A new method of image noises removal

The paper proposes a new method to remove Gauss noises in an image while preserving the image details well, Based on the four-order cumulant, a 5 times 5 window is selected around a pixel in the image, and the window is divided into four sub-windows without superposition. The four-order cumulants of the four sub-windows are calculated individually. The minimum of the cumulants corresponds to a sub-window. The grey value of processed pixel is replaced by the average grey value of the four pixels in the sub-window. The method can remove or decrease the Gauss noises in an image well and preserve the details in the image.