Integrating quality in fuzzy reasoning edge detection
Vincent Bombardier, Oliver Perez-Oramas, J. Brémont · 2002
We describe an edge detection operator based on fuzzy linguistic rules. The aim of the work is to introduce "high level information" in low level image processing such as edge detection in order to adapt image processing to image context conditions so as to improve the detection. First, we present the fuzzy reasoning edge detection operator and secondly, we explain the two stages where we integrate information about image quality. We consider two ways of obtaining image quality either by expert assessment or by histogram analysis. The image quality information is used for choosing the most adapted homogeneity extraction function and modifying the membership functions of the operator.