A precise pose monitor is required for high precision target positioning in inertial-confinement-fusion (ICF) experiments. Conventional visual monitoring methods are sensitive to illumination and target pose and often need artificial judgment. This paper introduces seale invariant feature transform (SIFT) algorithm into ICF target monitoring domain and implements an on-line target monitoring method based on SIFT features matching. The experimental results of an on-line setting demonstrate that the translational and angle accuracy are better than 6 μm and 0.5′,respectively.