Light Pen Measuring Three-dimensional Modeling Study

Due to the tense international competition in the market, rapid and precise 3D modeling technology plays a more and more important role in product development and has attracted considerable research attention. Based on the study of Binocular Stereo Vision 3D modeling technology, a light-pen dual-cameras 3D modeling technology has been devised for this purpose. It was a kind of hybrid modeling based on geometry and image, which can overcome the inefficiency of the Geometry-based modeling, and improve the short precision of the Image-based modeling, thus realize the purpose of fast and accurate 3D modeling.With binocular vision theory, image data preprocessing, binocular vision matching, camera calibration,3D reconstruction problem of Light-pen vision 3D modeling technology has been discussed in detail. The main work are as follows:Since the mathematical expression of the primitive is generally refer to gray value, the basic information of image is extracted in group to calculate the gray value of every pixel point. For object shape feature recognition, the edge feature of image is detected by using an improved Canny edge detector. Two improved feature extraction approaches based on edge was raised. One method was achieved by introducing the concept of the connection number, which was ready to be used in the chain code tracking of the edge. Then, by the use of polygon approximation algorithm, the intersection point of each two side was taken as the feature point. Harris corner detection algorithm was inserted to the other method. The value of Harris corner detector on binary edge was computed, the maximum value greater than the threshold was chosen as the feature points.Under the premise of the highest Normalized cross correlation constraints, stereo matching on the edge-based feature points was carried out with cross-correlation inspection. In order to contribute to dense disparity, a new light pen measurement seed growth interpolation algorithm is proposed. It started from a set of matched feature points to near interpolate points on a region growth principle. To counter the low accuracy of weak texture region matching, with the combination of light pen measure date, disparity gradient restriction was increased in matching process.Camera calibration experiment with Zhang algorithm is used to obtain the internal and external camera parameters, which were linked up with parallax to acquire the spatial coordinates of matching points. By the adoption of triangular mesh,3D model was established with the discrete point set. With a view of rapid and precise modeling,3D model surface was corrected with light pen measured data. Finally, by means of Open Graphics Library, the 3D visualization of the model was realized. Equipped with standard file exchange interfaces of STL format document and BMP Image File, the system can be integrated with other commercial CAD/CAM softwares conveniently and efficiently.