METHOD FOR GENERATING 3D SURFACE RECONSTRUCTION MODEL USING MULTIPLE GPUS AND APPARATUS OF ENABLING THE METHOD

摘要

PURPOSE: A method for generating a 3D surface reconstruction model using multiple GPUs(Graphics Processing Unit) and an apparatus for the same are provided to determine whether an object region is included based on a central point, not an apex of a voxel cube, and to efficiently compensate peripheral information, which can be lost, in a mesh generation step, thereby reducing time required for calculation. CONSTITUTION: An input data generator(110) divides an image data, which is inputted through at least one or more cameras, into a foreground image and a background image. A real-time model generator(120) generates a 3D volume model using the foreground and background images through multiple GPUs, and generates a 3D mesh model by processing the 3D volume model. A real-time rendering unit(130) generates an animation image through texture mapping and composing to the 3D mesh model. [Reference numerals] (110) Input data generator; (120) Real-time model generator; (130) Real-time rendering unit; (AA) Input; (BB) Output