【摘要】：從三維點云數(shù)據(jù)中重建出高精度的植物葉片模型并進行葉片形變模擬一直是林學、計算機圖形學、植物生理生態(tài)學、景觀設計等眾多學科研究的熱點問題之一。植物葉片不僅是植物最重要的器官之一,同時也是植物葉片生理功能、形態(tài)特征分析、冠層光分布計算等研究的基礎。但由于室外植物葉片的點云數(shù)據(jù)存在噪聲、葉片邊緣不光順、葉片表面存在孔洞等問題,所以不易重建出高精度的植物葉片模型。針對存在的問題,本文首先對室外植物葉片點云數(shù)據(jù)進行去噪處理,然后根據(jù)植物葉片點云數(shù)據(jù)的邊緣點云數(shù)據(jù)擬合出新的葉片邊緣點云數(shù)據(jù),其次對葉片表面進行擬合并采用Delaunay三角剖分算法生成植物葉片的網(wǎng)格模型,最后對存在孔洞的植物葉片網(wǎng)格模型進行修補。另一方面圍繞植物葉片的三維形變模擬研究已經(jīng)有了幾十年的歷史,經(jīng)過幾十年的研究努力,目前已經(jīng)取得了顯著的成果。但在這些研究中,大部分植物葉片模型都是通過虛擬現(xiàn)實技術生成的,植物葉片模型也就不具有真實性。且植物葉片模型形變往往只是采用簡單的彈簧-質點模型實現(xiàn)葉片形變,方法過于簡單、單調。針對以上問題,本文首先將植物葉片網(wǎng)格模型實體化并添加主脈,然后對葉肉部分和葉脈部分賦予不同的材料屬性,最后采用目前比較流行的幾種形變方法,實現(xiàn)植物葉片模型形變。本文主要研究內容如下:(1)點云數(shù)據(jù)去噪方法研究。提出了一種適合于室外植物葉片點云數(shù)據(jù)的去噪算法,該算法將噪聲點分為三類。第一類是距離大片中心點云較遠、小而密集的點云。第二類是偏離點云較遠,懸浮在點云上方的離散、稀疏點。第三類是噪聲點和真實點混合在一起的點,該類點云的形態(tài)呈分層狀。并分別對這三類噪聲點進行去噪。(2)植物葉片邊緣擬合算法研究。提出了一種適合于室外植物葉片點云數(shù)據(jù)的邊緣擬合算法。該算法首先提取出植物葉片的邊緣點云數(shù)據(jù),然后根據(jù)空間曲線的定義擬合出光順的植物葉片邊緣。(3)植物葉片表面擬合與重建方法研究。提出了一種適合于室外植物葉片點云數(shù)據(jù)的表面擬合與重建方法。該方法首先采用雙三次Bezier曲面擬合葉面,使葉片表面更加光順,然后通過二維Delaunay算法實現(xiàn)植物葉片的三維重建。(4)植物葉片網(wǎng)格模型表面孔洞修補方法研究。提出了一種適合于室外植物葉片點云數(shù)據(jù)的表面孔洞修補方法。該方法首先識別出植物葉片網(wǎng)格模型中的孔洞邊緣,然后通過波前法實現(xiàn)孔洞的修補。(5)植物葉片形變方法研究。提出了一種適合于室外植物葉片點云數(shù)據(jù)的葉片實體化方法,并將目前流行的幾種形變方法運用到植物葉片模型中,以實現(xiàn)室外植物葉片模型形變方法的多樣性。
[Abstract]:It has been one of the hot issues in forestry, computer graphics, plant physiology and ecology, landscape design and so on to reconstruct high precision plant leaf model from 3D point cloud data and simulate leaf deformation. Plant leaves are not only one of the most important organs of plants, but also the basis of physiological function, morphological characteristics analysis and calculation of canopy light distribution. However, due to the noise in the point cloud data of outdoor plant leaves, the edge of the leaves is not only smooth, there are holes in the surface of the leaves, so it is difficult to reconstruct the high precision plant leaf model. In order to solve the existing problems, the data of outdoor plant leaf point cloud are de-noised firstly, and then the new data of leaf edge point cloud are fitted according to the edge point cloud data of plant leaf point cloud data. Secondly, the meshes of plant leaves were generated by Delaunay triangulation algorithm, and the mesh models with holes were repaired. On the other hand, the three-dimensional deformation simulation research around plant leaves has been decades of history, after decades of research efforts, has achieved remarkable results. However, in these studies, most plant leaf models are generated by virtual reality technology, and the plant leaf models do not have authenticity. And the deformation of plant leaf model is usually realized by simple spring-mass model. The method is too simple and monotonous. In order to solve the above problems, the model of plant leaf grid is first materialized and the main vein is added, then the mesophyll part and the vein part are given different material properties. Finally, several deformation methods which are popular at present are adopted. The deformation of plant leaf model was realized. The main contents of this paper are as follows: (1) the method of point cloud data denoising. A denoising algorithm for outdoor plant leaf point cloud data is proposed, which divides the noise points into three categories. The first is a small, dense point cloud that is far from the center of a large area. The second is the discrete and sparse point suspended above the point cloud, which is far away from the point cloud. The third is the point where the noise point and the real point are mixed together, the shape of this kind of point cloud is stratified. The three kinds of noise points were de-noised respectively. (2) the edge fitting algorithm of plant leaves was studied. An edge fitting algorithm for outdoor plant leaf point cloud data is proposed. Firstly, the edge cloud data of plant leaves are extracted, and then the smooth leaf edges are fitted according to the definition of spatial curve. (3) the method of surface fitting and reconstruction of plant leaves is studied. A surface fitting and reconstruction method for outdoor plant leaf point cloud data is proposed. In this method, the surface of the leaf is fitted with a bicubic Bezier surface to make the leaf surface more smooth, and then the three-dimensional reconstruction of the plant leaves is realized by using the two-dimensional Delaunay algorithm. (4) the method of repairing the holes in the surface of the plant leaf mesh model is studied. A surface hole repair method for outdoor plant leaf point cloud data is proposed. In this method, the hole edges in the plant leaf mesh model are first identified, and then the holes are repaired by the wavefront method. (5) the deformation method of plant leaves is studied. A method of leaf materialization suitable for outdoor plant leaf point cloud data is proposed, and several deformation methods are applied to plant leaf model in order to realize the diversity of outdoor plant leaf model deformation methods.
【學位授予單位】：南京林業(yè)大學
【學位級別】：碩士
【學位授予年份】：2015
【分類號】：TP391.41

【參考文獻】

相關期刊論文 前4條

1 張超,王雪峰,李�？�;立體視覺技術在森林資源調查中的應用[J];北京林業(yè)大學學報;2004年03期

2 李江雄;反求工程中復雜曲面邊界線的自動提取技術[J];機械設計與制造工程;2000年02期

3 岳杰;陸聲鏈;孫智慧;郭新宇;魏學禮;;三維點云孔洞修補算法及在植物形態(tài)重建中的應用[J];農機化研究;2013年05期

4 孫殿柱;朱昌志;李延瑞;;散亂點云邊界特征快速提取算法[J];山東大學學報(工學版);2009年01期

相關碩士學位論文 前1條

1 曹園園;植物葉片萎蔫變形實時仿真技術研究[D];燕山大學;2013年

，

本文編號：2410308