A Recursive Digital Filter Implementation for Noisy and Blurred Images

A part of low level image processing consists of the detection (by filtering) of sharp level variations of luminance in the image to obtain the edge segments, extremities and angles. To select the geometrical information related to the structure of the objects present in an image, a key step consists of extracting the edges of the considered objects. Several accurate numerical algorithms have been proposed in this field: Roberts [1], Prewitt [2], Canny [3], and Deriche [4] have already been implemented in software or with dedicated hardwares for several applications [1,5,6].