Significance waveband selection method based on structure similarity

Abstract

The invention relates to a significance waveband selection method based on the structure similarity. The method comprises following steps of step 1, using high-spectral remote sensing image data subjected to de-noising processing; step 2, by selecting the significance difference degrees of a target region in each waveband, ranking the full wavebands of an image from high to low; step 3, based on the structure similarity, carrying out screening in pairs on the ranked waveband sets, and removing wavebands with the great similarity and low significance; and step 4, selecting multiple front wavebands of the screened waveband sets, thereby finishing waveband selection. According to the invention, the difference of the target region in each waveband is calculated by use of distance significanceexchange, thereby ranking all wavebands; wavebands with the quite great similarity are screened through the structure similarity; and finally, the front wavebands after the screening are selected. Inthis way, the provided method can be used for different targets, different waveband selections are obtained and following target detection effects are well improved.

Description

technical field [0001] The invention relates to the field of remote sensing, in particular to a band selection method for improving the subsequent target detection effect. Background technique [0002] Saliency refers to the degree of difference between the target and the background based on the human visual cognition system. When the target is easier to be noticed by the human eye, its brightness on the saliency map will be higher and more significant. The saliency measures the degree of difference between the target and the background. Obviously, the higher the saliency, the easier it is for the target to be detected. Visual saliency transformation is to transform the original image into a saliency distance map through the distance transformation function, and the pixel value of each point on the saliency map is the saliency of the point on the original image. [0003] If the hyperspectral image is to be put into application, it is first necessary to remove the redundant ...

AI Technical Summary

Problems solved by technology

However, the existing methods do not target specific target information and select bands that are more suitable for specific target detection