The Performance Analysis Based on Heterogeneous Parallel Processors for Anisotropic Diffusion Filters

A noise in digital image degrades the performance of image processing. These images are most often used in medical field for diagnosis and treatment. Thus, there is a huge demand for high quality images from the medical field. The current algorithms to process useable images are derived using Gaussian blur filter. However, using such isotropic diffusion technique, the filter cuts away pertinent information within the image. The high noise level in an image will blur organs in medical images such as in MRI. In order to accommodate the loss of information while maintaining the quality of the image, an anisotropic diffusion filter has been developed. The anisotropic diffusion filter uses heat equation and the only downfall of the technique is lengthy process time. The purpose of this study is to decrease the processing time using parallel processors through the optimization of GPU.