June 1995
Intermediate to advanced
438 pages
10h 53m
English
Jean-François Doué and Ruben Gonzalez Rubio, HEC Paris, France; University of Sherbrooke Canada
In computer graphics applications, it is often convenient to be able to convert a shape from its bitmap representation to a vectorized form. Indeed, a vectorized form has many advantages over a bitmap: It can be transformed without loss of resolution, and it is sometimes better suited for the needs of computer vision. This gem presents an algorithm to perform the vectorization of arbitrary 2D shapes (a shape is defined here as “a set of contiguous pixels of the same color”). We developed the algorithm for a computer vision application: The purpose was to study how to train a neural ...
Read now
Unlock full access