Hyperspectral feature selection method based on simulated annealing algorithm

Abstract

The invention discloses a hyperspectral feature selection method based on a simulated annealing algorithm, wherein the hyperspectral feature selection method comprises the following steps: step 1, formulating a coding mode according to spectral characteristics; step 2, selecting a target function according to spectral characteristics; step 3, initializing simulated annealing algorithm parameters,and starting to search for an optimal result; step 4, continuously comparing the values of the historical results and the current result, and continuing to perform the algorithm on the optimal value;and step 5, repeating the step 4 until the optimal result does not change any more in the preset stable algebra or reaches the preset algebra. The method has the advantages that the differential solution is received with a certain probability to jump out of the local optimal solution; the global optimal solution is achieved to complete hyperspectral feature selection; and a large amount of manpower, material resources and financial resources can be saved.

Description

technical field [0001] The invention relates to the field of hyperspectral remote sensing, in particular to a hyperspectral feature selection method based on a simulated annealing algorithm. Background technique [0002] Hyperspectral remote sensing, also known as imaging spectroscopy, originates from multispectral remote sensing technology, which can use dozens to hundreds of continuous and narrow spectral bands for imaging. With the continuous improvement of hyperspectral remote sensing technology, the application fields are constantly expanding, mainly involving atmospheric detection, environmental monitoring, earth resources, geological science, ecological science, hydrological science and military fields. [0003] The reflectance information of each spectral band contained in each pixel point data in the hyperspectral image technology image can be obtained after preprocessing. Therefore, by extracting the spectral reflectance information, it is possible to identify the...

AI Technical Summary

Problems solved by technology

The features selected by the band ranking method may have great redundancy, resulting in little meaningful information complementary to the existing information.

Using the band clustering algorithm, 2) often because of the selection of the clustering approximate center point, it will inevitably lead to the deviation between the selected band and the real center point

And this method also has the problem of not considering the spectral continuity of HSI

No matter which method is used above, it needs to consume a lot of manpower and power, and the efficiency is low, and it is easy to fall into the local optimal solution.

Images