Research On Semantic Feature Extraction Method Of Complex Space Surface

Geological surface reconstruction is an important part of 3D geological modeling.Many current methods are based on data-driven.However,due to the lack of key information in the original data and the uncertainty in the data,the reconstructed surface is caused The deformation and distortion of the main structural morphology in the middle is inconsistent with the understanding of actual geological surfaces by geologists.The reconstruction results obtained through pure data driving are unreasonable and fail to meet expectations.Based on this actual situation,this thesis first extracts structural semantic features on sparse geological point cloud data.Structural semantic features are important morphological manifestations in the data.Structural semantic features are the result of human understanding of geological data and are geologically related.The embodiment of experience can improve the effect of surface reconstruction by converting people's understanding of structural semantic features into effective control information,and adding it to surface reconstruction.In addition,this thesis evaluates the effect of surface reconstruction from the perspective of constructing semantic features.The research of this thesis focuses on the extraction method of semantic features.The main research work and innovations are as follows:1.A method of extracting semantic features from sparse geological point cloud data is proposed.In view of the characteristics of sparse geological point cloud data,the direction of survey lines is dense and the lines between survey lines are sparse.After extracting feature points for classification,through human intervention,select feature points that can reflect the hidden semantic features in the data for fitting,and get important Then,the structural features are converted into effective control information and added to the surface reconstruction process.To a certain extent,it can solve the problem of the shape deformation of the reconstructed surface caused by the lack of key information and errors in the original data,so as to improve the quality of reconstruction.2.A method of extracting semantic features from rectangular grid data is proposed.The method in this paper is based on the contour data structure.On the basis of extracting local features,through the editing form of human-computer interaction,the feature is judged and the most important part is selected.Features,regression to get the final surface structure features,and the method is compared with the traditional flow simulation method by simulation experiment,which proves the effectiveness of the method proposed in this paper.Based on the semantic features extracted from the reconstructed surface,a method for evaluating the effect of surface reconstruction based on semantic features is proposed.The semantic features on the reconstructed surface are studied and compared with the potential structural features before reconstruction.It is a way to evaluate whether the surface reconstruction effect is as expected.If the structural features before and after reconstruction are consistent,it means that the reconstruction effect has reached the expectation.If they are inconsistent,there will be characteristic errors,which can be further reconstructed based on the characteristic errors.