SPATIAL STRUCTURE PREDICTION BASED ON HIDDEN MARKOV CONTOUR TREE MODEL

In order to realize the merging of complex dependency structures, a spatial structure prediction method based on Hidden Markov contour tree model is proposed. The common hidden Markov model was extended from totally ordered sequence to partially ordered multivariate sequence. By capturing the complex contour structure on the surface, the flow direction between all positions on the three-dimensional surface was reflected. A node folding learning algorithm based on contour tree was proposed. An extension of the model was proposed, which was extended from generative type to discriminant type, so that the model could be used as post processor. Experiments were carried out on the real flood map data set. The results show the superiority of the proposed method.