Reduction And Surface Reconstruction Of Point Cloud Data Of Automotive Seat Based On B-spline

Along with the rapid development of digital technology, reverse engineering has become the principal means of parts modeling. Reverse engineering technology is based on the existing product model, using the specialized equipment and software to obtain the digital product model to develop new products, with some processing, innovation and re-creation to the digital model. That is the Process form the ohysical model to the CAD model. Reverse engineering process in general can be divided into four steps: Data measurement, data processing, model surface reconstruction and CAD Model Checking. One of points processing and surface reconstruction are key points, is also difficult Points.Along with economic development and the new technological revolution, international trend of automobile production has been more and more obvious. As an important component of vehicle, seat's design has developed rapidly. Surface reconstruction of automotive seat at home and abroad mainly depend on the software of CAD.etc. Designers use a variety of measurement tools to obtain the point cloud data of automotive seat. Then process the point cloud data of the seat and recontruct surface through manual operation. This design method requires a longer design time, and accuracy is low, efficiency is bad. Therefore, this article through the research of reading, displaying, reduction, curve fitting, data segmentation and surface reconstruction techniques to implement the automated design of the reduction and surface reconstruction of the automitive seat. This method is to use the computer to replace the traditional manual method of operation. The design process has high efficiency, high precision and high speed. On the reserch of automotive seat's design has high value of application and theoretical significance.In this paper, the point cloud of automotive seat is large. Therefore, need processing to the point cloud data before the surface reconstruction. Taking into account the processing of the two-dimensional point cloud data than the three-dimensional is simple and the point cloud data of seat showed the distribution of scan line. Therefore, this paper makes the point cloud data stratification through definiting one group of cut-off plane. The point cloud data of seat are spatial points. when using the cut-off plane to make stratification, the possibility of points falling on the cut-off plane exactly is very small. Only the points landing on the cut-off plane exactly usually can not reflect the actual shape of the entity model. In this paper, Firstly to detemine the corresponding point cloud data of every layer according to the distribution of the point cloud data of the parts of the seat. Then get the points of every layer by usingthe method of projection. The points which get using projection method move a small distance in the same direction. The basic profile of the model is unchanged and little impact on the results. For the follow-up processing to the point cloud data, use binary insertion sort to make the point cloud data of every layer sort according to clockwise circle.Along with the development of scanning technology, the speed of data acquisition has a sharp increase. The huge amount of points brought surface reconstruction great difficulties. Therefore, it needs reduction. The traditional methods of reduction include the method based on wrap around box, the method using grid the method based on curvature and the method of triangular grid. These methods have certain disadvantage. Therefore, this article put forward a method through the comparision of the distance of point to cubic Beizer curve and the error provided according to the characteristics of the point cloud of seat. To the point cloud data of every layer, reading a group of points in poper order. Then pick up four points as the control points of the cubic Beizer curve. Calculating the distance of the points of that group to the cubic Beizer curve and comparing with the error provided. If the distance of every point of the group is smaller than the error provided, using the selected four control points in place of this group of point cloud data. This method of reduction achieve the goal of reduction while ensure the shape of the model. This algorithm has an important question is how to calculate the distance of the point to the cubic Beizer curve.Cubic Beizer curve is a cubic polynomial based on the parameter. About the distance of the point to the polynomical curve, obtaining the distance formula of the point to the curve by the mathematical formulas. Then using the method of calculated the differential coefficient to get the extremum. Through this method, we get the distance formula is five polynomial. Calculating the solution of five polynomials is difficult work. On the basis of the features that the surface is indicated by lines, the line is indicated by points in the computer graohics. We calculate the distances of the point to the points which is indicatint the cubic Beizer curve. Then get the smallest distance through comparison. Cubic Beizer curve is a special case of cubic B-spline curve. So this paper uses cubic B-spline method to fit the curve model of the automotive seat.In Reverse Engineering, the surface of the product carry out a complete description by one surface is usually impossible. It is composition of multiple surfaces. Therefore, it needs segmentation for the entire point cloud data. Though segmentation, data processing and surface reconsturction will be more simple. There are two commonly methods for data segmentation. One is the manual method, and the other one is automated segmentation method. The study of the automated segmentation mainly includes data segmentation based on the side and data segmentation based on the surface. The base of data segmentation is feature extraction. Feature extraction is mainly extracted the feature points for generate the feature curve and surface from the measured point cloud data. At present, the research on feature extraction, there is no ideal method of feature extraction and recognition. In this paper, we put forward a senmi-automatic method of data segmentation by the combination of manual and automated segmentation, according to the characteristics of distribution of point cloud data and production assmbly of the part of automotive seat. Fistly extract the boundaries area by manual extraction to get a rough segmetation of point cloud data. Then extract the feature points by the curvature analysis of the points in the boundaries area to get the precise data segmetation. The feature points usually are the one that is the mutation curvature point. This paper makes the local parabolic fitting in the neighborhood points, and calulates the curvature of the parabplic as the curvature of the point. Then we extracte the feature points by comparing the curvature and the fixed value.The point cloud data is divided into several different four-side areas by data segmentation. It needs the control vertexs and knot unity, then uses cubic B-spline to fit the surfaces. In this paper, we implement the unity of the control vertexs by each interpolation on the direction of each parameter. For the suture of the surfaces, by definiting the boundary points as four same konts, and the surfaces interpolate the boundary points. About the rough boundary, generating the transition surface by interpolation based on the boundary points of the base surfaces.Using this method, implementing the smooth suture of the base surface. And get the complete surface model of the parts of the seat.In this paper, we develop the software of point processing ande surface reconstruction of the automotive seat by the tools of Visual C++ and OpenGL. This software can finish points reading, points reduction, curve fitting and surface reconstruction. This software has implemented the points processing and surface reconstrucion of automotive seat automatically design. And by definting the different views to convennient observation. The software get to the disired result.