A new unsupervised multiresolution edge detection technique was presented,which combines the morphological filtering,pyramid data structure and fuzzy technique.The algorithm first constructed a pyramid by morphological filtering,then it considered the parent child linking relationship of edge nodes between the two adjacent levels as fuzzy model.Via this model,the coarse and low resolution edge map was propagated and refined in the pyramid,thus the map with fine and high resolution edge was obtained.Finally,some experiments on synthetic and real world images were given with satisfactory results.