Decomposition method of raster geographic data based on parallel computing

Abstract

The paper mainly studied decomposition method of raster geographic data based on parallel computing. Firstly, we structured computational transformation model of raster geographic data; Then, we designed a computational experiment to validate the computational transformation model and evaluate the performance of k-NN classification algorithm. Results of parallel computational experiment show that the model can be applied to decompose a heterogeneous spatial computational domain representation into a balanced set of computing tasks; the speedup performance of parallelizing k-NN classification algorithm based on the transformation model is superior to the results from traditional method.