Multispectral remote sensing image variation detection method based on spectral reflectivity variation analysis

Abstract

The invention discloses a multispectral remote sensing image variation detection method based on spectral reflectivity variation analysis. The method mainly solves the problem of whole luminance difference sensitivity between different time phase images in the prior art. The method comprises the steps that (1) two time phase multispectral image sets registered in the same region are input, and subjected to wiener filtering, noise removal and normalization processing; (2) the processed image sets are converted into relative ground object spectral reflectivity image sets; (3) the variance and the modulus of a spectral reflectivity variation are calculated, a variogram and a modulus value figure of the spectral reflectivity variation are obtained and enhanced respectively, and an enhanced variogram and an enhanced modulus value figure are obtained; and (4) the enhanced variogram and the enhanced modulus value figure are divided, and a variation detection result figure is obtained by fusing divided binary images. The method requires no manual participation, is high in detection accuracy, and can be used for land utilization and cover monitoring, vegetation cover monitoring, and water resource and mineral resource monitoring.

Description

technical field [0001] The invention belongs to the technical field of optical remote sensing image processing, and relates to unsupervised change detection when there is a large overall brightness difference between two-temporal multi-spectral remote sensing images, and can be used for land use, vegetation coverage, water resources, mineral resources, etc. Change monitoring. Background technique [0002] With the openness of remote sensing image data and the continuous improvement of processing technology, the use of multi-temporal remote sensing image data to monitor changes in land use, vegetation cover, water resources, and mineral resources has become more and more popular. It is of great significance in many aspects such as water conservancy and mineral resources. Since the changes of different surface object types may be reflected in different spectral ranges, and multispectral remote sensing image data has multiple receiving frequency bands from visible light to inf...

AI Technical Summary

Problems solved by technology

Compared with principal component analysis, independent component analysis has certain robustness to registration error and brightness difference, but independent component analysis transformation makes different types of change regions distributed on several independent component images, which usually needs to rely on experience and prior knowledge. Manually select certain component images for subsequent classification based on empirical knowledge, which makes it difficult to completely detect change information

Images