Research On Color Texture Reconstruction Method For The Object To Be Tested

Three-dimensional reconstruction technology is to accurately restore the real appearance of objects,which has been widely used in various fields in the current information age.The binocular vision system can reconstruct the three-dimensional shape of the object according to the parallax principle,which has the advantages of high efficiency,simple system structure and low cost.However,the three-dimensional model reconstructed by two black and white cameras cannot show the original color information of the object,so a color camera is added to the system to collect the color texture information of the object.Finally,the real three-dimensional model of the object is obtained through space mapping.In this thesis,the spatial mapping relationship is used to map the image texture information from the single view Angle to the 3D model.Then,adding color elements,the color information of the collected color images is mapped to the three-dimensional model through the above texture mapping method,and the color texture reconstruction from a single perspective is completed.Then,the three-dimensional models collected from multiple perspectives are stitched together into a more complete model,and the model is triangulated.Finally,the color images collected from each perspective are mapped to the complete model mesh surface,and a more complete threedimensional color texture model is obtained.The main research contents are as follows:Firstly,according to the basic principle of binocular vision system imaging,the transformation relationship between binocular images two-dimensional coordinate points and three-dimensional model space points is obtained.Based on this,the corresponding relationship between point and image is derived and the texture mapping methods are designed.A high-resolution binocular camera was used to collect black and white images of the object to be tested and generate three-dimensional point clouds.According to the designed texture mapping method,the binocular left two-dimensional coordinate points were mapped to the three-dimensional model to complete the mapping of texture information to three-dimensional space,and a three-dimensional model with texture information was obtained.In order to impart color information to the point cloud model,two new color texture reconstruction algorithms are designed when the pixel of binocular camera is larger than that of color camera.The color camera with lower resolution than the binocular camera is used to collect the color image of the object to be measured.Because the coordinate system of the image collected by the color camera is different from that of the binocular camera,the point cloud reconstruction process is based on the binocular left camera.Only after the interpolation and registration process of the collected color image can the color texture reconstruction be accurately realized.In this paper,the double cubic interpolation algorithm is used to interpolate the color image to the same resolution as the reference image,and then SIFT algorithm is used to extract the feature points of the color image and binocular left image after interpolation,then the feature matching is carried out,the transformation matrix is calculated to complete the image registration,and then the spatial information transfer is realized according to the texture mapping method,and the color texture model under the single perspective is obtained.Finally,in order to obtain multi-view color texture model,a new multi-view cloud color texture reconstruction algorithm is designed.Binocular camera and color camera simultaneously collect the information of objects under different poses,so that the data collected for many times can cover the multi-angle surface of the object.Firstly,the collected point clouds are spliced,and the sign-based point registration algorithm is used for rough registration,and then the ICP improved algorithm is used for fine registration.After the complete point cloud model is generated,the Delaunay algorithm is used for triangular meshing,and the texture color information of each color image is mapped to the triangular surface.And finally the multi-view real color three-dimensional model is obtained.