This paper introduces an architecture of parallel remote sensing image processing software, with advantages including high scalability, platform-independence, language-independence, and so on. It helps achieve high-performance computing in this field. MPI is used as the fundamental distributed message passing protocol. An object-oriented wrapper, Boost.MPI library, is used in the software to manipulate MPI. Open Source libraries such as GDAL and Open-CV are studied in this paper to help develop detailed image processing programs and implement classification algorithms. A number of experiments are conducted to test the parallel classification programs. The results indicate that in most cases the performance is significantly improved, especially for multi-spectral remote sensing image classification, in which a highest speed-up of 3.92 is reached. © 2012 Springer-Verlag.
CITATION STYLE
Wang, X., Li, Z., & Gao, S. (2012). Parallel remote sensing image processing: Taking image classification as an example. In Communications in Computer and Information Science (Vol. 316 CCIS, pp. 159–169). https://doi.org/10.1007/978-3-642-34289-9_19
Mendeley helps you to discover research relevant for your work.