This project focuses on the comparison of multiple meta-heuristic 3D mesh optimization algorithms. The aim is to address the challenges of simplifying complex polygonal meshes while preserving essential characteristics for efficient rendering.
- Problem Statement: The project tackles the NP-Hard problem of mesh simplification in computer graphics, aiming to reduce the computational burden during rendering.
- Algorithm Comparison: The team evaluates the performance of three meta-heuristic algorithms: Selection-Reproduction, Genetic Evolution, and Multi-Objective Optimization.
- Converts 3D mesh to 2D space.
- Uses Delaunay Triangulation to create triangles.
- Iteratively minimizes errors by selectively adding or removing vertices.
- Defines a chromosome representing mesh points.
- Utilizes Delaunay triangulation for visual representation.
- Applies mutation and crossover operations for evolution.
- Maintains a population of individuals representing simplified meshes.
- Evaluates fitness based on error minimization and vertex count.
- Conducted experiments on different datasets, revealing a trade-off between time complexity and accuracy.
- Multi-objective optimization algorithm demonstrated superior results.
- Computation time is a concern, suggesting potential for optimization.
The project concludes that the implemented algorithms effectively simplify 3D meshes, with the multi-objective optimization algorithm showcasing the best performance. While computation time remains a challenge, opportunities for parallelization and optimization exist to enhance efficiency.
The team suggests exploring less computationally intensive error functions without compromising accuracy. Additionally, parallelization and optimization techniques could further improve the efficiency of the algorithms.