Temporal Surface Reconstruction

Abstract

This thesis investigates the problem of estimating the three- dimensional structure of a scene from a sequence of images. Structure information can be recovered from images through a number of visual mechanisms such as shading, motion and stereo. Image information is commonly available in a time-continuous fashion and this work proposes a method for estimating structure information in a temporally continuous manner for a variety of visual mechanisms. Structural information about a scene is represented in a dense depth map in which the distance to the scene is stored for each pixel location in the image. In addition, uncertainty about the structure values is represented explicitly by the estimate covariance. This representation is maintained over time by a stochastic recursive estimator, the Kalman filter. The estimator consists of two stages which are repeated for each new arriving image. The update stage improves the current depth estimate by incorporating the latest image measurement. It depends on the particular visual mechanism being employed and amounts to an iterative relaxation algorithm similar to conventional single- frame algorithms. The prediction stage transforms the current depth estimate into the next time-step to account for changes in the depth values that may occur if the camera moves relative to the (rigid) scene during the acquisition of the sequence.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
May 03, 1991
Accession Number
ADA259494

Entities

People

  • Joachim Heel

Organizations

  • Massachusetts Institute of Technology

Tags

Communities of Interest

  • Air Platforms
  • Cyber
  • Energy and Power Technologies
  • Ground and Sea Platforms
  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Algorithms
  • Artificial Intelligence
  • Computational Complexity
  • Computational Science
  • Computer Science
  • Computer Vision
  • Coordinate Systems
  • Estimators
  • Filters
  • Geometry
  • Kalman Filtering
  • Kalman Filters
  • Mathematical Filters
  • Optimal Estimators
  • Probability Distributions
  • Random Variables
  • Three Dimensional

Readers

  • Adaptive Control and Estimation with Uncertainty in Dynamic Systems.
  • Computational Modeling and Simulation
  • Computer Vision.