3D modeling for an airplane with close-range sequence images

摘要

This paper presents a method of 3D modeling for an airplane based on the typical theory of close-range photogrammetry and the principle of the multi-ocular vision in computer vision, including two main steps: computation of image parameters and creation of 3D airplane model. Because the taken images are provided with the characteristics of short-baseline and multi-images, multi-baseline stereo matching technique should be applied to solve the problems of the distortion discordance between close-range and far-range images in the close-range photogrammetry and dissatisfied intersection angle. In this way, it could greatly enhance the automatization and reliability of matching. After the image parameters are calculated, the 3D spatial coordinates of discrete points on the airplane surface are computed by forward intersection of multi-images. A Delaunay Triangle Network (DTN, satisfying the Circle Rule of DTAD) with boundary constraint is constructed from the points. Among the process, the idea of the partition of surface pieces based on spatial structure relationship is applied. Real three dimension airplane model is visualized in OpenGL finally. Experiment shows that this method of modeling for an airplane provides high precision results.