Extraction of Photovoltaic Electric Farms from Multi-source Remote Sensing Images Based on Object-based Image Analysis

Yan-fu LI,Yong LIU

DOI: https://doi.org/10.3969/j.issn.1672-5867.2016.03.020

2016-01-01

Abstract:As one of representative of new energy, photovoltaic electric farms are extending rapidly in China, especially in desert area of the western China.It isof realistic significance to extract photovoltaic electric farms automatically, quickly and accurately.In this artical, the photovoltaic electric farms at Ningxia Desert Photovoltaic Industry Park, Zhongwei City, China, were extracted through ob-ject-based image analysis ( OBIA)method through employing spectral and textural features in Landsat 8 OLI, WorldView II and GF-1 images.It is shown that OBIA is effective on extraction of photovoltaic electric farms from either middle or high resolution remote sensing images with high precision.For Worldview II and GF-1 images, solar cell arrays can be extracted firstly based on four fea-tures:RRI, Length/Width, dissimilarity and brightness, and then internal road and other entries can be extracted based on the con-text, geometry, spectral features.Rule sets constructed in this way is accessible, transferable and have fewer features.For Landsat 8 OLI images, photovoltaic electric farms can be extracted using RRI, NDBI, and brightness features.