Parameter extraction method of virtual plant growth model based on Improved Particle Swarm Optimization

•A common problem in plant modeling is to create varying shapes similar to an initial plant model.•A parameter extraction method for a virtual plant growth model is proposed based on an improved particle swarm (PSO) algorithm.•The experiments show that suitable parameters for simulating a specific tree can be obtained through multiple iterations.•The proposed method provides a unified description of plant geometrical structure and a new idea for generating different shapes of a specific tree. A common problem in plant modeling is to create varying shapes similar to an initial plant model. Although significant advancement of 3D plant modeling has been achieved, there are still some problems such as vague descriptions and execution speed. In view of those problems, a parameter extraction method for a virtual plant growth model is proposed based on an improved particle swarm optimization (PSO) algorithm. This method first defines a description framework to set the parameters of a virtual plant growth model, and then the optimized parameters are obtained by using the improved PSO algorithm. This algorithm helps particles leave the local optima by adding a stochastic perturbation in solution space. The experiments show that suitable parameters for simulating a plant abstraction can be obtained through multiple iterations. Meanwhile, the fitness value and standard deviation are better than from other methods under four test functions and the convergence rate is also improved. The proposed method provides a unified description of plant geometrical structure, which is relatively intuitive, and provides a new idea for generating different shapes of a plant geometrical structures.