Due to the complex geometry

scattered position distribution

and obvious curvature change of the defect holes in aero-engine blades

it is difficult to extract the boundary feature points completely by using a single criterion. To realize the accurate boundary extraction of blade defect holes and smooth fitting of the holes boundary line

an algorithm for extracting boundary feature points of aero-engine blade defect holes based on double criteria was proposed. The fitting plane was established based on the spatial topological structure of the sampling point and its K neighborhood points

which transformed the three–dimensional coordinates into two–dimensional coordinates by the local coordinate system. By fusing the criterion of the normal angle of sampling points with the adjacent field power force criterion

the eigenvalue and discrimination threshold of sampling points are obtained by the weighted method

which can be used as the criterion for extracting boundary feature points. The validity and accuracy of the algorithm were verified by two sets of experiments

the experimental results demonstrate that the method can effectively extract the boundary feature points in complex holes model.