Traitements d'images sur architectures parallèles et hétérogènes : Traitement hétérogène d'images

Image processing algorithms present a necessary tool for various domains related to computer vision. These algorithms are hampered by their high consumption of computing times when processing large sets of high resolution images. In this work, we propose a development scheme enabling an efficient exploitation of parallel (GPU) and heterogeneous (Multi- CPU/Multi-GPU) platforms, in order to improve performance of image processing algorithms. The proposed scheme enables an efficient scheduling of hybrid tasks and an effective management of heterogeneous memories. We present also parallel and hybrid implementations of edge and corner detection methods. Experimental results showed a global speedup ranging from 5 to 25, when processing different sets of images, by comparison with CPU implementations.