Abstract: In reverse engineering, so as to obtain a feature- based parametric model, surface feature extraction techniques are deeply studied in recent years. However, another same important problem that how to extract sectional feature curves from slicing data points is almost disregarded. Aiming at solving the issue, the discrete curvature estimating, data segmentation, feature recognition and constraint-based global optimization of slicing data are studied and realized. With these techniques, the best fitting curves can be acquired, and the continuity constraints between the curves can be satisfied too. The application proves that surface construction based on feature curves can greatly improve the efficiency of modeling and accuracy of model in reverse engineering.

Key words: Sectional feature curve, Global optimization, Reverse engineering, Slicing data