This thesis addresses the problem of producing three-dimensional constrained Delaunay triangulated meshes from the sequential two dimensional MRI medical image slices. The approach is to generate the volumetric meshes of the scanned organs as a result of a several low-level tasks: image segmentation, connected component extraction, isosurfacing, image smoothing, mesh decimation and constrained Delaunay tetrahedralization. The proposed methodology produces a portable application that can be easily adapted and extended by researchers to tackle this problem. The application requires very minimal user intervention and can be used either independently or as a pre-processor to an adaptive mesh refinement system. / Finite element analysis of the MRI medical data depends heavily on the quality of the mesh representation of the scanned organs. This thesis presents experimental test results that illustrate how the different operations done during the process can affect the quality of the final mesh.

No description available.

隨著互聯網上3D模型的大量增加，產生了開發3D模型搜索引擎的需求。本論文提出了一個基於草圖和3D模型的3D模型搜索引擎。 / 對於使用3D模型作檢索條件的搜索系統，我們提出了兩種新的3D模型描述子，分別叫做Sphere Image 描述子和Bag-of-View-Words (BoVW)描述子。Sphere Image描述子是由一系列投影圖的特徵組成。我們將每一個視角看到的圖形都當作是一個"像素"，把視角的位置看作像素點的位置，把所看到的圖形的特徵值看作是像素值。我們同時也提出了一種基於概率圖的3D模型匹配算法，並開發了一個3D模型檢索系統來檢測我們的算法。BoVW描述子通過3D模型投影圖出現的次數來描述3D模型。我們用一種自適應的聚類算法，對3D模型的所有投影圖進行分類，然後用一個多層次的柱狀圖來描述一個3D模型。我們同時提出一種新的金字塔匹配算法來比較3D模型。我們使用SHREC和普林斯頓的3D模型庫來檢驗我們的系統，實驗結果證明我們的系統在檢索效率和精度上都優與現今的3D模型檢索系統。 / 對於使用草圖作檢索條件的3D模型搜索系統，我們提出Bigger ExposureOpportunity Views (BEOV) 描述子來表示3D模型，同時提出Shape-Ring描述子來表示草圖。BEOV描述子是由一些特徵圖組成，這些圖的特點是更容易被人們看到。Shape-Ring描述子保留了圖形的輪廓和內部特徵。我們使用SHREC2012草圖數據庫來檢驗我們的系統，實驗結果證明我們的系統在精度和計算複雜度上都優與現今的3D模型檢索系統。 / The large number of 3D models on the Internet encourages us to develop 3D model search engines. In this dissertation, we present a 3D model retrieval system using both the 3D model query and the sketch query. / For 3D model query based retrieval system, we propose two new 3D model descriptors, named the Sphere Image and the Bag-of-View-Words (BoVW) descriptor. The Sphere Image is defined as a collection of view features. A viewpoint of a 3D model is regarded as a "pixel": (1) The position of the viewpoint is denoted as the coordinate of the "pixel". (2) The feature descriptor of the projected view is denoted as the value of the "pixel". We also propose a probabilistic graphical model for 3D model matching, and develop a 3D model retrieval system to test our approach. The BoVW descriptor describes a 3D model by measuring the occurrences of its projected views. An adaptive clustering method is applied to reduce the redundancy of the projected views of each 3D model. A 3D model is represented by a multi-resolution histogram, which is combined by several BoVW descriptors at different levels. The codebook is obtained by unsupervised learning. We also propose a new pyramid matching method for 3D model comparison. We have conducted experiments based on the SHape REtrieval Contest (SHREC) 2012 Generic 3D model benchmark and the Princeton Shape Benchmark (PSB). Experimental results indicate that our system outperforms some state-of-the-art 3D model retrieval systems with respect to the retrieval precision and the computational cost. / For sketch query based retrieval system, we propose a Bigger Exposure Opportunity Views (BEOV) descriptor and a Shape-Ring descriptor, for representing the 3D model candidates and the sketch query, respectively. The BEOV descriptor represents a 3D model by several characteristic views, which have more chances to be exposed to people. The Shape-Ring descriptor preserves the features of the contour and the inside detail of the sketch query and the BEOV. Experiments have been conducted based on the SHape REtrieval Contest (SHREC) 2012 and SHREC 2013 sketch track data sets. Our approach outperforms the existing 3D model retrieval methods in terms of the retrieval precision and the computational cost. / Detailed summary in vernacular field only. / Detailed summary in vernacular field only. / Detailed summary in vernacular field only. / Ding, Ke. / Thesis (Ph.D.) Chinese University of Hong Kong, 2014. / Includes bibliographical references (leaves 107-120). / Abstracts also in Chinese.

Image classification is a fundamental information extraction procedure in remote sensing that is used in land-cover and land-use mapping. Despite being considered as a replacement for manual mapping, it still requires some degree of analyst intervention. This makes the process of image classification time consuming, subjective, and error prone. For example, in unsupervised classification, pixels are automatically grouped into classes, but the user has to manually label the classes as one land-cover type or another. As a general rule, the larger the number of classes, the more difficult it is to assign meaningful class labels. A fully automated post-classification procedure for class labeling was developed in an attempt to alleviate this problem. It labels spectral classes by matching their spectral characteristics with reference spectra. A Landsat TM image of an agricultural area was used for performance assessment. The algorithm was used to label a 20- and 100-class image generated by the ISODATA classifier. The 20-class image was used to compare the technique with the traditional manual labeling of classes, and the 100-class image was used to compare it with the Spectral Angle Mapper and Maximum Likelihood classifiers. The proposed technique produced a map that had an overall accuracy of 51%, outperforming the manual labeling (40% to 45% accuracy, depending on the analyst performing the labeling) and the Spectral Angle Mapper classifier (39%), but underperformed compared to the Maximum Likelihood technique (53% to 63%). The newly developed class-labeling algorithm provided better results for alfalfa, beans, corn, grass and sugar beet, whereas canola, corn, fallow, flax, potato, and wheat were identified with similar or lower accuracy, depending on the classifier it was compared with. / vii, 93 leaves : ill., maps (some col.) ; 29 cm