基于稀疏解混的Landsat影像城区提取方法

With the continuous development of the city, the original vegetation, arable land and soil are replaced, to some extent affect the ecological diversity, hydrological system, climate and ecological cycle. Remote sensing technology as an effective means is widely used in urban monitoring, but also faces some problems. When extracting urban areas using the medium resolution Landsat image, there is often a phenomenon of mixed pixels. The traditional mixed pixel decomposition method is mainly used to extract the endmember, which can be divided into geometric method and statistical method based on the statistical method. However, with the spectral library, the mixed pixel decomposition becomes a semi-supervised method, and the image observation pixel can be expressed as a linear combination of a small amount of pure spectrum in a known spectral library, which is sparse unmixing. The SUnSAL, SUnSAL-TV and SSCSUn algorithms are applied to urban extraction of Landsat images. The results show that the SUnSAL-TV algorithm has the highest accuracy.