A random-valued impulse noise detector by using level detection <Abstract> In this paper, we propose a new random-valued impulse noise detecter from images using some shape windows. In our system, an window that is selected from multi-windows includes flat region. Thus, selected window has low variation in pixel values, and random-valued impulse noise moves to the both ends of order statistics. Therefore noise detection in selected window is easy. Consequently proposed method reduces undetected noise pixel without increasing mis-detection. Extensive simulations indicate that proposed method performs significantly better than traditional methods. |