Macro-mesoscopic modeling method for multi-source data fusion by using attention mechanism to extract features

The invention discloses a multi-source data fusion macro-mesoscopic modeling method for extracting features by using an attention mechanism, and the method comprises the steps: collecting multi-source data through multiple devices, and then achieving the multi-modal modeling of point cloud data collected by a laser scanner and image data obtained by a linear array camera through a feature extraction algorithm based on the attention mechanism. The method effectively solves the problem that an existing three-dimensional scanner is high in cost and long in consumed time in three-dimensional modeling of a hydropower station outgoing line field, and has the advantages that a live-action three-dimensional model can be constructed through multi-modal modeling, the scanning frequency of the three-dimensional scanner is reduced, the cost is reduced, the consumed time of data acquisition is shortened, the precision is improved through mutual verification of multi-source data, and the real-scene modeling efficiency is i