Parallel fractal image compression using quadtree partition with task and dynamic parallelism

Fractal image compression is a lossy compression technique based on the iterative function system, which can be used to reduce the storage space and increase the speed of data transmission. The main disadvantage of fractal image compression is the high computational cost of the encoding step, compared with the popular image compression based on discrete cosine transform. The aim of this paper is the development of parallel implementations of fractal image compression using quadtree partition. We develop two parallel implementations: the first one uses task parallelism over a multi-core system and the second uses dynamic parallelism over a GPU architecture. We show performance comparisons of the parallel implementations using standard images to compare the capabilities of these parallel architectures. The proposed parallel implementations achieve speedups over the serial implementation of approximately 15x using the multi-core CPU and 25x using the GPU.