| Relics is an important symbol of the historical heritage, is a non-renewable cultural resources, an important carrier of traditional culture education. Heritage conservation is a selection of the inevitable social development. Archaeological excavations are the activities for scientific research to explore the sites of ancient culture or ancient tombs and discover the historical relics. The spatial structure of the scene information and texture information record is extremely significant in the process of archaeological excavation. The three-dimensional reconstruction techniques not only can be used to save the data of the three-dimensional structure of artifacts and scenes, and the reconstruction of three-dimensional information can also be used for dynamic virtual display, so in recent years, these techniques are gradually introduced in digital archaeological work. Due to the non-renewable characteristics of ancient culture, this article studies aimed at how to reconstruct ancient scene with three-dimensional reconstruction technique of high-resolution images based on binocular vision.Firstly, epipolar rectification fails to reach ideal situation because of calibration error. The error in the application of high-resolution images is enlarged and affects the effectiveness of the stereo matching algorithm. In response to these issues, we propose a novel image rectification algorithm based on SIFT features constraints. We use SIFT feature prior information to rectify mage and optimize focal length of camera. It makes image pair forward parallel.Secondly, owing to stereo matching algorithms for high-resolution images on the memory and speed limitation, we propose an original stereo match algorithm based on pyramid structure. We reduce the search range of disparity with this technology. And this can reduce the algorithm complexity and space complexity.Finally, we have implemented a prototype system. And we apply this system in some of the ancient culture sites. |
PDF Full Text Request