THRESHOLD IMPULSE DETECTOR BASED ON LUM SMOOTHER (LUMsm DETECTOR)

Rastislav Lukáč - Stanislav Marchevský

   In this paper we present a new impulse detector based on a LUM smoother (LUMsm detector) for detection of an impulsive type of noise. This detector is controlled by a threshold. Our detection is based on a LUM smoother filter, which is a subclass of a rank-order-based, lower-upper-middle (LUM) filter. A subclass of LUM smoother is used in image-noise smoothing applications. The output of this filter is determined by comparing a lower- and upper-order statistics to the middle of the sample in the observation window. We demonstrate that the proposed algorithm of the LUMsm detector executes the detection operation very well and the performance of impulsive noise removing along with a 3x3 median filter shows excellent capabilities of image-detail preservation. We yield comparisons to other impulsive noise detectors demonstrated by the mean-absolute error (MAE) and mean-square error (MSE). Description of the LUM smoother and impulse detectors is included.

Keywords: LUM filters, impulsive type noise, impulse detector

[full-paper]