Triangular mesh simplification on the GPU

摘要

We present a simplification algorithm for triangular meshes, implemented on the GPU. The algorithm performs edge collapses on manifold triangular meshes driven by a quadric error metric. It uses data parallelism as provided by OpenCL and has no sequential segments in its main iterative structure in order to fully exploit the processing power of the GPU. Our implementation produces results faster than a corresponding sequential implementation and the resulting models are of comparable quality.