The fitting of curves in computer aided geometric design is generally regarded as an optimisation problem. According to the application, conditions to be satisfied can make the problem difficult to solve with classical methods. Stochastic ones, such as gen
If offspring_fitness > worst_fitness
Then Discard it;Else Insert it into the population;
Select randomly an individual for deletion;Endif;
Endif;Endwhile;
End.
The selection procedure is based on a binary tournament selection scheme. The probability rule fixing themutation rate, is the same as the one used in [10]. Let pm the mutation probability of an individual. Its calculationprocedure is :
If offspring_fitness mean_fitnessThen pm = 0.5 (offspring_fitness – best_fitness) / (mean_fitness – best_fitness);
Else pm = 0.5;Endif ;
The generated populations, in the experiments we performed, comprise 30 individuals. Figure 3 shows two fittingresults on a same data points set, by taking a two and three segments composite curve. The number of performediterations was respectively 166 and 244. In accordance with the practice, a closest approximating curve isobtained by increasing the number of segments. Such a subdivision, as illustrated on the curvature diagrams,works as a shape refinement and allows to reveal some details on the curvature variation. The Curvaturevariations of the obtained curves clearly shows the high quality of the typical curves shape.
5. Conclusion and future work
In this work, we have developed the basis of an original curve fitting method dedicated to the reverse engineeringof styling shapes, using a stochastic approach. The first results obtained are promising. Nevertheless, somedifficulties we encountered remain. As previously outlined, the domain of some variables, such as the tangentdirection at the curve endpoint, can strongly influence the algorithm running and cause convergence on localoptimum. Tests have shown that the mutation does not always suffice to prevent such a situation. The problemalso seems to lie in the correlation between the initial curve variables and the segments variables. The influenceof these last variables on the fitness function depends on the initial variables, thus generating an epistasis [2] likephenomenon.
Without considering the needed improvements to make our algorithm more robust, further new applications ofour method could be addressed in a next step. Our template curve definition can be indeed extended to theconstruction of large variety of shapes, by combining typical curves with increasing or decreasing curvaturevariations. Curves with non monotonous curvature could be generated in that way. Shape analysis or recognitionon clouds of points, by testing different template curve definitions, could be, in the same way, considered.
Abstract Adaptive T-spline Surface Fitting to Z... Computational Geometry and Object Modeling—curve,...styling, NC machining, GIS, and computer vision....
Abstract Conceptual design in the automotive ...for a design application where creativity should ...The resulting 2D curve is projected back into 3D...
deforming 3d shapes in biomedical and styling ...Abstract In contrast to machined mechanical parts,... Computational Geometry and Object Modeling—Curve,...
deforming 3d shapes in biomedical and styling ...Abstract In contrast to machined mechanical parts,... Computational Geometry and Object Modeling—Curve,...
热门文档