Title :
The Study of Parallel Ortho-rectification Method of Line-Array Image Based on GPU
Author :
Yuxia Yang ; Zhaohua Liu ; Jingyu Yang
Author_Institution :
Sch. of Archit. & Surveying &Mapping Eng., Jiangxi Univ. of Sci. & Technol., Ganzhou, China
Abstract :
This paper first briefly introduces the principle of Ortho-Rectification of line-array image, then designed a parallel processing method based on GPU and proposes a shared memory optimizing strategy of POS data to avoid performance bottle-neck due frequently accessing data in global memory, at last do a system experiment using ADS40 image based on Tesla C2050 GPU and invalidate the parallel processing advantage of GPU.
Keywords :
graphics processing units; image processing; parallel processing; shared memory systems; ADS40 image; POS data; Tesla C2050 GPU; global memory; line-array image; parallel ortho-rectification method; parallel processing method; shared memory optimizing strategy; Arrays; Graphics processing units; Instruction sets; Kernel; Mathematical model; Memory management; Parallel processing; GPU; Image; Line-Array; Ortho-Rectification; Parallel Processing;
Conference_Titel :
Computer Sciences and Applications (CSA), 2013 International Conference on
Conference_Location :
Wuhan
DOI :
10.1109/CSA.2013.148
