Research On The Model With Growth Features For Virtual Plants

The individuals or communities are the studying object for virtual plants, in which its goal is the morphogenesis of plants and simulating the process of plant growth and development. In management science and engineering discipline, the modeling and simulation of virtual plants is an important research direction.In recent years, the research of virtual plants has gradually become a hot interdisciplinary research field including botany, agriculture, computer graphics, information science, applied mathematics and visualizing technology and so on. The model played a significant role in research and application of virtual plants. On the one hand, it is the abstract description of plant form and structure, physiological and ecological processes empirically. On the other hand, it applies qualitative or quantitative method to analyze the growth and development process of plants based on mathematical theory. This process refers to the produce and development of topology and geometrical form, the mutual feedback mechanics between their structure and function, the regulation of genotype by plant genes, and eventually visualizing in computer. We carried out qualitative or quantitative analysis for plant growth and development process and established the corresponding mathematical model by utilizing advanced computational theory and technology under, in which it can provide support for the computational model and visualization of system biology and help to predict and guide the production of agriculture and forestry in a way.Our study comes from the innovative special projects in Shandong Province(No: 2006ZZ10), which its presider was my tutor and initially simulated to visualize of fractal. Later, affected by P. Prusinkiewicz’ research team in Calgary University, Canada, we began systematically to study virtual plant model based on L-system. Meanwhile, in order to develop and utilize biological data effectively, we extracted the features of plant leaves image and analyzed them so that 3D geometric model of the leaves had been constructed and visualized by computer. This work benefited from the system biology key laboratory of Shandong province which was led by Professor Xianzhong Feng. After long times, by reading a lot of literature, we systematically studied fractal model, organ model and integral plant form, and eventually an adaptive intelligent evolution model of plant morphology was proposed on the basis of generalized L-system, in which it combined with the concepts of gene expression programming. Therefore, our final destination is to reconstruct systematic model for virtual plants by the study of different growth mechanism of plant phenotype, it strived to build a virtual plant model from a view of systemic point. It finally built up a computational model of virtual plant with clear physical significance, simple structure coherent and easy to understand and programming while meeting the design requirements of the plant optimum plant type. It was able to perform computer visualization so that it can be able to predict and guide the production of crops and support the research and application of precision agriculture in some degree.It has achieved the desired goal and satisfactory results after long-term study. This thesis mainly focues on the five aspects to study, as follows:1. Followed the self-similarity of plant growth, a quantitative fractal growth model of plant phenotype was proposed based on fractal geometry theory, which is an important way for the rapid construction of plant morphology. In nature, plants are showed as a multi-level self-similarity system at a certain scale. It is very difficult that we recognize and study the fractal characteristics in morphological structure, the distribution of individuals and growth process of plants from linear, smooth view. Then, these fractal features are complicated and nonlinear. But, the processes of modeling and simulation for fractal plants mainly take the recursive or iterative algorithms, so the plants with fractal characteristics are simple if we analyze the algorithmic complexity of plant morphogenesis. Therefore, followed the original principle simplifying complex issue, a quantitative model of fractal plants easily implementing in computer was proposed based on fractal theory after deep study on the fractal features of plants and the method of fractal plant reconstruction. Then, it can provide a new means and perspective for solving the problem of plant classification and identification, the impact on the plant by environment and others.2. Focused on the application, this paper put forward an improved Snake model to extract and calculate the leaf features of plants based on image feature analysis and representation, and as a constraint, conducted the 3D reconstruction of soybean leaves. On condition that pre-processing of picture file and analytical representation of inamge features, the work mainly includes two aspects:(1) the processing and analysis of characteristics of plant leaves: Firstly, after the gray and binary processing of vane image, it extracted and calculated the image features of plant leaves using the classical image processing algorithm in RGB color space; Secondly, it extracted the corner features of the color image of plant leaves with complex background by the Harris operator based on the HSI color model. On the basis, it used the improved Snake model algorithm to carry on the contour fitting of the plant leaf, and adopted the chain code method to solve the corresponding biomass, such as leaf area, perimeter, length width ratio, etc.(2) The 3D reconstruction and transformation of the plant leaves: According to the calculated results of the biomass, vane contour has been interpolated and fitted based on the Bspline curve and surface and the fitted error was evaluated by the Hausdorff distance, then we adjusted the vane contour by node interpolation algorithm to match the feature points, in which made the fitting and interpolation of vane contour to achieve the best optimization. In the end, the 3D visualization model of the plant leaf is constructed and it achieved free transform in 3D space. The simulation results also show that the computational analysis and 3D reconstruction based on image features is a very effective method for the modeling of biological morphology, and it is an important part of the plant dynamic model system.3. In order to develop the simulation and calculation function of L-system, a generalized L-system model is proposed by means of combining structure-function model with it, and it analyzed and discussed the modeling ability of the L-system. The virtual plant modeling method based on L-system, which was originally a descriptive empirical model of fractal plant. At present, it has been transformed into a dynamic interpretation model for the integration of plant growth mechanism. As a result, based on the definition and composition of L-system, this paper made a thorough study of the plant modeling and simulation of L-system in branching pattern, geometric form, growth mechanism and the interaction between surroundings and plants, even then proposed a generalized L-system model which abstracted the plant growth and development into a multi-level, structural dynamic functional transformation system. First, The model of plant branching is constructed by using the location information of the branching nodes based on the architectural model, and a descriptive empirical model of plant branching structure is established. The geometrical morphology of branch based on bicubic B-spline subdivision of multi-resolution curve method. Secondly, the biomechanical mechanism of the branching axis is introduced, and it structured the functional branching pattern of plants based on the reaction-diffusion-convection equation. Ultimately, followed the state transfer strategy of Markov process, this paper constructed a dynamic visual physical model of virtual plant in order to meet the requirements of the general plant considering the interaction between plant and environment, the structure-function of plant growth model, the modeling method of a plant organ based on differential turtle geometric interpretation. The proposed model improved the function of L-system for modeling and simulation of virtual plants, and extended the application range of L-system, then laid the foundation of L-system united with the other model.4. For the construction of effective mechanism to communicate biology with modeling form of plants, we put forward a soybean branch morphological quantitative model based on L-system using the Numerical Nomenclature Coding System. Proceeded from experimental data of biology, on the basis of structure-function principle, followed the pipe model theory, soybeans as model plant, after the acquirement of an empirical model and distribution interval for soybeans height by nonlinear regression analysis, we quantified the main stem, branch and internode of soybeans using the taper stem equation and S-curve equation. A novel L-system was finally proposed to construct the branching form by means of the numerical nomenclature coding system and the visual simulation was performed. This part is the basis of constructing intelligent evolutionary model of plant morphogenesis.5. Based on gene expression programming and L-system, an adaptive intelligent evolutionary model is proposed for virtual plant morphogenesis so that it fit to diversity and plasticity of plant phenotype. On the basis of morphological quantitative computation, this part put forward an adaptive intelligent evolutionary model of plant genotype based on the ideas of gene expression programming and the generalized L-system. The proposed model can minimize manual intervention in the modeling process, get the modeling parameters automatically and built the initial population of plant morphological evolutionary process rapidly. It also could improve search efficiency of the algorithm. At the same time, combined with the cluster analysis principle to compute the individual fitness, it can automatically filter out fine individual of per generation. The model is not only to meet the diversity and plasticity of plant genotype, but also to inject vitality for optimal plant type automatic design. Meanwhile, it can provide new computing ideas and patterns for assistant decision in modeling, analysis and computation of system biology.This research is the summary of long-term work. In order to systematically and comprehensively build complete plant phenotype, we set the fractal phenomenon of natural plants as start in this paper. According to the biological modelling method, faithful to the actual plant growth mechanism, we carried out research for these models of virtual plants, including model of plant organs, the quantitative model of the experimental data, computer visualization model. Furthermore, we discussed the intelligent evolution model of plant phenotype. These research results will also provide strong support for the study of the plant phenotype regulated by the genes in the follow-up work.