Sparse Pixel Vectorization: An Algorithm and Its Performance Evaluation

  • Dov Dori ,
  • Liu WenYin

Published by Institute of Electrical and Electronics Engineers, Inc.

Publication

Accurate and efficient vectorization of line drawings is essential for their higher level processing. We present a thinningless Sparse Pixel Vectorization (SPV) algorithm. Rather than visiting all the points along the wire’s black area, SPV sparsely visits selected medial axis points. The result is a crude polyline, which is refined through polygonal approximation by removing redundant points. Due to the sparseness of pixel examination and the use of a specialized data structure, SPV is both time efficient and accurate, as evaluated by our proposed performance evaluation criteria.