M. Sainz and N. Bagherzadeh (USA), and A. Susin (Spain)
3D reconstruction, Structure from Motion, Singular Value Decomposition, Voxel Carving.
In this paper we present an automatic method for the recon struction of a 3D volumetric representation of real world scenes from a set of multiple uncalibrated images. The process is divided in two steps (1) an automatic calibra tion of the cameras and (2) a scene reconstruction consis tent with the input views. The calibration of the cameras is performed using automatically tracked 2D features, and consists in calculating a projective approximation and up grading it to an Euclidean structure by computing the pro jective distortion matrix in a way that is analogous to esti mate the absolute quadric. Moreover, in contrast to other approaches our process is essentially a linear one. The un derlying technique is based on the Singular Value Decom position (SVD) and the process is enhanced with a careful study of the rank of the matrices involved in order to get the excellent results shown in the paper. The volumetric re construction of the scene is performed using an improved voxel carving algorithm. The result is a voxel-based model of the external surface of the physical objects present in the scene. Optimized data structures and graphics hardware acceleration are used to achieve a substantial reduction in computation time. Furthermore, the spatial information ob tained from the the camera calibration process about the 2D tracked measurements is used to automatically set the in ternal thresholds of the carving algorithm, achieving a full automation of the method.
Important Links:
Go Back
