Heuristics for selecting gray scale morphological structuring elements
Paul Fetter · RIT Scholar Works (Rochester Institute of Technology) · 1994
This thesis explores some heuristics for choosing 8 bit gray scale morphological structuring elements for reducing noise. The variables of size, shape and volume that enter into the choice of structuring elements create a very large number of possible structuring elements. Some general heuristics to guide the choice of an appropriate structuring element will make the task easier. Both the absolute error of the image and the appearance of the image will be used to judge the results. The experiments were performed on 3 images. Each of the images had noise added before processing; one set of data had 10 percent of the pixels disturbed by noise, the other had 20 percent of the pixels disturbed by noise. The resulting 6 images were then filtered with 10 different structuring elements and the resulting images were then compared against the respective baseline image. The conclusions were guided by the resulting absolute error values.