of curve itself, inventive algorithm can be same etc.
Reason, simplification curve all can correctly reflect the morphological feature of curve, solve traditional algorithm asking in a plurality of curve while simplification
Topic.In addition in calculating process, without computing repeatedly the point on curve to the distance for corresponding to line segment, it is only necessary to retain the last time
Distance is recalculated in curve section where point, and the distance of curve to the corresponding line segment in other sections can direct basis
The calculated result of last time obtains, and therefore, can efficiently reduce calculation amount.
Preferably, for the embodiment of the present invention, for the curve of spectrum, the point of certain description spectral signatures needs to protect
It stays, such as certain wave crests, trough etc., the process that the present invention is simplified using SARI characteristic point come constraint curve, first extraction curve
Several upper SARI characteristic points regard these points as initial value, are used in improvement DP algorithm proposed by the present invention, so that simplifying bent
Line retains the Absorption Characteristics of original spectrum curve as much as possible.Therefore, special according to X on every curve of spectrum fixed SARI
Point is levied, by calculating the N number of specific steps always put obtained on every curve of spectrum are as follows:
Step 1: sequentially straight line connects the first point of every curve of spectrum, first point in X point, second in X point
A point ..., X point and tail point in X point, formed X+1 line segment；
Step 2: calculating all the points on the corresponding curve of spectrum of each line segment to the distance of each line segment, and compare institute
Obtained all distance values retain that maximum point of distance value, and form X+1 point on the corresponding curve of spectrum；
Step 3: step 1 and step 2 are repeated, until light described in road according to the X+1 point formed on the curve of spectrum
The points retained on spectral curve are N.
As shown in connection with fig. 3, the curve of spectrum in Fig. 3 is extracted 20 SARI characteristic points, including 10 SARI absorption peaks
Value point and 10 SARI absorb valley point, are simplified using present invention method described above, as a result as shown in Figures 5 and 6
Simplification curve, total points that wherein Fig. 5 retains are 50, that is, the N is set as 50；Retain in Fig. 6 it is total points be
100, i.e., N value here is 100, afterwards as can be seen that the method for the present invention can completely retain the whole shape of curve compared with Fig. 3
State feature.
Preferably, in above-mentioned steps, by calculating when selection is retained apart from maximum point on each curve, if
The distance of multiple points is equidistant to corresponding line segment, then retains one of them apart from maximum point by following steps:
Each point is obtained to the distance value of the corresponding line segment and the length value of corresponding line segment；
It calculates and the distance value of each point to the corresponding line segment is worth ratio with the corresponding line segment length
Size, and obtain the smallest point of ratio；
Retain the point on the smallest curve of spectrum of the ratio.
Further, according to N number of total point on every curve of spectrum, spectrum to be retrieved is obtained by the method that wave band compares
The step of match point of curve and every curve of spectrum for needing to retrieve are as follows:
The wave where the curve of spectrum to be retrieved and the point for needing to be retained on the every curve of spectrum retrieved is calculated by wave band
Section, if the current signature point on each wave band, on the same wave band, the two points match.
Further, square of the difference of the spectral value for calculating the match point, and according to the light of the match point
The step of similarity for every curve of spectrum that square the calculating curve of spectrum to be retrieved and the needs of the difference of spectrum are retrieved are as follows:
The similarity of the curve of spectrum to be retrieved and every curve of spectrum for needing to retrieve
Wherein, C is the number of match point, RA_iFor i-th point on the curve of spectrum to be retrieved of spectral value, RB_iTo need to examine
I-th point of spectral value on the rope curve of spectrum.
The similarity of the curve of spectrum and the curve of spectrum to be retrieved retrieved is needed to be normalized by every of above-mentioned calculating,
And according to Sequential output search result from big to small, retrieval can be completed.
In order to verify the validity of search method of the present invention, as shown in fig. 7, at Purdue university remote sensing images
The AVIRIS Hyperspectral imaging in reason laboratory is tested, and raw video is derived from the indiana ,US of in June, 1992 shooting
A part of the Indian remote sensing trial zone in the state northwestward has 224 wave bands, and spectral region is 0.4 to 2.45 μm, spatial resolution
It is 20 meters, image size is that 145*145 pixel obtains 183 after removing water absorption bands and biggish wave band affected by noise
Wave band is as research object.
With reference to true line map, the search result in three classes easily mixed atural object (soybean, wheat and forest land) is provided, by these three types of ground
Three pixels of the every class of object are analyzed, while being compared with two methods of conventional SID and SAM, as a result such as 1~table of table
Shown in 3, wherein A1, A2, A3 indicate soybean, and B1, B2, B3 and C1, C2, C3 respectively indicate wheat and forest land, and context of methods retains
50 points on the curve of spectrum.
Wherein, oblique runic indicates the result of retrieval error.
Can be seen that method of the invention from above-mentioned Experimental comparison can correctly distinguish different atural object completely, and
But there are many mistakes in two methods generally used now, this also illustrates the validity of search method of the present invention.
Using technical solution of the present invention, the object of the curve of spectrum can more accurately be described based on spectral absorption reflection index
Feature is managed, equal number of characteristic point can be adaptively obtained based on DP algorithm is improved, with conventional retrieval method using all
Wave band carries out retrieval and compares, and high-precision retrieval can be realized by simplifying curve in the present invention.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the invention
Made any modifications, equivalent replacements, and improvements etc., should all be included in the protection scope of the present invention within mind and principle.
Claims (7)
1. a kind of Hyperspectral imaging search method, it is characterised in that: including,
The a plurality of curve of spectrum that the curve of spectrum and needs for obtaining atural object to be retrieved are retrieved, and in every curve of spectrum wave crest and
X SARI characteristic point is extracted respectively in the position of trough；
It is N number of total on every curve of spectrum of acquisition by calculating according to X on every curve of spectrum fixed SARI characteristic points
Point, N number of total point on every curve of spectrum include X described on the curve fixed SARI characteristic points；
According to N number of total point on every curve of spectrum, obtains the curve of spectrum to be retrieved by the method that wave band compares and need to examine
The match point of every curve of spectrum of rope；
Calculate square of the difference of the spectral value of the match point, and according to the difference of the spectral value of the match point square calculate to
The similarity of the retrieval curve of spectrum and every curve of spectrum for needing to retrieve；
The normalized value of the calculated result of similarity is obtained, and according to Sequential output search result from big to small；
Wherein, the SARI is spectral absorption reflection index, and the X is the first preset value, and the N is the second preset value, and institute
State N > X；
Square of the difference of the spectral value for calculating the match point, and according to square meter of the difference of the spectral value of the match point
The step of calculating the similarity for every curve of spectrum that the curve of spectrum to be retrieved and needs are retrieved are as follows:
The similarity of the curve of spectrum to be retrieved and every curve of spectrum for needing to retrieve
Wherein, C is the number of match point, RA_iFor i-th point on the curve of spectrum to be retrieved of spectral value, RB_iTo need to retrieve light
I-th point of spectral value on spectral curve.
2. Hyperspectral imaging search method according to claim 1, it is characterised in that: described according on every curve of spectrum
X fixed SARI characteristic points, by calculating the N number of specific steps always put obtained on every curve of spectrum are as follows:
Step 1: sequentially straight line connects the first point of every curve of spectrum, first point in X point, second in X point
Point ..., X point and tail point in X point, formed X+1 line segment；
Step 2: calculate all the points on the corresponding curve of spectrum of each line segment to each line segment distance, and relatively obtained by
All distance values, retain that maximum point of distance value, and M+1 point is formed on the corresponding curve of spectrum；
Step 3: step 1 and step 2 are repeated according to the X+1 point formed on the curve of spectrum, until the spectrum is bent
The points retained on line are N.
3. Hyperspectral imaging search method according to claim 2, it is characterised in that: by calculating on each curve
When selection is retained apart from maximum point, if the distance of multiple points being equidistant to corresponding line segment, is protected by following steps
Stay one of them apart from maximum point:
Each point is obtained to the distance value of the corresponding line segment and the length value of corresponding line segment；
Calculate and each point to the corresponding line segment distance value and the ratio size of the corresponding length along path angle value,
And obtain the smallest point of ratio；
Retain the point on the smallest curve of spectrum of the ratio.
4. Hyperspectral imaging search method according to claim 1, it is characterised in that: according to the N on every curve of spectrum
A total point obtains the match point of the curve of spectrum to be retrieved and every curve of spectrum for needing to retrieve by the method that wave band compares
Step are as follows:
The wave band where the curve of spectrum to be retrieved and the point for needing to be retained on the every curve of spectrum retrieved is calculated by wave band, such as
Current signature point on each wave band of fruit is on the same wave band, then the two point matchings.
5. Hyperspectral imaging search method according to claim 1, it is characterised in that: the spectral absorption reflection indexWherein M is SARI characteristic point on the corresponding curve of spectrum, ρMFor the reflectivity of M point, ρS1、ρS2Point
Not Biao Shi on the curve of spectrum the left shoulder point of M point, right shoulder point reflectivity, d is asymmetry parameter.
6. Hyperspectral imaging search method according to claim 5, it is characterised in that: the asymmetry parameterW=λS2-λS1；
Wherein, λS1、λS2And λMRespectively indicate the wavelength of S1, S2 and M point.
7. Hyperspectral imaging search method according to claim 1, it is characterised in that: the light for obtaining atural object to be retrieved
The a plurality of curve of spectrum that spectral curve and needs are retrieved method particularly includes: directly extracted from the high spectrum image of corresponding atural object
It obtains.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201511020652.2A CN105678229B (en) | 2015-12-29 | 2015-12-29 | A kind of Hyperspectral imaging search method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201511020652.2A CN105678229B (en) | 2015-12-29 | 2015-12-29 | A kind of Hyperspectral imaging search method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105678229A CN105678229A (en) | 2016-06-15 |
| CN105678229B true CN105678229B (en) | 2019-03-05 |
Family
ID=56298083
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201511020652.2A Active CN105678229B (en) | 2015-12-29 | 2015-12-29 | A kind of Hyperspectral imaging search method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105678229B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111324805B (en) * | 2018-12-13 | 2024-02-13 | 北京搜狗科技发展有限公司 | Query intention determining method and device, searching method and searching engine |
| CN110974191A (en) * | 2019-12-18 | 2020-04-10 | 深圳市精气生物技术有限公司 | Radial artery waveform data identification algorithm |
| CN113792082A (en) * | 2021-09-02 | 2021-12-14 | 深圳创景数科信息技术有限公司 | Fabric component retrieval method based on database |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101916377A (en) * | 2010-07-15 | 2010-12-15 | 北京航空航天大学 | Method for intelligently identifying mineral information based on spectral information |
| CN102609711A (en) * | 2012-02-21 | 2012-07-25 | 核工业北京地质研究院 | Information extraction method applicable to hyperspectral image |
| CN104331482A (en) * | 2014-11-07 | 2015-02-04 | 武汉大学 | Method and system for hyper-spectrum matching based on feature index |
-
2015
- 2015-12-29 CN CN201511020652.2A patent/CN105678229B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101916377A (en) * | 2010-07-15 | 2010-12-15 | 北京航空航天大学 | Method for intelligently identifying mineral information based on spectral information |
| CN102609711A (en) * | 2012-02-21 | 2012-07-25 | 核工业北京地质研究院 | Information extraction method applicable to hyperspectral image |
| CN104331482A (en) * | 2014-11-07 | 2015-02-04 | 武汉大学 | Method and system for hyper-spectrum matching based on feature index |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105678229A (en) | 2016-06-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhong et al. | Iterative support vector machine for hyperspectral image classification | |
| CN105809140A (en) | Method and device for extracting surface water body information based on remote sensing model | |
| CN110648347A (en) | Coastline extraction method and device based on remote sensing image | |
| CN105678229B (en) | A kind of Hyperspectral imaging search method | |
| CN104181170B (en) | fruit appearance detection method based on spectral image analysis | |
| CN105528580B (en) | A kind of EO-1 hyperion Curve Matching method based on absorption peak feature | |
| CN108399424A (en) | A kind of point cloud classifications method, intelligent terminal and storage medium | |
| CN105956544A (en) | Remote sensing image road intersection extraction method based on structural index characteristic | |
| CN103500343A (en) | Hyperspectral image classification method based on MNF (Minimum Noise Fraction) transform in combination with extended attribute filtering | |
| CN114398948A (en) | Multispectral image change detection method based on space-spectrum combined attention network | |
| Li et al. | An object-oriented CNN model based on improved superpixel segmentation for high-resolution remote sensing image classification | |
| CN106875407A (en) | A kind of unmanned plane image crown canopy dividing method of combining form and marking of control | |
| CN106529472A (en) | Target detection method and apparatus based on large-scale high-resolution and high-spectral image | |
| CN116310471A (en) | Hyperspectral image classification method | |
| CN116630225A (en) | Method and device for identifying underwater foundation damage of railway bridge and processing equipment | |
| CN107103295A (en) | Optical remote sensing image cloud detection method of optic | |
| CN108007880B (en) | Method for establishing tree species spectral curve polygon area index | |
| Ghiyamat et al. | Airborne hyperspectral discrimination of tree species with different ages using discrete wavelet transform | |
| CN111882573A (en) | Cultivated land plot extraction method and system based on high-resolution image data | |
| Lin et al. | In defense of iterated conditional mode for hyperspectral image classification | |
| Abraham et al. | Unsupervised building extraction from high resolution satellite images irrespective of rooftop structures | |
| CN114399807A (en) | Cross-spectrum face recognition method based on image conversion and monitoring equipment | |
| CN113205143A (en) | Multi-scale superpixel hyperspectral remote sensing image classification method based on space-spectrum coupling characteristics | |
| CN110188821A (en) | The hyperspectral image classification method of multi-parameter holding edge filter and multiple features study | |
| CN108984601A (en) | A kind of image search method and system based on probability histogram area similarity |
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 |