Dataset reduction in a three
dimensional home environment
Abstract:
Most geometrical surfaces are made up of large number of triangles. Many algorithms have been developed to reduce the large number of triangles that are required for the approximation of objects or models.
For a given three dimensional surface or model, dataset reduction algorithms continuously and iteratively remove triangles by triangulation. The distortion made to the original model should be minimal. Most reduction schemes use an upper bound which guarantees the reduction level attained with minimal distortion.
The order in which triangles are considered for reduction is based on a parameter at their vertices such as curvature, Hausdroff distance etc. Parameter weights are computed for every triangle. Triangle with lower weights are identified and placed in a dendrogram tree. This tree structure provides multi -resolution capabilities of the project. The tree structure facilitates appropriate edge removal to attain desired resolution.
For this project, dataset reduction scheme will be implemented on a home environment model and then tested. Testing is based on file size, time, accuracy, amount of user control, degree of resolution and degree of reduction attained.