Method and device for selecting hyperspectral image band based on key band extraction
A hyperspectral image and band selection technology, which is applied in the field of image processing, can solve the problem of low detection accuracy of hyperspectral image anomalies, achieve the effect of reducing the amount of calculation, achieving good results, and improving the efficiency of band selection
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Example Embodiment
[0059] Example one
[0060] Step 1: Use the HFC (Harsanyi-Farrand-Chang) method to determine the number of endmembers.
[0061] (1) Calculate the covariance matrix K of the image data L×L And autocorrelation matrix R L×L .
[0062] (2) Calculate the eigenvalue sets of the covariance matrix and the autocorrelation matrix respectively, denoted as {λ 1 ≥λ 2 ≥…λ L }with Where L is the number of spectral bands.
[0063] (3) Calculate the approximate noise variance value of the lth band of the spectral image Among them, M×N represents the number of elements in the image.
[0064] (4) Calculate the probability density function
[0065] (5) Given false alarm probability P F ,according to with Find τ l value
[0066] (6) Satisfaction The number of eigenvalues is the number of bands sought.
[0067] Step 2: Use FNSGA (Fast New Simplex Growing Algorithm) monomorph growth algorithm to achieve endmember extraction and obtain endmember spectrum curve.
[0068] (1) For each pixel r in the image...
Example Embodiment
[0104] Example two
[0105] Steps 1 to 4 of the second embodiment are exactly the same as those of the first embodiment.
[0106] Step 5: Use the band selection method based on the maximum amount of information for the data corresponding to the key band subset to perform band selection.
[0107] Assuming that there are k bands in the key band set, the data of k bands are denoted as Φ 2d ={B 1 ,B 2 ,...,B k }∈R MN×k , Where MN=M×N. The number of bands required for band selection is set to num(num
[0108] For the C matrix obtained in step 4
[0109] (1) Calculate maxC=max(C), define the set
[0110] (2) Calculate the minimum value of each row element of the C matrix and record it as minR i , (I=1,2,3,...,k);
[0111] (3) Calculation And adjust SET=SET∪g;
[0112] (4) Modify the elements of the C matrix so that C(g,i)=C(g,i)=maxC;
[0113] (5) If the number of elements in SET is less than k-num, skip to step (3), otherwise skip to step (6);
[0114] (6) Remove the elements in SET from...
PUM

Abstract
Description
Claims
Application Information

- R&D Engineer
- R&D Manager
- IP Professional
- Industry Leading Data Capabilities
- Powerful AI technology
- Patent DNA Extraction
Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.
© 2023 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap