Real-coded genetic algorithm in superquadric fitting

Superquadric parameter extraction is essential for superquadric-based reconstruction from 2D images and 3D data, but most of the search algorithms for superquadric parameter extraction are suboptimal and they are susceptible to being trapped into local optima. In this paper, we propose a search based on a real-coded genetic algorithm (RCGA) for parameter extraction, which applies the genetic algorithm to superquadric-based fitting computation. Numerical fitting experiments for comparison of GA parameters and genetic operators are carried out. Results obtained show the efficiency, robustness and accuracy of the RCGA-based search algorithm, which not only solves the problem of being trapped into local optima, but also performs quickly and reliably for superquadric fitting.