Unlock instant, AI-driven research and patent intelligence for your innovation.

Spectrum waveband selection method based on multi-modal fusion

A spectral band and multi-modal technology, applied in the field of hyperspectral image processing, can solve the problems of not being able to select hyperspectral images, large bands, and not fully considering the correlation and redundancy tradeoffs of spectral bands.

Active Publication Date: 2017-09-29
SHENYANG AEROSPACE UNIVERSITY
View PDF3 Cites 5 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] In view of this, the purpose of the present invention is to provide a spectral band selection method based on multimodal fusion, to solve the problem that the existing spectral band selection method cannot fully consider the trade-off between correlation and redundancy between spectral bands , cannot well select the band problem that is more useful for the classification of hyperspectral images

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Spectrum waveband selection method based on multi-modal fusion
  • Spectrum waveband selection method based on multi-modal fusion
  • Spectrum waveband selection method based on multi-modal fusion

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0068] S1. Receive hyperspectral image samples and category label information of the samples;

[0069] Take the indian data set as an example, input 145*145*200 hyperspectral image samples (the samples include 145*145 samples, 200 bands), and the corresponding category label information of the data set, a total of 1-16, Represents 16 sample classes.

[0070] S2. Fuse the spatial features and texture features in multiple spatial neighborhoods, and use the correlation measurement criterion of the bands to sort all the bands according to the correlation from low to high, and obtain the band sequence 1;

[0071] Specific steps are as follows:

[0072] S201. Take the five neighborhoods of 3*3, 5*5, 7*7, 9*9, and 11*11 of each sample. First, input all samples under each band under the neighborhood of 3*3 Filter in the LBP filter function, and output the filtered value of 145*145 samples of each band in the neighborhood of 3*3. Taking band 1 as an example, input 145*145 pixels in b...

Embodiment 2

[0085] Select 80% of the samples of each category on the indian data set for training, use this method to process image samples, and finally classify through the SVM classifier, which is higher than the correct rate of selecting all bands and then using the SVM classifier to classify 5%, which proves the effectiveness of the method in Example 1, and also shows that the selected 120 bands are bands with more useful information that are more conducive to the classification of hyperspectral images.

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a spectrum waveband selection method based on multi-modal fusion. The method comprises steps that a high spectral image sample and the class mark information are received; space characteristics and texture characteristics of multiple space neighborhoods are fused, all the wave bands are ordered through utilizing wave band correlation tolerance criteria according to correlation to acquire a wave band sequence 1; all the wave bands are ordered respectively based on a largest value, an average value and a variance value of each row of a correlation matrix to acquire wave band sequences 2,3 and 4; weighted sum of ordering sequence numbers of each wave band in the four wave band sequences is carried out to acquire a final ordering sequence number of each wave band; n wave bands having the ordering sequence numbers in the front are taken as selection wave bands. The method is advantaged in that the abundant multi-modal information is considered, supervised strategies are employed to calculate wave band correlation, not only can wave band priority be considered, redundancy among the selected wave bands is further considered, and the spectrum wave bands facilitating high spectral image classification can be acquired.

Description

technical field [0001] The invention relates to the field of hyperspectral image processing, and in particular provides a method for selecting spectral bands based on multimodal fusion. Background technique [0002] Hyperspectral remote sensing images have multiple modal information such as texture, correlation, and spectrum. Among them, the rich spectral information can reflect the diagnostic spectral features that distinguish different substances, so that hyperspectral remote sensing can detect more ground object information. It has greatly improved human's cognitive ability to the objective world. [0003] In the current research, more is to choose the relevant metric function as the metric criterion for spectral sorting, and then sort the spectral information based on the metric function. Among them, the band selection based on spectral sorting only considers the band priority of a given task, and often ignores In order to select the possible redundancy between bands, r...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06K9/62G01N21/25
CPCG01N21/25G06F18/2323G06F18/2411
Inventor 李照奎黄林赵亮王岩刘翠微张德园石祥滨徐一民
Owner SHENYANG AEROSPACE UNIVERSITY