Research Of Rapid Three-dimensional Reconstruction Based Digital Of Ultra-high Resolution Mural

Mural is one of the oldest form of humanity paintings, its rich content make it become an important research data for political history, cultural history and so on. However, because of weathering, erosion, rain erosion, fading and other diseases, Mural is disappearing. As Image Stitching can permanently save the information of murals, it has been used in the digital archaeological research of mural recently.However, since the image registration error and other factors, the stitch result appear lines distorted, or lines disconnected. Moreover, algorithms like multiresolution spline algorithm which cost a lot of time is used during image fusion, resulting in low time efficiency and image blur.To solve the above problem, this paper proposes a rapid three-dimensional reconstruction based mural digitizing framework, in the first time three-dimensional reconstruction is used in mural digitizing. The main work includes the following sections:Proposes a rapid three-dimensional reconstruction based mural digitizing framework. The framework improve data integrity by collecting the three-dimensional information of mural, which solve the lines distorted and lines disconnected problem, moreover, improve the mural digitizing speed and get a 3D model.Proposes a fast relative position based SIFT feature matching algorithm. The algorithm estimate the relative positional relationship between each image firstly, then search the possibly SIFT features in a Quadtree. The algorithm improve matching speed and accuracy.Proposes a fast render algorithm for ultra-high-resolution textures of larger 3D model. Due to the huge mural model, its texture can reach hundreds, the algorithm breakthrough the limitation of graphics memory, and improve the speed of rendering.