Research On Regular Surfaces Extraction In Feature Based Reverse Engineering

With the development and application of information technology, the product manufacturing takes on a global trend. The international competition is becoming keener, which is focus on the capability of product innovation of a country. The life cycle of a product gets shorter and shorter, and the product innovation is developing at an ever-increasing rate. Reverse engineering has become very popular in recent years, as an important technology and method that supports the rapid design of product to promptly occupy the commercial market, In order to reconstruct a CAD model to support the product innovation design, this dissertation will research the related key techniques on the regular surface feature extraction from point cloud.The existed re-modelling methods can be divided into two classes: non-feature based method and feature based method. In this dissertation, the basic theories and strategies on feature based reverse engineering are mainly researched. Then the feature classification and parameter representation in reverse engineering are pointed out. This dissertation specifies the types of surface in regular surfaces, which will be researched in detail. A new feature based re-modeling strategy is proposed to successfully reconstruct the CAD model.The statistic theory is introduced into the blend feature extraction to identify the type and determine the parameters of blend feature. And some important parameters values associated with the blend feature extraction are greatly modified, such that the improved algorithm is more stable and applicable. At the same time, the determination of initial value used in fitting cylinder is entirely changed, which makes the cylinder fitting algorithm more stable. On the basis of blend feature extraction, pipe surface extraction from point cloud is presented.The conception of principal direction Gaussian image (PDGI) is created. The PDGI of a surface reveals the information of the feature parameters of the surface. And the PDGI of a rotational surface possesses one and only one great circle associated with the direction of the rotational axis. So a PDGI based algorithm is presented to extract the direction of the rotational axis from point cloud. Then the positional point of the rotational axis is obtained from the normal lines of the rotational surface and the estimated direction of rotational axis through some calculations. Finally, an optimization method is applied to refine the translational direction and the rotational axis.Two principal direction based methods on translational surfaces extraction are...