Junhui Deng

Computerized 3D-reconstructions of the ligaments of the lateral aspect of ankle and subtalar joints*

Abstract3D-reconstruction images of the structures of lateral aspect of the ankle and subtalar joints were produced using plastination to make equidistant serial sections of 1.2 mm in thickness. A SGI workstation was employed to reconstruct the structures of the ligaments of the lateral aspect of ankle and subtalar joints in three dimensions. Reconstructed structures were displayed singly, in groups or as a whole, and these were rotated continuously at different velocities in 3D space. Different diameters and angles of the reconstructed structures could be measured easily. Improved results could be achieved with the use of a special sectional anatomical technique, i.e. contours + marching cubes algorithm.

The pre-styloid compartment of the parapharyngeal space: a three-dimensional digitized model based on the Chinese Visible Human

To build a digitized visible model of the parapharyngeal space of the Chinese Visible Human and to provide a sectional anatomic basis for radiological and clinical diagnosis of the parapharyngeal space, sectional anatomy data of the parapharyngeal space were selected from the Chinese Visible Human male and female to compare with MR imaging findings in the axial planes. From these data the parapharyngeal space and surrounding structures were segmented. They were then reconstructed in three dimensions on PC. In the axial planes of the sectional anatomy and MR imaging, the shape, content and relations of the parapharyngeal space were clearly displayed and the dominant plane for showing the parapharyngeal space was elicited. The three-dimensional reconstructed images displayed perfectly the anatomic relationships of the parapharyngeal space, parotid, muscles, mandible and vessels. All reconstructed structures can be displayed singly, in groups or as a whole; any diameter or angle of the reconstructed structures can be easily measured. The Chinese Visible Human male and female data set can provide complete and accurate data. The digitized model of the parapharyngeal space and its surroundings offers unique insights into the complex anatomy of the area, providing morphologic data for imaging diagnosis and surgery of the parapharyngeal space.

Parallel volume rendering on virtual machines

A volume rendering algorithm for parallel virtual machines is presented. To reduce the communication cost, and to guarantee the locality of subtasks, the volume data is divided and organized as a series of slices. By maintaining and employing a database of the performance index, the task subdivision algorithm produces an acceptable load balancing. A binary asynchronous method is introduced to merge all partial images in O(nlogn) time. Our algorithm was implemented, exploiting the classical client/server paradigm. Scalabilities of both the task size and the number of hosts are tested and experimental results are demonstrated and analyzed.

A B-Spline Surface Interpolation Technique for Reconstructing 3D Objects from Serial Arbitrary Shaped Planar Contours

A new method is presented to the B-spline surface presentation of an object defined by a set of parallel slices. For the application of B-spline inversion procedure, the methods about the data points generated are mainly introduced. The mesh of the data points is generated from interpolation curves of the vertex of the contour, which enable the interpolating surface to approximate the original object. It has the advantage that the reconstructed surface keep the smoothness in total longitudinal direction. The proposed method is also capable of handling the branching problem. Several experimental results corroborate the theory. The results show the image with high fidelity and the rendering speed are satisfactory and pleasing.

A probabilistic model for robust face alignment in videos

A new approach for localizing facial structure in videos is proposed in this paper by modeling shape alignment dynamically. The approach makes use of the spatial-temporal continuity of videos and incorporates it into a statistical shape model which is called constrained Bayesian tangent shape model (C-BTSM). Our model includes a prior 2D shape model learnt from labeled examples, an observation model obtained from observation in the current input image, and a constraint model derived from the prediction by the previous frames. By modeling the prior, observation and constraint in a probabilistic framework, the task of aligning shape in each frame of a video is performed as a procedure of MAP parameter estimation, in which the pose and shape parameters are recovered simultaneously. Experiments on low quality videos from web cameras are provided to demonstrate the robustness and accuracy of our algorithm.

Parallel volume rendering on distributed-memory multiprocessor system

Several approaches have been proposed to visualize 3D volume data on multiprocessor systems in recent years. Most of these algorithms are tailored for special architectures and hard to immigrate to new architectures, such as Tsinghua High-Performance Cluster System (THPCS), a distributed-memory multiprocessor system. In this paper, a parallel volume rendering algorithm is presented for THPCS. To reduce the communication cost and the complexity of blending local intermediate images, a static subtask distribution strategy is proposed. Combined with the prediction for the subtask load and node performance, our strategy yields a good load balance. An asynchronous binary compositing method is adopted to provide flexibility for our algorithm. The experimental results demonstrate the efficiency and practicality of our algorithm. As a result, THPCS using our algorithm achieves performance ofrendering 512 x 512 color image for 128x 128x 197 volume within 1 second.

New watermarking method using convex chain

Digital image information security is a new research topic which is appearing with the rapid progress of computer network and multimedia. Becoming main measure of information representation, digital image meets the problem how to protect the security of the information it carries. In particular, how to keep the information after kinds of image processing procedure is still an open problem which is not solved. The common way to hide information into an image is to change the values of certain places in the space domain or frequency domain. The novel technique introduced in this thesis, also being a new idea, is to use the topological characteristics of certain chosen points, in detail, the sequence of points along a convex-chain, in the picture to carry information. And this makes the information quite hard to be destroyed by many kinds of image processing.

Parallel shear-warp factorization on THPCS

Several approaches have been proposed to visualize 3D volume data on multiprocessor systems in recent years. Most of these algorithms are tailored for special architectures and hard to immigrate to new architectures, such as the Tsinghua High-Performance Cluster System (THPCS). In this paper, a parallel volume rendering algorithm is presented for THPCS. To reduce the communication cost and the complexity of blending local intermediate images, a static subtask distribution strategy is proposed. Combined with a prediction for the subtask load and node performance, our strategy yields a perfect load balance. An asynchronous binary compositing method is adopted to provide flexibility for our algorithm. The experimental results demonstrate the efficiency and practicality of our algorithm. As a result, THPCS using our algorithm achieves performance of rendering 512/sup 2/ color images for 128/sup 2//spl times/197 volume within 1 second.

New algorithm for contour connection

In this paper, a new solution to the problem for reconstructing the surface of 3D objects over a set of cross-sectional contours is proposed. An algorithm for single branch contours connection, which is based on the closest local polar angle method, is first presented. Then an approach to solving the branching problem is proposed. Next, a method for surface shading is given. Finally, these methods have been applied to the reconstruction of the external surface of a complex shaped object such as the cellar region of human brain. The results show that the presented methods are practical and satisfactory.

Constructing the convex hull of a planar density-bounded integral-points set in linear time

A special kind of points sets named DPDIS is defined in this paper, followed by an elegant algorithm for constructing the convex hull of this class of points sets. Taking advantages of the characteristics of this sets category, this algorithm finishes its work within an O(N) complexity. The crux of this algorithm is a new sorting method by which all points can be sorted along x direction in linear time.