Title :
A New Type of NDVI Algorithm Based on GPU Dividing Block Technology
Author :
Wang Juan ; Sun Jianchao
Author_Institution :
Comput. & Inf. Eng. Coll., Henan Univ., Kaifeng, China
Abstract :
According to the matrix computation of the traditional NDVI algorithm calculate slowly on CPU, the paper puts forward a new type of NDVI algorithm based on GPU. And we use the CUDA programming language to realize it based on NVIDIA GPU. Experimental results show that the new type of NDVI algorithm significantly improved the execution speed of the algorithm. Compared with the serial CPU program, the maximum acceleration ratio is up to 8X in our experiment.
Keywords :
geophysical image processing; graphics processing units; matrix algebra; parallel architectures; programming languages; vegetation mapping; CUDA programming language; GPU dividing block technology; NDVI algorithm; NVIDIA GPU; algorithm execution speed; compute unified device architecture; graphics processing units; matrix computation; maximum acceleration ratio; normalized difference vegetation index; serial CPU program; Acceleration; Graphics processing units; Instruction sets; Kernel; Random access memory; Remote sensing; Vegetation mapping; GPU; Matrix; NDVI;
Conference_Titel :
Computational and Information Sciences (ICCIS), 2013 Fifth International Conference on
Conference_Location :
Shiyang
DOI :
10.1109/ICCIS.2013.192
