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
Curve fitting for styling application
by genetic algorithm
Yves Mineur. Marc SevauxUniversity of Valenciennes –LAMIH
BP311 59313 Valenciennes Cedex 9 – France
Abstract
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 classicalmethods. Stochastic ones, such as genetic algorithms, therefore appear as a relevant way of solving. We considerhere a curve fitting problem aiming the generation of shapes with specific curvature variations. A particular curvemodel have been developed with this aim. Its implementation within a genetic algorithm have been carried out.We describe its main characteristics and present a first result.
1.Introduction
The design of freeform shapes occurs in the engineering of an increasing number of products such as cars, mobilephones or any consumer goods whose shape has been specially designed with a styling intent. A recurrent task insuch a design consists in the fitting of curves and/or surfaces on measured data in a reverse engineering process.Geometrical requirements attempted in freeform shapes design are very often numerous, more or less compatibleand sometimes severe. The problem of data fitting, which can be turned in terms of optimal design, is thereforegenerally addressed through optimisation techniques[1] [4] [5]. There are however cases where one faces withhighly constrained non linear problems whose solution is cumbersome and difficult to handle with classicalmethods.
Solving techniques such as genetic algorithms appear here as an appropriate and alternative way, as someexisting works done in the field of curve fitting show [6] [10]. They are however dedicated to smoothingproblems in general design and do not proceed to a fine control of the shape as required in aesthetical shapesdesign.
The application we are developing and we will present here concerns the fitting of constrained Bézier curveswhose curvature variation insures the generation of visually pleasant shapes.
2.Composite curve definition.
Shapes playing an aesthetical role must fulfil constraints mainly related to their curvature variation. Such visualfeatures are traditionally expressed through the profile of the section lines used in surfaces construction.Harmonious curvature variation are thus expected on these lines to ensure the attractiveness of the consideredshape, which is mathematically translated in terms of curvature monotony and/or convexity [3].
In a previous work [7], we defined a particular planar bezier curve called typical curve aiming at the generationof shape with a monotone curvature variation. A typical curve possesses a control polygon with a same angle anda same length ratio between each adjacent polygon edge. Such properties provide the typical curves with amonotone and highly smooth curvature variation, quite the same as involutes of circle, and with other interestingmathematical properties.
To cope with a high variety of shapes and to give to the typical curves much more versatility, we develop in another work [8], a piecewise curve model, curvature continuous G2, where each piece is a cubic typical curve (fig.
1). The definition of such a curve is given by a set of shape-parameters which are from one side, the location of aend point Q0, the tangent direction W0 and curvature 0 at this point, and from an other side, the arc length Siand the angular variation of the tangent direction Wi of each curve segment.Fitting such a curve to a data points set is not easy for many reasons. First of all, the function giving the curvelocation is a non linear function of the curve shape-parameters. The curve nodes position being variable, mainlythrough the variation of the arc length parameters, the curve parametrization cannot be fixed arbitrarily. Thus,that ordinary least-squares methods cannot work in this case [9].
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