E. Bayro-Corrochano and E. Hancock (Eds.): CIARP 2014, LNCS 8827, pp. 1030–1038, 2014.
© Springer International Publishing Switzerland 2014
An Efficient GPU-Based Implementation
of the R-MSF-Algorithm for Remote Sensing Imagery
David Castro-Palazuelos
1,2,*
, Daniel Robles-Valdez
1
, and Deni Torres-Roman
1
1
Center for Advanced Research and Education of the National Polytechnic Institute of Mexico
CINVESTAV Unidad Guadalajara, Mexico
Avenida del Bosque #1145, Colonia el Bajío, 45019, Zapopan, Jalisco, México
2
Culiacan Technological Institute, Department of Electrical-Electronic Engineering,
Sinaloa, Mexico
dcastro@gdl.cinvestav.mx
Abstract. This paper presents an efficient real time implementation of the regu-
larized matched spatial filter algorithm (R-MSF-Algorithm) for remote sensing
(RS) imagery that employs the robust descriptive experiment design (DED) ap-
proach, using a graphics processing unit (GPU) as parallel architecture. The
achieved performance is significantly greater than initial requirement of two
image per second. The performance results are reported in terms of metrics as:
number of operations, memory requirements, execution time, and speedup,
which show the achieved improvements by the parallel version in comparison
with the sequential version of the algorithm.
Keywords: Remote sensing, GPU, real time implementation.
1 Introduction
Currently, sensor array signal processing (SP) for imaging radars have been focus of
great interest in many research works that now are available in [1, 2]. Such algorithms
are computationally expensive; consequently, the majority of the sequential imple-
mentations are not suitable to achieve real time or near real time. As many of the re-
quired operations needed by these algorithms are: correlation, convolution, filtering,
and matrix operations, parallelization technique and theory can be applied in order to
improve their performances [3]. The implementation of real time systems in the field
of high performance computing (HPC) is limited by the data dependencies of algo-
rithms to be implemented and the features of graphic processing unit (GPU) used [4].
Hence, the advent of the GPU with user-friendly programming environments has
allowed the employment of a GPU as parallel mathematical co-processor. In this doc-
ument, the proposed DED framework based on the matched spatial filter (MSF) SP
technique [2] are implemented (referred here as R-MSF-Algorithm). The R-MSF-
Algorithm is composed of three parts: the average, the array correlation function, and
the matched spatial filter (MSF). In addition, R-MSF-Algorithm is aimed to form a
*
Corresponding author.