Visualization of Proximity Objects and Refinement of Characteristic Line from Biased Samples

Z. Izhakian (Israel)


Visualization and HMI, Multi-dimensional Visualization, Parallel Coordinates, Line Refinement, Approximated Objects


Applications in many applied areas of modelling are based on recurrent sampling of geometric patterns. These specimens are usually biased, thus a preliminary goal in order to obtain a clear and comprehensible im age is refinement of a characteristic object from a col lection of distorted samples. This is done by establish ing a visualization method which enables the simulta neously representation of multiple occurrences of ele ments. Generally, objects that are linearly defined and embedded in 3-dimensional space can be characterized by their edges ("slices" of lines). Consider, a collection of lines generated by a family of proximity edges. The question discussed here is this: how can we visualize and then approximate (in terms of line's coefficients) this collection by a single line ? Namely, refine (i.e. recover) the source line. A solution to this issue is the basis for refinement of objects. The visualization of linearly defined objects in Parallel Coordinates is constructed from the funda mental point line duality, thus their representa tions are determined by the coefficients of their defin ing equations. Composing this representational advan tage with additional geometric methods yields simple planar visualization, and an efficient solution to the refinement problem, where both can be applied to any space dimension. Algorithm based on this concept of representation which uses dual images of lines and de termines an approximated line is also obtained. Addi tionally, it turns out that our approach can be nicely generalized to more complex multi-dimensional regions and approximated curves.

Important Links:

Go Back