Semi-automatic classification method for timing sequence remote sensing images based on continuous wavelet transforms

Abstract

The invention relates to a semi-automatic classification method for timing sequence remote sensing images based on continuous wavelet transforms. The semi-automatic classification method first builds a plurality of within-a-year sequential variation initial maps of known surface features, conducts continuous wavelet transforms on the within-a-year sequential variation initial maps, and then obtains wavelet coefficient maps and on this basis builds wavelet variance maps based on time dimension and wavelet variance maps based on scale dimension. Meanwhile, wavelet variance maps based on time dimension and wavelet variance maps based on scale dimension of all the research units in the whole research area are built, and then the optimum time domains and the optimum scale domains for images classification are respectively confirmed according to an inter-class otherness maximization principle of the wavelet variance maps of known surface features, and finally semi-automatic classification of remote sensing images is achieved through building a comprehensive evaluation system. The semi-automatic classification method can effectively draw the characteristics of timing sequence remote sensing images in time dimension and scale dimension and has the advantages of being less in reliable on prior knowledge, good in robustness, high in classification accuracy, high in the degree of automation and the like.

Description

technical field [0001] The invention relates to the technical field of remote sensing image information processing, in particular to a semi-automatic classification method for time-series remote sensing images based on continuous wavelet transform. Background technique [0002] At present, satellite remote sensing systems such as MODIS can provide daily remote sensing image data covering the whole world, providing a detailed data basis for monitoring the characteristics of land cover changes. How to reasonably use the timing information of remote sensing images for automatic and semi-automatic classification is a very important task. Remote sensing image classification methods can be roughly divided into two categories: the first category is the traditional algorithm based on the idea of ​​spatial clustering, many commonly used algorithms belong to this type, such as maximum likelihood discriminant method, neural network classification method, fuzzy classification method me...

AI Technical Summary

Problems solved by technology

The disadvantage of the first type of method is that it is very dependent on the experience of the user and the quality of the single-phase image data itself, and due to the limited dimension of information, the phenomenon of the same object with different spectra and different objects with the same spectrum is very common, and has become an in-depth application. The biggest bottleneck; the advantages of the second type of method are obvious, but because the method is not proposed for a long time, it is in the development stage, and more in-depth research work is needed

However, each different type of ground object has its characteristics in multiple dimensions such as different scale dimensions and time dimensions. The above methods cannot extract these features completely and effectively, which affects the accuracy and efficiency of classification.

Images