ABSTRACT
This chapter reviews the related work, and the problem definition of massive remote sensing image processing. It presents the design and implementations of our proposal of generic parallel skeletons for massive Remote Sensing (RS) algorithms on message passing interface (MPI)-enabled multi-core clusters. The chapter discusses the experimental analysis of program performance. It describes RS_GPPS is proposed to provide both a template for distributed RS data and generic parallel skeletons for RS algorithms. The RS image-wrapper skeleton is designed for the algorithms including pixel-based processing algorithms, neighborhood-based processing and band-based algorithms, the computation of which will causes, not cause the geometrical warping of image data. Incorporation of multi-core cluster based high-performance computing models in remote sensing image processing is an effective solution to address these computational challenges. The chapter examines the main issues related to parallel programming for massive remote sensing image processing applications on an MPI-enabled multi-core cluster.
