Open Access
Parallel and deterministic algorithms from MRFs (Markov Random Fields): Surface reconstruction and integration. Memorandum report
Davi Geiger,Federico Girosi +1 more
TLDR
In this paper, the authors derived deterministic approximations to Markov Random Fields (MRFs) models, which can be applied in the output of the visual processes to reconstruct surfaces from sparse and noisy depth data, or to integrate early vision processes to label physical discontinuities.Abstract:
In recent years many researchers have investigated the use of Markov random fields (MRFs) for computer vision. They can be applied for example in the output of the visual processes to reconstruct surfaces from sparse and noisy depth data, or to integrate early vision processes to label physical discontinuities. Drawbacks of MRFs models have been the computational complexity of the implementation and the difficulty in estimating the parameters of the model. This paper derives deterministic approximations to MRFs models. One of the considered models is shown to give in a natural way the graduate non convexity (GNC) algorithm. This model can be applied to smooth a field preserving its discontinuities. A new model is then proposed: it allows the gradient of the field to be enhanced at the discontinuities and smoothed elsewhere. All the theoretical results are obtained in the framework of the mean field theory, that is a well known statistical mechanics technique. A fast, parallel, and iterative algorithm to solve the deterministic equations of the two models is presented, together with experiments on synthetic and real images. The algorithm is applied to the problem of surface reconstruction is in the case of sparse data. A fast algorithm ismore » also described that solves the problem of aligning the discontinuities of different visual models with intensity edges via integration.« lessread more
Citations
More filters
Book
Markov Random Field Modeling in Computer Vision
TL;DR: This book presents a comprehensive study on the use of MRFs for solving computer vision problems, and covers the following parts essential to the subject: introduction to fundamental theories, formulations of MRF vision models, MRF parameter estimation, and optimization algorithms.
BookDOI
Geometry-Driven Diffusion in Computer Vision
TL;DR: This paper presents a meta-analyses of differential Invariant Signatures and Flows in Computer Vision: a Symmetry Group approach P. Sapiro, A. Tannenbaum, and a Differential Geometric Approach to Anisotropic Diffusion.
Journal ArticleDOI
Motion detail preserving optical flow estimation
TL;DR: In this article, a novel optical flow estimation method is proposed, which reduces the reliance of the flow estimates on their initial values propagated from the coarser level and enables recovering many motion details in each scale.
Journal ArticleDOI
Generalized Deformable Models, Statistical Physics, and Matching Problems
TL;DR: Techniques from statistical physics are used to exploit the power of statistical techniques to put global constraints on the set of allowable states of the binary matching elements and be preferable to existing methods of imposing such constraints by adding bias terms in the energy functions.
Journal ArticleDOI
Analog hardware for detecting discontinuities in early vision
TL;DR: This work discusses the first hardware circuit that explicitly implements either analog or binary line processes in a deterministic fashion, and successfully designed, tested, and demonstrated an analog CMOS VLSI circuit that contains a 1D resistive network of fuses implementing piecewise smooth surface interpolation.