The use of geospatial workflows to support automatic detection of complex geospatial features from high resolution images

A complex geospatial feature is spatially composed of member elementary features. The methods extracting elementary features from high resolution imageries are being mature. However, the detection of complex feature is still a nearly-blind point. This paper proposes an automatic approach using geospatial workflows to detect complex feature by identifying the spatial patterns like the contained member elementary features and the spatial relationships between member features and the surrounding environment. A prototype online system is realized for evaluation.