Abstract: Advances in laser scanning technologies have facilitated sampling part surface data with speed and accuracy. It is necessary to manipulate these large amounts of point data. The adaptive minimum distance (AMD) method is proposed in this paper which is a kind of data reduction algorithm that balances efficiency and precision. Firstly the noise data is eliminated by median filtering and wavelet packet filtering, and then the curvature of the point data are analyzed and different zones are separated where a suitable minimum distance is selected. The points where the distance is larger than the given distance are neglected. This novel algorithm has merits of both precision and efficiency with the ratio of reduction being 36％.The method is applied to two sample models, and experimental results illustrate the feasibility of the new algorithm.

Key words: reverse engineering, curvature analyzing, minimum distance method, data reduction