Segmantace obrazu pomocí difuzních metod na platformě CUDA

Abstract

This diploma thesis deals with image segmentation algorithm and its implementation for CUDA device. Image segmentation is operation during which we try to split digital image into areas(foreground, backgroudn), or into individual objects. There are many image segmentation algorithms for this task. I deal wiht algorithm called k-mean. I studied this algorithm and then made implementation for CUDA device, which are graphic cards enables paralel running of some algorithm parts.