A new impulse detector based on order statistics

The acquisition or transmission of digital images through sensors or communication channels is often interfered by impulse noise. It is very important to eliminate noise in the images before subsequent processing, such as image segmentation, object recognition, and edge detection. In this letter, a novel impulse noise-detection algorithm is presented, which can remove impulse noise from corrupted images successfully, and at the same time, without eliminating image details. The algorithm is based on the order statistics within a local window. Although our method is low in complexity when compared with some other complicated algorithms, experimental results show that it produces better restored images than many other existing techniques.