A Study on the Acceleration of a Speckle Reduction Algorithm using GPUs
Geon Lee, Myung‐Jun Lee, Nam-Hoon Jeong, Jae-Ho Choi, Kyung‐Tae Kim · The Journal of Korean Institute of Electromagnetic Engineering and Science · 2021
When a synthetic aperture radar (SAR) image is created, if there are some scatterers in the resolution cell, the reflected signals from the scatterers interfere with each other, causing irregular changes in the intensity of pixels in the image called speckles. Speckles increase the false alarm rate and decrease the pixel detection rate. Therefore, speckle reduction filters have been developed to reduce the influence of speckles. However, filters that reduce large amounts of speckles have a long processing time; therefore, they are costly to use in real-time applications, in which quick processing is crucial. This study used a graphics processing unit (GPU) to increase the processing speed of speckle reduction filters and optimized the existing speckle reduction algorithm to run in parallel with a GPU. Furthermore, we compared the processing time and data capacity of existing and optimized algorithms.