Method of the Square — A new algorithm for image vectorization

This paper presents a new algorithm `Method of the Square´ (MS) for automatic image vectorization, which is implemented and tested here against some classical vectorization algorithms (ArcInfo, Contour Fitting and Sparse Pixel Vectorization). The comparison between these algorithms is performed by using widely accepted performance metrics introduced by Wenyin and Dov Dori (such as pixel recovery index pixel - PRI, pixel detection rate - D_p, false pixel detection rate - F_p) and shows that the algorithms discussed here achieve comparable performances. The experimental results qualify the proposed MS algorithm as a cheaper alternative for image vectorization.