Skip to Main content Skip to Navigation

Three-dimensional structure from a monocular sequence of images

Abstract : In this paper, we address the following problem: given a camera moving in an unknown environment, we want to obtain a 3D description of the environment. We present a unifying approach to the problem of the lUoving camera by deriving a unique formalism to pro cess uniformly different but complementary features. namely points and linear segments, in order to solve important dynamic vision problems such as tracking and structure from motion. We first present different concept for tracking features: (1) 2D tracker: 2D features (points and segments) are tracked using an order 1 dynamic model for their evolution; (2) 2D ..i.. esümation tracker: 3D fusion of 2D features is performed recursively, then the value predicted at time t for these 3D features is projerted at time t + 1 onto the camera focal plane and replaces the dynamic model used in the 2D tracker. This allows for introducing 3D informations into the 2D features tracker without prior kllowledge of the envirolllllellt. (3) .JD trader: the 2D tracker disappears totally, and aH computations are :3D. This tracker combines the silllplicity of the 2D tracker and the dficiency of the 2D-'-estimation tracker. We then describe the mechallisllls of fusion that integrate 2D measurements into an estimate of the feature 3D parameters. Uncertaillties are .taken into a.CCOUllt through extended Ka.lman filtering. F",ature parameterizatioll are chosen such as to reduce the algorithmic complexity. simplify the linearization process and ensure numerical stability.
Document type :
Complete list of metadata
Contributor : Rapport de Recherche Inria Connect in order to contact the contributor
Submitted on : Wednesday, May 24, 2006 - 5:52:04 PM
Last modification on : Friday, February 4, 2022 - 3:13:38 AM
Long-term archiving on: : Tuesday, April 12, 2011 - 10:16:08 PM


  • HAL Id : inria-00075277, version 1



Jean-Luc Jezouin, Nicholas Ayache. Three-dimensional structure from a monocular sequence of images. [Research Report] RR-1282, INRIA. 1990. ⟨inria-00075277⟩



Record views


Files downloads