Reverse engineering of pipe layouts and 3D point set damage models

This paper focuses on obtaining pipe layout information from depth images, and on the development of damage models for 3D shapes. Techniques are given for recognizing various pipes and pipe features. Methods for generating compact geometric descriptions of pipes and pipe features are discussed. These methods and techniques are analyzed for error, and error measurements are given. 3D shape damage models are proposed based on a separable Gaussian kernel model. Experiments on synthetic and real data were performed and results discussed.