In order to handle this difficulty, we thus classify algae images into two groups: a rod shape and a
non-rod shape groups. Algae in Anabaena, Oscillatoria, Closterium, Pleurotaenium and Euglena are
classified into the rod shape group, while the rest are classified into the non-rod shape group. Algaeimages in the rod shape group require high degree of image smoothing so that it can extract the contour
as close as possible to the true algae boundary. The non-rod shape group requires less degree of image
smoothing in order to preserve their spines or flagellums. Thus, in a segmentation process, the two
groups of algae image are treated separately. In this work we propose a single-resolution edge detec-
tion for segmenting images in the non-rod shape group (because they require small amount of image
smoothing) and propose a multi-resolution edge detection to handle with images in the rod shape group
(because they require higher amount of image smoothing).