Intelligent terrain recognition method fusing pixel-level image segmentation and reinforcement learning

The invention provides a terrain intelligent identification method fusing pixel-level image segmentation and reinforcement learning, and the method comprises the following steps: S1, building an unstructured BIM three-dimensional terrain model based on a BIM technology by means of a digital three-dimensional design platform; classifying the BIM three-dimensional topographic map, arranging topographic edge information tags, and obtaining various types of topographic target data sets; s2, separating and identifying various terrains in a target scene by adopting an image segmentation task type convolutional neural network, and obtaining a terrain segmentation map; and S3, inputting the original image and the terrain segmentation map into a reinforcement learning module, and outputting a finer-grained terrain segmentation map by the evaluation network. And S4, visualizing an algorithm result, and providing an external interface. And the specific terrain condition of the planning area can be analyzed in a finer-gr