Shape recognition of CAD models via iterative slippage analysis

• We present a new shape recognition method by iterative slippage analysis.
• The exact normal is found to be one of the key points for slippage analysis.
• The appropriate region is found to be the other key points for slippage analysis.
• A knowledge guided region growing method is used to get the appropriate region.
• An iterative normal modification method is used to obtain the exact normal.

A new slippage analysis method for recognizing basic primitive surfaces of CAD models is presented in this paper. Obtaining the exact normal and searching the appropriate local region of each point are found to be the key steps for determining the local slippage motion type. First, the tensor voting-based boundary point recognition method is integrated to preprocess the original points. Then, the local slippage analysis method is used to initialize the point type. Furthermore, the appropriate region of each point is acquired by the region growing method. Meanwhile, the middle level information (the basic primitive surface types and the representative parameters) is found, guiding the modification of the normal of each point and the iterative detection of the surface types. Finally, the middle level information-based smooth method is introduced to refine the boundary of each basic primitive surface. The empirical results show that the proposed algorithm is efficient and robust for recognizing primitive shapes from CAD models of mechanical parts.