Linear spectral unmixing urban impervious surface remote sensing extraction method

Abstract

The invention provides a linear spectral unmixing urban impervious surface remote sensing extraction method comprising the steps that step one, remote sensing image material files are acquired and data preprocessing and water body masking are performed; step two, first data files are obtained from the remote sensing image material files through minimum noise fraction rotation; step three, the purity of each pixel in the first data files is calculated through the pixel purity index; step four, the pixels greater than the preset purity value are selected from the first data files through screening according to the preset purity value so that the pixels of high purity, i.e. second data files, are obtained; step five, pure pixels are selected by using the characteristic scatter two-dimensional model between different components of the second data files through combination of high-resolution images; step six, the proportion of each end member of each pixel is solved based on the end members of the step five; and step seven, mask processing is performed on low-albedo surface features in the unmixing result, and finally the processed low-albedo surface features and high-albedo surface features are added and then an impervious surface coverage graph is obtained.

Description

technical field [0001] The invention relates to the field of remote sensing image processing, in particular to a method for extracting urban impermeable surface remote sensing by linear spectrum unmixing. Background technique [0002] In the process of using remote sensing data to extract urban impervious surfaces, first select the appropriate data source, and after preprocessing such as geometric correction and atmospheric correction, the methods for classification and extraction of ground objects on remote sensing images include linear spectral unmixing method, supervised classification, Unsupervised classification, decision tree classification, artificial neural network, etc., the commonly used extraction method is the linear spectral unmixing method (LSMA). In the process of remote sensing to extract urban impervious surfaces, the traditional linear spectral unmixing method has two problems: First, in the forest area, there are a small number of impervious surfaces such ...

AI Technical Summary

Problems solved by technology

In the process of remote sensing to extract urban impervious surfaces, the traditional linear spectral unmixing method has two problems: First, in the forest area, there are a small number of impervious surfaces such as individual houses and other high-albedo features, which are easy to be ignored when selecting end members. Neglecting it will affect the accuracy of endmember selection; 2. In some areas, the pixel albedo is relatively low, such as vegetation, bare soil, etc., which have a greater impact on the low albedo part of the impermeable surface, and are prone to large-scale unmixing abnormal areas , when extracting the impermeable surface by linear spectral unmixing, it is necessary to use the threshold method to remove the abnormal area of ​​the extraction result

The selection of the threshold directly affects the extraction accuracy of the impermeable surface, and it is empirical, and the human factors are relatively large

Images