A Novel Method for Vectorization

摘要

Vectorization of images is a key concern uniting computer graphics andcomputer vision communities. In this paper we are presenting a novel idea forefficient, customizable vectorization of raster images, based on Catmull Romspline fitting. The algorithm maintains a good balance between photo-realismand photo abstraction, and hence is applicable to applications with artisticconcerns or applications where less information loss is crucial. The resultingalgorithm is fast, parallelizable and can satisfy general soft realtimerequirements. Moreover, the smoothness of the vectorized images aestheticallyoutperforms outputs of many polygon-based methods