Research On Simplification Approach Of Linear Features Based On Domain Knowledge

Line simplification is an important operator of map generalization.In simplifying tasks,cartographers usually select the appropriate one from scattered and disordered line simplification algorithms based on subjective decisions while seldom paying attention to the orderly management and automatically choosing existing line simplification algorithms.However,the existing algorithms cannot obtain good results in all application scenarios.For example,the rapid development and extensive collection of high-precision maps require the research of road network simplification algorithms that can meet specific scenarios and the data's characteristics.Therefore,it is still necessary to continuously improve the existing and research new line simplification algorithms.Based on the abstraction and induction of the current line simplification algorithms' knowledge and experience,this paper studies and improves the existing algorithms for linear features to solve the problem of effective identification of curved structures,simplification of urban high-precision roads network,and maintenance of spatial directional relations.The main research results and innovations are as follows :1)The knowledge representation method for simplifying the line elements on the map is studied.Scientific induction and formal expression of linear simplification knowledge and experience,and encapsulation of knowledge in ontology,can automatically select the algorithm that meets the simplification requirements of a specific linear target based on the knowledge,which is beneficial to the management,maintenance,and update of the existing algorithms.Taking the simplification of coastline and road network as examples,a relevant knowledge graph was constructed.2)The new algorithm for bending recognition is proposed and applied to the progressive simplification of lines.We analyze in detail the two problems when dividing the bending directly based on the extreme points and study the specific solutions to form a basic bending unit that conforms to the law of spatial cognition.Then select and calculate the bending indicators according to the needs,and then the bending is simplified progressively.Taking the coastline as the simplified object,the results prove that the method we have presented maintains the integrity of the bending structure on both sides of the line,while the progressive simplification results maintain the topological relationship and the overall structure of the line target.3)Improved an Opheim algorithm that takes into account navigation constraints.The algorithm first analyzes the navigation road network's model expression,extracts the key points that meet the semantic constraints,and then simplifies it progressively according to the road section's length,direction,and topological characteristics.The experimental results can maintain the spatial relationship and shape characteristics of the road to the greatest extent and mark out the problems that cannot be dealt with eventually.4)The intelligent maintenance method of the spatial direction relationship in reducing the accuracy of the urban high-precision road network is proposed.In view of the road network structure's characteristics,this article defines multiple directly connected nodes as particular nodes and other nodes as common nodes.After analyzing the problems caused by the existing methods to maintain the spatial directional relationship of special nodes,the genetic algorithm with elite strategy is introduced to find the optimal solution.The results of the four cities show that the method is reasonable and effective.The road shape after maintenance is similar to the original road shape,and the topological relationship is correct.