The energy transition to more sustainable forms is currently ongoing worldwide, because of the environmental impacts produced by the non-renewable energy sources employed in the last decades. Among the main alternatives, wind plays a key role and, nowadays, innovative instruments, such as small-scale turbines allow for installation of wind turbines in urban areas. Their energy potential assessment requires high-accuracy simulations of the turbulent flows in the urban canopy layer, which, in turn, require detailed information about the geometrical properties of the basic element to classify urban surfaces, i.e., the urban canyon, often not available. In this work, we propose a novel automatic method, based on Voronoi graph, to univocally identify urban canyons and to extract their geometrical parameters from online available GIS (Geographic Information System) data, and test it on four European cities that differ in size, story and location. Results show the capability of the method to identify the single urban canyon and to properly extract its geometrical parameters, which tend to assume similar values for the largest cities. Moreover, we first attempt to propose and test some curves to generally describe the data probability distribution, which may be useful for turbulence simulations for urban wind energy assessment and planning. The best results are found for the canyon aspect ratio.