GENETIC ALGORITHMS FOR APPROXIMATION OF DIGITAL CURVES WITH LINE SEGMENTS AND CIRCULAR ARCS

A method for error-bounded approximation of digital curves with line segments and circular arcs is proposed. In the proposed method, a chromosome is used to represent an approximation by a string, and the operations of genetic algorithms are designed especially to obtain the approximation for which an error is bounded by a given norm. Many experiments show that the convergence is guaranteed, and optimal or near-optimal solutions can be obtained. The proposed method successfully reduced the total number of line segments and circular arcs required to approximate the digital curves under the error-bound criterion.