| Research Interests |
CAD/CAM/CAE, Geometric Modeling, Differential Geometry, Computer Graphics, Solid Modeling, Physically Based Modeling |
| Current Projects |
Robust Solid ModelingFull automation of the design of complex structures requires reliable geometric computations. This reliability is absent from current CAD/CAM systems due to their implementation in floating point arithmetic, which leads to inconsistent and unstable numerical computations. We are developing a solid modeling system based on interval splines, which differ from classical splines in that the control point coordinates are represented as intervals, which in combination with rounded interval arithmetic guarantees the robustness of geometric computations. A current project involves conversion of legacy geometric models to robust interval solid models that are topologically valid and geometrically consistent. Producing rectified models free from gaps and inappropriate intersections is necessary for effective analysis and manufacturing.Click here for more information about I-TANGO. Click here for more information about SGER. Shape Intrinsic Fingerprints for Free-Form Object MatchingThe objective of this project is to develop an intrinsic watermark technique for solids bounded by NURBS surfaces. The key idea is to extract intrinsic properties of solids, which are not affected by coordinate transformations, random noise and malicious action of the user. This watermark can be destroyed only if the digital model describing the shape is changed so much that the newly represented object cannot any longer be considered approximately identical to the original solid in the database. Click here for more information about this project. |
| Past Projects |
Laser Measurement and Reconstruction of Curved PlatesAn automated system has been developed which acquires range image data from a curved plate, analyzes them and yields a suitable mathematical representation. The plate is assumed to be topologically equivalent to a four sided free-form surface representable in an explicit, non-parametric form, whose tangent values in every direction are small. The system consists of three parts: a laser scanner, a robot and a workstation. Each system is connected through TCP/IP and a serial line so that all operations are synchronized. The limitation of the scanner that the size of the rectangular array of points produced by the scanner is too small to capture the whole shape is avoided by attaching the scanner to the robot so that enough shots to cover the whole object can be taken by placing the scanner at appropriate positions. The procedure starts with detecting roughly the edges of the plate, which is necessary to compute the number and poition of the shots that are needed to obtain the full model. Edges are detected by investigating changes of tangent values along each raster line of the scanner. The rough estimation of boundary is followed by scanning of the plate that starts from a corner and sweep through the whole plate in a zig-zag manner. Registration technique is used to accomplish the congruence of geometric representation of the model with a set of arrays of points. Data points sampled along raster lines of digitized points are approximated by B-spline curves, which are lofted in the least squares approximation scheme so that the B-spline surface approximating the plate is obtained. Examples are given and comparison between the measured data from the plate and the B-spline surface representation is made. Computer Program showing a surface of a plate ![]() Surface Construction from Scanned Data ![]() |