Parallel implementation of an integrated edge-preserving smoothing algorithm in clusters of workstations

An iterative, adaptive image smoothing algoritluu may he timexcmsumhg for a large amount of remotely sensed images. Because of the high computational cost of this type of algorithm, it is essential for the algorithm to be implemented and evaluated in a parallel processing framework. The Message Passing Interface @IPI) running on an Ethernet connected network of workstations was tested on the image smoothing filter. Depending on the size of the image, MPI can reduce the computation~riented algorithms significantly by properly selecting a number of processors. Testing results on different numbers of processors for the image smoothing filter were shown in this study.