基于生长模型的苗期大豆植株三维可视化研究

发布时间：2018-03-18 00:01

本文选题：Gabor滤波　切入点：大豆植株　出处：《西北农林科技大学》2017年硕士论文　论文类型：学位论文

【摘要】：利用生长模型参数实现农田大豆植株的三维可视化是虚拟植物建模研究的重要内容。为实现苗期阶段的农田大豆三维可视化模拟,本研究选取西北农林科技大学水保所试验田培养的中黄13号大豆植株为研究对象,通过对苗期大豆植株各个器官的三维重建研究,从而实现农田大豆单株及群体的三维可视化及农田环境漫游交互的模拟,本文的主要研究内容如下:(1)苗期大豆植株器官形态建模。利用索尼DSC-W800数码相机获取原始大豆叶片图像信息,采用Sobel算子实现图像的边缘检测,然后利用中值滤波去除噪声。结合三次参数样条曲线拟合叶片轮廓,并基于Delaunay三角剖分逐点插入实现多边形的网格化,运用改进的Gabor小波算法提取叶片纹理能量特征,采取SPM算法描述纹理渐变线索和场景深度的关系,选用柱体的拓扑结构及量化公式模拟大豆主茎,结合模型分级化原理重建分枝模型,最后通过中轴线结构及二次曲线方程重建侧枝形态。(2)基于参数化L系统和有限状态机模型实现单株大豆可视化。利用获取的苗期大豆植株各器官三维重建模型,结合大豆植株的总线型空间结构特点,基于参数化的L系统在器官尺度上模拟苗期大豆植株的拓扑结构,并实现了L系统的并行生长及绘制技术,结合大豆的生长特性,同时使用多层次自动机构建苗期大豆生长的有限状态机模型,最终实现苗期单株大豆的三维可视化。(3)群体大豆植株可视化。基于L系统和有限状态机实现的大豆个体可视化模型,结合获取的大豆植株节间距采样和几何形态数据,采用实例化渲染技术实现了群体大豆的可视化。为了优化群体植株数据的存储结构及真实感渲染,采用OpenGL显示列表封装苗期大豆植株的各个器官,结合农田生长环境加上光照、地形等因素的影响,同时采用天空盒及雾化效果渲染天空场景,提高系统可用性。(4)大豆植株漫游系统。首先将虚拟农田图像分成N等份,然后结合行间十字型路线进行漫游交互,通过设定好的漫游控制点和漫游路径,结合转向处的Hermite曲线对控制点进行拟合,随着路径漫游,为了节约虚拟漫游系统的资源,将曲线进行插值离散化,同时将“固化”后的数据保存到数组中,并在计算过程中引入转换系数μ使漫游转向更加光顺自然,最后结合漫游路径拼接技术实现大豆农田的漫游效果。
[Abstract]:Using growth model parameters to realize 3D visualization of farmland soybean plants is an important part of virtual plant modeling. In this study, the soybean plant Zhonghuang No. 13, cultivated in the experimental field of soil and Water Conservation Institute of Northwest University of Agriculture and Forestry Science and Technology, was selected as the research object, and the three dimensional reconstruction of each organ of soybean plant at seedling stage was studied. In order to realize the 3D visualization of individual plant and population of farmland soybean and the simulation of field environment roaming interaction, The main contents of this paper are as follows: 1) Modeling the organ morphology of soybean plant at seedling stage. Using Sony DSC-W800 digital camera to obtain the original soybean leaf image information, and using Sobel operator to realize edge detection of the image. Then the median filter is used to remove the noise, and the cubic parameter spline curve is used to fit the blade contour. Based on the Delaunay triangulation, the polygon is meshed, and the improved Gabor wavelet algorithm is used to extract the energy feature of the leaf texture. SPM algorithm is adopted to describe the relationship between texture gradation cues and scene depth. The topological structure and quantification formula of columns are used to simulate the main stem of soybean, and the branching model is reconstructed by combining the principle of model classification. Finally, based on parameterized L system and finite state machine model, the visualization of single soybean was realized through the reconstruction of lateral branch morphology by the central axis structure and conic equation. The 3D reconstruction model was used to reconstruct the organs of soybean plant at seedling stage. Based on the characteristics of bus-type spatial structure of soybean plant, the topological structure of soybean plant at seedling stage was simulated by parameterized L system on organ scale, and the parallel growth and drawing technology of L system was realized, and the growth characteristics of soybean were combined. At the same time, the finite state machine model of soybean growth in seedling stage was constructed by using multi-level automata. Finally, the visualization of soybean plant is realized by 3D visualization of single plant soybean in seedling stage. Based on the individual visualization model of soybean realized by L system and finite state machine, the sampling of node spacing and geometric shape data of soybean plant are obtained. In order to optimize the storage structure and realistic rendering of plant data, OpenGL display list was used to encapsulate all organs of soybean plant in seedling stage, combined with field growing environment and illumination, in order to optimize the storage structure and realistic rendering of population soybean, the instantiation rendering technique was used to realize the visualization of soybean. The terrain and other factors are used to render the sky scene with sky box and atomization effect to improve the availability of the system. 4) Soybean plant roaming system. Firstly, the virtual farmland image is divided into N equal parts. Then the roaming interaction between the interline cross-type routes is carried out. By setting the roaming control points and roaming paths and combining with the Hermite curve of the steering place, the control points are fitted. With the roaming of the path, in order to save the resources of the virtual roaming system, The curve is interpolated and discretized, and the "solidified" data is saved into the array, and the conversion coefficient 渭 is introduced in the calculation process to make the roaming steering more smooth and natural. Finally, the roaming effect of soybean farmland is realized by combining the technology of roaming path splicing.
【学位授予单位】：西北农林科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：S565.1;TP391.41

【参考文献】