Parallel Laplacian filter using CUDA on GP-GPU
Mishal Almazrooie, Rosni Abdullah, Lim Yun Yi, Ibrahim Venkat, Zahraa Adnan · 2014
Parallel programming has been extensively applied to different fields, such as medicine, security, and image processing. This paper focuses on parallelizing the Laplacian filter, an edge detection algorithm, using CUDA. We have conducted a performance analysis to benchmark the sequential Laplacian version against the CUDA parallel approach. Results show that the parallelization of Laplacian filter achieves better performance than sequential code. CUDA approach achieved 200 fold speedup for an image size of 50MB when 216 processing elements were deployed.