Parallel Implementation of a Robust Optical Flow Technique


Line of Research



The accuracy and performance of current variational optical flow methods have considerably increased during the last years. The complexity of these techniques is high and enough care has to be taken for the implementation. The aim of this work is to present a comprehensible implementation of recent variational optical flow methods. We start with an energy model that relies on brightness and gradient constancy terms and a flow-based smoothness term. We minimize this energy model and derive an efficient implicit numerical scheme. In the experimental results, we evaluate the accuracy and performance of this implementation with synthetic images. We show that it is a competitive solution with respect to the most accurate methods. In order to increase the performance, we use a simple strategy to parallelize the execution on multi-core processors. In this page you can find the code for the implementation of a robust variational optical flow method. These are based on energy functionals that include the brightness and gradient constancy assumptions and a flow driven smoothness regularization. These methods replace the traditional quadratic penalty function by a continuous L1 norm, which makes them more robust to noise and outliers. We use the OpenMP library for the parallelization of the algorithms. It is a simple parallelization that does not modify the original code. The code is developed in standard C++ and it has been compiled in Windows and Linux, using the GNU gcc compiler.