The　edge　detecting　ability　of　anti-symmetrical　biorthogonal　wavelets　is　theoretically　analyzed.　Then　a　multiscale　edge　detection　algorithm,　which　integrates　edge　information　of　different　resolution　levels,　is　proposed　to　obtain　the　highly　accurate　one-pixel-width　image　edges.　By　applying　the　algorithm　to　real-world　images,　the　effectiveness　of　our　method　is　tested,　which　provides　a　fundamental　basis　for　employing　edge　information　to　support　image　indexing　in　wavelet-based　compressed　domain.