Research And Realization Of Automatic Acquisition Algorithm For Deriving L-system's Rules Based On Genetic Algorithm

Virtual plants has always been an important issue in the field of computer graphics.It not only plays an important role in modern agriculture,forestry ecology and landscape planting,but also has significant value in the field of education,entertainment,and commercial.In recent years,with the increasing government' sinvestment in digital agriculture,L-system wich is one of the main method of the virtual plant,has gain more and more attention of scholars.But most of the L-system rules are derived by manual amendmenting and experience.There are many defects,such as inefficiency and can't see the graphics.With the continuous development of the virtual plant in recent years,this kind of method has turned to unable to meet the needs of applications.Therefore,the study of automatic acquisition algorithm of L-system,especially for the specified three-dimensional plants,has become a scientific problem urgently to be solved.Based on the National Natural Science Fund,the automatic acquisition algorithm of L-system production rules,wichi based on genetic algorithm,are discussed deeply.The main research works are as follows:1.Considering the characteristics of the three-dimensional tree structure,this paper presents a similarity calculation method based on the particle size of three-dimensional tree structure.The tree structure is firstly divided into thick,medium and thin three particle size by the algorithm:similar of outline,topology and internal node.Surrounding the tree structure with the paraboloic and it will reflect the different of the tree's outlines.Then get the similarity of the tree's topology can be calculated by comparing the relationship of the ondes in the treess.After that,it will gain the similarity of the internal space can be calculated by calculating the minimum space distance based on the topology relation.Finally,the similarity between the two tree structures can be calculated by combinating the outline similarity,the topological similarity and the internal space similarity.2.A spacial encoding solution of the chromosome base on the structure of the character and bracket in the L-system rules was propose in the genetic algorithm The form of gemini body was used for the coding solution,and the length of the chromosome is not fixed.One of the chromosome is used to encode characters besides brackets and it's position,and the other one is used to encode the brackets.Symbolic coding was used in this paper and a map between alphabet of L-system and the integers was given.3.The method of base on baseline rules to generate population and the method of randomly generated population were combined in this paper.The method which is based on plant taxonomy,can not only get the high quality individuality but also can keep the variety of the population.4.Based on the gemini structure ecoding solution,the operations of crossover and mutation are proposed.The chromosome of characters and the chromosome of brackets have it's own crossover and mutain rates respectively,Additionally,the operates only can be applicated to the same type of chromosome.Therefore,this method not only realizes more delicate operation of the mutation process,but also maintain the diversity of the population.5.On the basis of the above researches,a L-system automatic acquisition and redraw system is realized.According to the data of real plants inputted by the user,the system can cquire the L-system production rules,then draw and render a virtual model.