請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/27475
標題: | 非完整三維多面體網格模型之回復 Recovering Incomplete 3D Polygon Mesh Model |
作者: | Chun-Yen Chen 陳俊諺 |
指導教授: | 傅楸善(Chiou-Shann Fuh) |
關鍵字: | 網格平滑化,濾波器,幾何特徵表現,網格修補,幾何外型修正, mesh smoothing,filtering,feature representation,mesh repairing,geometric correction, |
出版年 : | 2007 |
學位: | 博士 |
摘要: | 平滑化和修補是三維多面體網格模型處理中最基本的問題。在此論文中,提出一個以尖銳度相關之濾波器設計方法來回復非完整三維多面體網格模型。本論文所提出的濾波器設計是以平滑表面法向量為基礎,藉由濾波演算法自動的選擇一種濾波器。依據區域尖銳度和尖銳度相關權重函式的選擇,此濾波器可以是一個平均濾波器(mean-filter)、一個最小值濾波器(min-filter),或是一個介於兩者之間的濾波器。為了將一個具有雜訊的模型回復成原來的外形,濾波演算法在面對平滑區域時會選擇平均濾波器,在面對尖銳區域時則選擇最小值濾波器。演算法中,尖銳度相關權重函式必須符合包含雜訊之網格模型的尖銳度數值分佈,因此Gaussian、Laplacian,和El Fallah Ford三種型式的函式可以被使用做為尖銳度相關權重函式。此外,演算法在權重函式中採用一個尖銳度係數來控制特徵回復的程度。此尖銳度係數的適當數值可以透過貝氏分類法(Bayesian classification) 分析尖銳度數值分佈來求得。
為了解決包含大量雜訊資料的多面體網格模型平滑化問題,本論文提出另一個增進的濾波器設計:具有特徵方向定位特性的尖銳度相關濾波器。此濾波器由一個前處理步驟和一個濾波程序所組成。在前處理步驟中,由具有雜訊的輸入網格模型得到一個平滑化參考模型。此平滑化參考模型具有和原始沒有雜訊干擾之完美模型幾乎一樣的特徵方向流場(flow direction field)。接下來的濾波程序是以等向性平滑化和非等向性平滑化所組成的一個線性組合。此設計可以讓濾波器利用重建輸入模型的表面特徵方向使它們和平滑化參考模型一致,逐漸地回復模型中的細部結構並去除雜訊。由實驗結果顯示,本論文所提出的濾波演算法在處理多面體網格模型之平滑化與保留尖銳特徵的效能均優於其它的方法。 在修補非完整三維網格模型方面,本論文提出一個以尖銳度為基礎的破洞修補方法。此方法包含兩個程序:以插補為基礎(interpolation-based)的破洞修補程序產生一個初始修補模型;和一個調整初始修補模型外形的後處理程序,使修補後的模型能符合原始模型的外形。在以插補為基礎的破洞修補程序中,使用radial basis function為基礎的表面插補演算法產生一個平滑的隱函數表面(implicit surface)填補破洞。接著,使用regularized marching tetrahedral演算法將隱函數表面做三角網格化。最後再利用縫合和規律化(regulating)步驟將破洞補片和原始多面體網格模型的破洞邊緣接合產生一個適用於後處理程序使用的規律化初始修補模型。在後處理程序裡,尖銳度相關濾波器被作用在初始修補模型上。後處理是一個反覆執行的程序,透過每一個反覆執行的步驟調整每一個多面體表面的法向量來回復隱含在被修補模型中的尖銳特徵。實驗結果顯示此方法可以有效的修補非完整三維網格模型。 Smoothing and repairing are the most fundamental problems for 3D polygon mesh processing. In this dissertation, a sharpness dependent filter design is proposed to recover incomplete 3D polygon mesh model. The proposed filter design is based on the fairing of surface normal, whereby the filtering algorithm automatically selects a filter. This may be a mean-filter, a min-filter, or a filter ranked between these two, depending on the local sharpness value and the sharpness dependent weighting function selected. To recover the original shape of a noisy model, the algorithm selects a mean-filter for flat regions and a min-filter for distinguished sharp regions. The selected sharpness dependent weighting function has a Gaussian, Laplacian, or El Fallah Ford form that approximately fits the sharpness distribution found in all tested noisy models. A sharpness factor is used in the weighting function to control the degree of feature preserving. The appropriate sharpness factor can be obtained by sharpness analysis based on the Bayesian classification. In order to smooth a noisy polygon mesh model that contains large noise, an improved filter design – “direction-oriented sharpness dependent filter” is presented. This filter consists of a pre-processing step and a filtering process. In the pre-processing step, a smoothed reference model is derived from the input noisy mesh model such that the flow direction field of the smoothed reference model is almost identical to that of the original model. The subsequent filtering process is a linear combination of isotropic smoothing and anisotropic smoothing. This design allows the filter to gradually recover fine structures and remove noise by reconstructing the face direction of the input noisy mesh model so that it is the same as that of the smoothed reference model. Our experiment results demonstrate that the proposed filtering algorithm is superior to other approaches for smoothing a polygon mesh, as well as for preserving its sharp features. For repairing incomplete 3D polygon mesh model, a sharpness-based method for hole-filling is presented. The method involves two processes: interpolation-based hole-filling, which produces an initial repaired model; and post-processing, which adjusts the shape of the initial repaired model to conform to that of the original model. In the interpolation-based hole-filling process, a surface interpolation algorithm based on the radial basis function creates a smooth implicit surface that fills the hole. Then, a regularized marching tetrahedral algorithm is used to triangulate the implicit surface. Finally a stitching and regulating strategy is applied to the surface patch and its neighboring boundary polygon meshes to produce an initial repaired mesh model, which is a regular mesh model suitable for post-processing. During post-processing, a sharpness dependent filtering algorithm is applied to the initial repaired model. This is an iterative procedure whereby each iteration step adjusts the face normal associated with each meshed polygon to recover the sharp features hidden in the repaired model. The experiment results demonstrate that the method is effective in repairing incomplete 3D polygon mesh models. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/27475 |
全文授權: | 有償授權 |
顯示於系所單位: | 資訊工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-96-1.pdf 目前未授權公開取用 | 4.81 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。