Feature Extraction And Shape Detection In Reverse Engineering

Feature curve extracting from triangular mesh should be regarded as an important algorithm in constructing surface model from unorganized point cloud. A two-step feature point collecting algorithm was introduced to reduce the error due to the disproportion of triangular surfaces. Then feature curves were generated from feature points, and some adjust and smoothing algorithm like branch cutting, feature points and feature curves collapsing, feature points bending, feature curve close up, were applied to gather better quality feature curve from the original curves. Surface segmentation and recognizing algorithm can be improved by applying the feature techniques proposed in this paper.In reverse engineering, the extraction of ridge lines was an important step in the process of triangular mesh converted to surface models. This paper presents a robust and fast method for extracting ridge lines on surfaces approximated by dense triangular mesh. The main direction of the surface and curvature derivatives was estimated combines with local polynomial fitting The ridge points were identified by used the curvature of the maximum/minimum, then got the initial ridge by connected thest ridge points. Finally, the spurious ridges were filtered by used threshold.The method presented in this paper is fast since it estimate necessary surface derivatives via local polynomial fitting, and can deal with the model with more noise,the algorithm is robust and efficient.The extraction of geometric primitives was very important in reverse engineering. This paper present an algorithm to detect primitive shape in unorganized point clouds.The algorithm decomposes the point cloud into a hybrid structure of inherent shapes and a set of remaining points.This method is based on random sampling,to identify plane,cylinder,cone,sphere and torus.The extraction algorithm can be used with multiple geometric primitives.