The Design Of Parallel Algorithm Of Remote Sensing Images Based On JPEG2000

With the continuous advancement of science and technology, remote sensingimages have played an increasingly important role in people’s daily lives, such asgeological exploration, urban planning, construction, military reconnaissance,disasterforecasting, environmental monitoring, remote sensing images can provide a wealth ofuseful information for researchers. According to relevant statistics, remote sensingimage plays a huge role in nearly30areas. With the increasing amount of remotesensing image data, how to decompress the remote sensing images fast and accuratebecomes more important. Taking into account the equipment’s processing speed andthe stability, with the project as an example,blade server is used to build a hardwareplatform, take the use of MPI+OpenMP to build good computing environment,whichare practical and with high performance for parallel computing, In thisenvironment,experiment of remote sensing image transmission and decompressionwere made.At first, the basic theory of parallel computing is explained,and the differencesbetween the MPI and OpenMP were compared. Considering the features and theporpose of this article, carried out a feasibility analysis of MPI+OpenMP hybridparallel mode.Then, it summarized the basic conceptions of JPEG2000imagecompress processing,after a detailed analysis of the most time-consuming part, whichare wavelet transform and EBCOT coding, the basic idea and methods of theparallelization of the decompression have been proposed. Ultimately, the concreteimplementation of the MPI+OpenMP hybrid parallel algorithm is given. The basicidea of the hybrid parallel consists of two parts,MPI the parallelization between thesub-compute nodes, but the parallelization in sub-compute code itself is actualized byOpenMP. At last, the influences of the data transmission rate to decompression isstudied, the effect of serial program is analyzed and the performance of serialalgorithm is compared to MPI and MPI+OpenMP two parallel algorithm mode.The results show that the transmission delay is too small to be ignored. With thestability and the fast processing speed, this system can greatly shorten thedecompression time. Compared to the serial program, the two parallel model’sprocessing speed has greatly improved, In particular, the MPI+OpenMP hybridparallel model is better than the MPI parallel model. This device has some reference and practical value for real-time processing of remote sensing data.