Adaptive threshold based impulse detection for restoration of digital images
Umesh Ghanekar, Rajoo Pandey · 2016
The choice of appropriate value of threshold in median-based impulse detection method becomes difficult due to its dependence on the noise density and image characteristics. In the case of random valued noise(RVIN), if a fixed value of threshold is used then it will result into large percentage of missed and false detection. Therefore, a variable threshold governed by the local image characteristics, is required for detection of RVIN. Here, we present an impulse detector in which the value of threshold depends on window under observation. The extensive simulations exhibit the efficacy of the method in respect of both random valued as well as salt and pepper noise.