| dc.description.abstract | In recent years, the volumetric scene representation has been widely used in the fields of computer graphics and computer vision. In computer graphics, the representation is used to visualize and reconstruct 3D internal structures of 2D human-body medical data obtained from Magnetic Resonance Imaging (MRI) or Computed Tomography (CT). On the other hand, the volumetric representation is used to model 3D objects from a sequence of 2D images in computer vision. In both cases, eliminating some unnecessary part of the volumetric data is sometimes needed. For example, we need to remove skin or tissue which hides the target organ of human in volume rendering, or empty region of 3D objects being modeled in image-based modeling. In this paper, we propose a fast carving algorithm which can be applied in volume rendering and image-based modeling. The memory usage is minimized by using 1-bit field of a voxel to represent whether the voxel is carved or not. Our experiments show that we can carve out any part of the 5123 volume data in about a second. The carving speed is not affected by the complexity of objects. | en_US |
