We present a pipeline to reconstruct the 3D structure and color from underground image sets and to present the heritage with an interactive system.Our approach employs a Structure from motion (SfM) algorithm to recover the pose of the cameras which will be used to reconstruct the 3D structure of the scene and to render the images in a visual reality system based on state-of-the-art image based rendering techniques.Then the polished pose of cameras and the coordinates of 3D points from triangulation through bundle adjustment are used for the representation of the relics.The contribution of our work is that we never treat the reconstructed scene and the images separately as approaches to record those heritages,on the contrary,we combine them together to take the great advantage of the 3D information provided by the reconstructed point clouds and the textured 3D models and the more accurate color and texture information provided by the images simultaneously.