As an optical remote sensing system is not perfect, noise can enter the data collection system at several points. Necessary corrections include the removal of shot noise (random bad pixels), correcting line or column drop-outs, accounting for line-start problems and radiometric correction of n-line striping caused by detector miscalibration. SAR data have global, random speckle noise. Speckle filters are designed to adapt to local image variations in order to smooth values, thus reducing speckle and enhancing lines and edges to maintain the sharpness of an image. A widely used way to reduce speckle is to apply spatial filters to the images. Typical approaches for speckle filtering include Laplace filtering for smoothing and sigma filters that preserve more of the signal with a lesser effect of smoothing.
Identify different types of noise and associated methods for their reduction
New