Hyperspectral processing method and waveband selection method

Abstract

The invention relates to a hyperspectral processing method and a waveband selection method. A hyperspectral classification model is trained by using a first training data set which is a full-band hyperspectral training data set, each training sample in the first training data set is input into the trained hyperspectral classification model, a second training data set for band selection is acquiredthrough a hyperspectral classification model, the second training data set is input into a waveband selection model for training, the waveband selection model outputs a selected waveband sequence number set, a corresponding waveband is extracted from the first training data set according to the selected waveband sequence number set to form a third training data set, and the hyperspectral classification model is trained by using the third training data set. According to the method, the purpose of hyperspectral band selection can be achieved, the band selected by the method is favorable for being matched with the hyperspectral classification model, and when the finally obtained hyperspectral classification model is applied, the precision can be effectively guaranteed.

Description

technical field [0001] The invention relates to the field of hyperspectral processing, in particular to a hyperspectral processing method and a band selection method. Background technique [0002] Hyperspectral imaging is an optical detection technology that divides a given wavelength interval into narrow and dense bands for imaging in the spectral dimension. Expanding the image information obtained by each narrow band in the spectral dimension can greatly increase the amount of information. Compared with multispectral imaging, which only cuts out several broad bands for spectral detection, the number of bands generated by hyperspectral imaging can reach hundreds or even thousands, so as to have higher spectral resolution to obtain more specific spectral details. Therefore, Even though the cost is high, there are still a wide range of application cases in defense and monitoring, food safety, medical diagnosis, geological exploration, agricultural testing, drug screening and ...

AI Technical Summary

Problems solved by technology

The ranking-based band selection method quantifies the importance of each band, and selects the first few bands in order of importance. The difficulty of this type of method is that the bands selected by quantitative indicators may still have a large overlap with each other. High redundancy; search-based band selection method, due to the exponential relationship betwe

Images