Reliable estimation of dense optical flow fields with large displacements

[EN] In this paper we show that a classic optical flow technique by Nagel and Enkelmann can be regarded as an early anisotropic diffusion method with a diffusion tensor. We introduce three improvements into the model formulation that avoid inconsistencies caused by centering the brightness term and the smoothness term in different images use a linear scale-space focusing strategy from coarse to fine scales for avoiding convergence to physically irrelevant local minima, and create an energy functional that is invariant under linear brightness changes.  Applying a gradient descent method to the resulting energy functional leads to a system of diffusion-reaction equations. We prove that this system has a unique solution under realistic assumptions on the initial data, and we present an efficient linear implicit numerical scheme in detail. Our method creates flow fields with 100% density over the entire image domain, it is robust under a large range of parameter variations, and it can recover displacement fields that are far beyond the typical one-pixel limits which are characteristic for many differential methods for determining optical flow. We show that it performs better than the classic optical flow methods with 100%  density that are evaluated by Barron et al. (1994). Our software is available from the Internet.

3-D geometry reconstruction using a color image stereo pair and partial differential equations

[EN] In this work, we present a new model for a dense disparity estimation and the 3-D geometry reconstruction using a color image stereo pair. First, we present a brief introduction to the 3-D Geometry of a camera system. Next, we propose a new model for the disparity estimation based on an energy functional. We look for the local minima of the energy using the associate Euler-Langrage partial differential equations. This model is a generalization to color image of the model developed in, with some changes in the strategy to avoid the irrelevant local minima. We present some numerical experiences of 3-D reconstruction, using this method some real stereo pairs.

Sobre un problema de optimización no-lineal en visión estereoscópica

[EN] Presentamos un método no lineal para la estimación de la geometría 3-D de una escena a partir de imágenes esteroscópicas. El problema principal consiste en calcular la posición relativa de las 2 cámaras a partir de un número de puntos que se corresponden en ambas cámaras. La posición relativa de las 2 cámaras viene dada por un vector de 7 parámetros : -X=(s,l,m,n,tx,ty,tz)-. Para calcular estos parámetros hay que minimizar una energía no-lineal del tipo E(x)=kAqxj donde A es una matriz 9x9 y q(X) es un vector función de X. En este trabajo presentamos un algoritmo para la busqueda de mínimos locales de E(X) basado en una modifcación del método de gradiente de paso óptimo. Presentamos algunas experiencias comparativas con otros métodos clásicos.