Abstract: The behavior that the crowded areas tourism is quite sudden and randomicity, key to reflect the characteristics of susceptible to keep out, there is monitoring blind Angle, the traditional approach to monitoring recognition ability, lack of independent behavior is affected by the block, on some suspected behavior lack of accurate identification, puts forward a kind of based on Internet of things crowded areas tourism trampling risk image monitoring method, analyzes the Internet monitoring platform, to extract the crowded areas tourism trampling risk Hu moment invariant features of the image and the input of RBF neural network learning and classification, crowded areas for various types of tourism trampling behavior, will test samples in the input of RBF neural network model is established to test, and the RBF network output corresponding test sample trample risk image is the tourism concentration areas, so as to realize tourism cluster trample image monitoring risk.The simulation results show that the proposed method in crowded areas tourism trampling risk monitor image, accuracy and efficiency than traditional methods have greatly improved.