Aiming at the detection of IR point target,correction of IR image fuzziness are carried out by inhibiting nonlinear influence in temperature field with the same row average subtracting.Inhibiting background and enhancement of point target are implemented by using Top-Hat operator morphological;segmenting threshold is obtained by comparing characteristics of filtered IR images histograms under different background.In the end,a point target segmentation image contained few false alarm points is obtained after binary segmentation.The experiments show the method can detect IR point target under different background with low signal-to-noise ratio.