Surface Reconstruction Technology Based On 3D Stereo Calibration

With the increasingly development of the laser measurement technology,the 3D topographic reconstruction of surfaces based on structural optics has received great attention,especially combined with digital image processing technology,because it is convenient to complete the reconstruction,and the method has high precision,stable results,and won't damage the object's surface.So this new technology is widely used to quality inspection,mechanical manufacture and the reverse engineering field which has special request for precision and environment.This thesis completes the 3D reconstruction of the object's surface using the method of linear structured light scanning.First of all,this thesis uses a template to calibrate CCD camera,then uses CCD camera which is calibrated to collect an image of light band that is modulated by the surface of the object and extracts the center of the light band.After that,the three-dimensional coordinate information of the coverage area of the optical strip center can be calculated.At last,let the linear structured light to scan the whole surface of object,the 3D contour information on the surface of the object can be obtained.The main work and innovation points of this thesis are as follows:(1)To analyze the characteristics and error sources of different calibration templates,and design a calibration template.(2)To study the algorithm for extracting the center of the ellipse,and present an elliptical random fitting algorithm.Then optimize the two factors,the number of fitting points and the number of fitting calculations,which influence the accuracy and speed of the algorithm most,and obtain the optimal values for this thesis's experiment.The experiment proved that using the elliptical fitting method of this thesis can improve the positioning accuracy of the ellipse and the positioning accuracy is about 0.03 pixels.(3)To study the mathematical model of camera imaging.Then analyse Tsai's two-step method and Weng's iteration which are representative in detail.And propose a simple and fast stereo calibration algorithm.The results turned out that the average pixel error can reach 10~-22 orders,at the same time,the computing time is shortened obviously.(4)To study the central extraction algorithms of linear structured light,analyze the Steger's algorithm based on Hessian matrix and its improved algorithms,and propose a marginal algorithm based on Hessian matrix.Using this algorithm,the number of calculated data points can be effectively compressed to ensure the high precision and high robustness of the algorithm,meanwhile the calculation speed of the algorithm is improved.