Research on Vehicle 3D Panoramic Model and Texture Mapping Algorithm Based on Low Distortion and Uniform Field of View
Shanshan Feng, Yongtao Jiang, Haibo Huang, Liguo Chen · 2018
The 3D panoramic vision technology of automobile is of great significance in obstacle recognition, obstacle ranging, driving alarm, active avoidance, intelligent driving and unmanned driving, and has become a research hotspot of future automobiles. In this paper, aiming at the difficulties of 3D modeling and texture mapping in 3D panoramic vision technology, a 3D panoramic model consisting of a bottom plane, a spherical connecting surface and an annular conical surface is proposed. The texture mapping algorithm based on texture step is used to map the 2D panoramic image to the surface of 3D mesh model by OpenGL, and a panorama with good 3D visual effects is achieved. Compared with other 3D mesh models the experiment shows that the 3D mesh model proposed in this paper can not only realize the smooth transition between the three parts of the model, but also avoid the texture distortion produced by the spherical 3D model due to its arc surface non-expandability. It can avoid the edge convergence phenomenon of the cylindrical model in the top view, so that the 3D scene around the vehicle is more realistic and can realize any switching of the observation angle. Furthermore, the texture mapping algorithm based on texture step can achieve better texture mapping effect for various models.