Contribution to the implementation of computer vision application on a GPU

Largely driven by cinema and video games companies, research and development of hardware tools for the generation of synthetic images such as graphics card (or GPU, Graphics Processing Units), they experienced a tremendous expansion. The characteristic of this card allows handled all tasks that require parallel computation. This opens the doors to other application areas. GPUs have become more valued when they made the principle of GPGPU (General Purpose computation on GPU or generic programming on GPU) which was developed in the laboratories of the manufacturer NVIDIA. It has made a revolution, bringing the world of research to interest in the parallel programming on GPUs. When it introduced the unified treatment notion of processors, which is at CUDA. Motivated by the computational needs considerable related to computer vision area, we propose in this paper to apply a GPGPU concept at a specific application of this domain by using a GPU graphics card.