Hypertension is a major risk factor for coronary artery disease, stroke, and kidney disease. Many studies suggest that elevated intramural stresses caused by hypertension may stimulate inflammatory changes, but little has been done to ascertain whether inflammation and stress are spatially correlated. Such correlations are a first step in identifying the mechanisms that may relate intramural stress to disease so that more effective clinical treatments may be developed. Arterial branches exhibit local stress peaks and are focal points for the onset of disease. They are thus a logical place to examine whether high stresses spatially correlate with increased inflammation. This research seeks to 1) develop a histology-based method to reconstruct small arterial branches; 2) use finite element analysis to evaluate intramural stresses where experimental testing is of limited use; 3) quantify biological measures of inflammation; and 4) visually and statistically compare the distribution of stress with the distribution of inflammation. Hypertension was induced in Sprague-Dawley rats by implanting Angiotensin II pumps for 7 days or 21 days. Normotensive rats were used as controls. To preserve morphology the mesentery was pressure-fixed in situ, harvested, processed, and embedded in glycol methacrylate resin. Branch geometry was reconstituted from serial sections. This involved: correcting deformations caused by sectioning; aligning sections into an image stack; identifying vessel boundaries; creating a surface suitable for finite element analysis; reducing the branch geometry to a midplane surface; and using Ansys (Ansys, Inc.) to mesh the midplane surface with a variable-thickness shell element. The pattern of inflammation was characterized by measuring the local density of monocytes and macrophages. Cell density was expressed as a distribution on the branch surface, which simplified visualization and facilitated statistical comparisons of inflammation with stress. Both intramural stresses and inflammation were greater near branches during hypertension. In most cases, however, high stresses and high cell density were not spatially collocated. The onset of an adaptive response may reduce the strength of this correlation. Maximal wall tension, defined as the maximal midplane stress multiplied by the wall thickness, was elevated near branches and strongly correlated with cell density.

In the process of continuous scaling improvement under a single system-on-chip which contains millions of logic gates, testability of the design becomes more and more important and thus multiple scan tree test architecture can effectively reduce test time and test data simultaneously. In the current two-dimensional structure of the system-level chip, the interconnect has become one of the main factors in delay and power consumption, and thus optimizing interconnect becomes a very important topic. Especially, three-dimensional ICs, stacked multiple chips vertically by through-silicon-via technique, can be effective in reducing the length of the interconnects, power consumption and offering features of heterogeneous IC integration. In this research study, we consider three-dimensional chips in both respects of wire length and the scan output limits, and propose the test synthesis algorithm of multiple scan trees to reduce test cost for three dimensional integrated circuits.

This thesis presents a scripting language interface for modeling manifold meshes represented by a Doubly Linked Face List (DLFL).With a scripting language users can create procedurally generated meshes that would otherwise be tedious or impractical to create with a graphical user interface. I have implemented a scripting language interface for the user to create stand-alone scripts as well as script interactively within a graphical environment.

Digital Earth is a kind of GIS application which use lots of computer techniques, including ¡§Computer Graphic¡¨, ¡§High quality Satellite Image¡¨, ¡§Multi-resolution¡¨, ¡¨ Metadata¡¨, ¡¨Level of detail¡¨, etc. In the past, geographers always tend to use triangle or grid structure to build 3D GIS. Now 3D GIS tend to use VRML to display virtual terrain data. Nevertheless, this display method is short of three-dimension analysis. It just uses 2.5 dimension to simulate 3 dimension space structure. This thesis is base on the theory of GeoVRML Quad-tree structure and use JAVA to implement ROAM algorithm. And this method can improve Quad-tree structure morphing error and raise virtual terrain display speed. This thesis use ETOPO5 5-minute DTM data to build Earth model (provide by NGDC) and Taiwan 40M DTM data, 200M, 100M, 50M, 25M satellite image to build Taiwan 3D GIS model. Meanwhile, this thesis addresses an unique three level structure to display 3D GIS model. One is Satellite Image layer, one is virtual terrain layer, and last one is 3D feature object.

In this paper, we propose an area-time efficient design for redundant CORDIC-based SIN/COS evaluation by predict on the polarity of micro-rotations using a novel technique called ¡§Base Transfer angle Decomposition Algorithm¡¨(BTDA). The proposed design benefits from a constant scaling and requires no correcting iterations. By predicting the polarity of the signed bit of the micro-rotation, the critical paths of the unfolded and the pipelined designs involve only the X and Y recurrences. The implementations of BTDA architectures for 24-bit wide CORDIC-Base SIN/COS generator were synthesized using FPGA tools (XILINX Foundation Series version 2.1i), and the area-time complexities are presented for unfolded as well as pipelined designs. The proposed design results save more than 25% hardware area with speed-up of more than 30% compared with the exiting methods. Keywords: CORDIC, BTDA, Redundant, SIN/COS, FPGA

How to efficiently and relistically generate 3D human face models is a very interesting and difficult problem in computer graphics. animated face models are essential to computer games, films making, online chat, virtual presence, video conferencing, etc. As the progress of computer technology, people request for more and more multimedia effects. Therefore, construct 3D human face models and facial animation are enthusiastically investigated in recent years. There are many kinds of method that used to construct 3D human face models. Such as laser scanners and computer graphics. So far, the most popular commercially available tools have utilized laser scanners. But it is not able to trace moving object. We bring up a technique that construct 3D human face model based on real image sequence. The full procedure can be divided into 4 parts. In the first step we use two cameras take picture con human face simultaneously. By the distance within two cameras we can calculate the depth of human face and build up a 3D face model. The second step is aimed at one image sequence which is taken by the same camera. By comparing the feature poins on previous image afterward image we can get the motion vector of human face. Now we can construct a template of animated 3D face model. After that we can map any kind of 2D new character image into the template, then build new character's animation. The full procedure is automatic. We can construct exquisite human facial animation easily.

This thesis presents an efficient and biologically informed 3D human body morphing technique through data-driven alteration of standardized 3D models. The anthropometric data is derived from a large empirical database and processed using principal component analysis (PCA). Although techniques using PCA are relatively commonplace in computer graphics, they are mainly used for scientific visualizations and animation. Here we focus on uncovering the underlying mathematical structure of anthropometric data and using it to build an intuitive interface that allows the interactive manipulation of body shape within the normal range of human variation. We achieve weight/gender based body morphing by using PCA. First we calculate the principal vector space of the original data. The data then are transformed into a new orthogonal multidimensional space. Next, we reduce the dimension of the data by only keeping the components of the most significant principal vectors. We then fit a curve through the original data points and are able to generate a new human body shape by inversely transforming the data from principal vector space back to the original measuring data space. Finally, we sort the original data by the body weight, calculating males and females separately. This enables us to use weight and gender as two intuitive controls for body morphing. The Deformer program is implemented using the programming language C++ with OPENGL and FLTK API. 3D and human body models are created using Alias MayaTm.

High-temperature gas-cooled reactors (HTGR) are passively safe, efficient, and economical solutions to the world’s energy crisis. HTGRs are capable of generating high temperatures during normal operation, introducing design challenges related to material selection and reactor safety. Understanding heat transfer and fluid flow phenomena during normal and transient operation of HTGRs is essential to ensure the adequacy of safety features, such as the reactor cavity cooling system (RCCS). Modeling abilities of system analysis codes, used to develop an understanding of light water reactor phenomenology, need to be proven for HTGRs. RELAP5-3D v2.3.6 is used to generate two reactor plant models for a code-to-code and a code-to-experiment benchmark problem. The code-to-code benchmark problem models the Russian VGM reactor for pressurized and depressurized pressure vessel conditions. Temperature profiles corresponding to each condition are assigned to the pressure vessel heat structure. Experiment objectives are to calculate total thermal energy transferred to the RCCS for both cases. Qualitatively, RELAP5-3D’s predictions agree closely with those of other system codes such as MORECA and Thermix. RELAP5-3D predicts that 80% of thermal energy transferred to the RCCS is radiant. Quantitatively, RELAP5-3D computes slightly higher radiant and convective heat transfer rates than other system analysis codes. Differences in convective heat transfer rate arise from the type and usage of convection models. Differences in radiant heat transfer stem from the calculation of radiation shape factors, also known as view or configuration factors. A MATLAB script employs a set of radiation shape factor correlations and applies them to the RELAP5-3D model. This same script is used to generate radiation shape factors for the code-toexperiment benchmark problem, which uses the Japanese HTTR reactor to determine temperature along the outside of the pressure vessel. Despite lacking information on material properties, emissivities, and initial conditions, RELAP5-3D temperature trend predictions closely match those of other system codes. Compared to experimental measurements, however, RELAP5-3D cannot capture fluid behavior above the pressure vessel. While qualitatively agreeing over the pressure vessel body, RELAP5-3D predictions diverge from experimental measurements elsewhere. This difference reflects the limitations of using a system analysis code where computational fluid dynamics codes are better suited.

This thesis presents a scripting language interface for modeling manifold meshes represented by a Doubly Linked Face List (DLFL).With a scripting language users can create procedurally generated meshes that would otherwise be tedious or impractical to create with a graphical user interface. I have implemented a scripting language interface for the user to create stand-alone scripts as well as script interactively within a graphical environment.

Hardware implementation is one of common solutions for accelerating 3D Graphics Pipelining Application. In this thesis, our purpose is to probe into the effect of 3D graphics system performance, according to the memory allocation of 3D graphics data and bus architecture for 3D graphics system-on-chip. And we also improve performance of whole application system efficiently by existent hardware resource. For getting the purpose, we use system level of simulation to observe and analyze the access of hardware accelerator in system and find out the key for improving performance. In this paper, we use ESL design to aid us for system simulation. Besides simulation time is much faster than RTL, abstract description is easy to implement and analyze. In memory organization, we must understand the relation of access data of 3D hardware with SDRAM, and reallocation memory. So, we divide each data and put them in different banks of SDRAM, scratch memory of system and built-in memory of hardware. Besides we increase the bandwidth of system bus by using multilayer architecture in system bus, we modify software to up the access times for improving performance. The experiment results point out that we speed up performance for 1.62 times.