Informace o publikaci

GPU Optimization of Convolution for Large 3-D Real Images

Autoři

KARAS Pavel SVOBODA David ZEMČÍK Pavel

Rok publikování 2012
Druh Článek ve sborníku
Konference Proceedings of the International Conference on Advanced Concepts for Intelligent Vision Systems (ACIVS’12)
Fakulta / Pracoviště MU

Fakulta informatiky

Citace
www http://dx.doi.org/10.1007/978-3-642-33140-4_6
Doi http://dx.doi.org/10.1007/978-3-642-33140-4_6
Obor Informatika
Klíčová slova gpu; convolution; 3-D; image processing
Popis In this paper, we propose a method for computing convolution of large 3-D images with respect to real signals. The convolution is performed in a frequency domain using a convolution theorem. Due to properties of real signals, the algorithm can be optimized so that both time and the memory consumption are halved when compared to complex signals of the same size. Convolution is decomposed in a frequency domain using the decimation in frequency (DIF) algorithm. The algorithm is accelerated on a graphics hardware by means of the CUDA parallel computing model, achieving up to 10x speedup with a single GPU over an optimized implementation on a quad-core CPU.
Související projekty:

Používáte starou verzi internetového prohlížeče. Doporučujeme aktualizovat Váš prohlížeč na nejnovější verzi.

Další info