Adaptive Mean Filter Technique for Removal of High Density Salt and Pepper (Impulse) Noise

Abstract

In this paper, the adaptive mean filter technique for removal of high density salt and pepper also termed as impulse noise is presented. It is desired to remove high density impulse noise from the images due to hot pixels produced or generated because of current leakage in the image sensors placed inside the camera. Adaptive mean filter technique processes the images affected with noise using variable filter size that results in better removal of high density impulse noise as compared with fixed filter size. Variable filter size comparatively takes more processing time but results in better evaluation parameters such as signal to noise ratio (SNR), mean absolute error (MBE) and image enhancement factor (IEF). Various experiments were performed on images having different entropy to measure the performance of the presented filter technique. The adaptive mean filter with variable size filters effectively removes high density impulse noise up to density 0.5. The presented adaptive mean filter technique requires simple arithmetic operations that achieve faster processing with less computational overheads. The presented adaptive mean filter technique for removal of high density salt and pepper or impulse noise can be adapted in image acquisition systems.