Noise and background clutter are inhibited by frequency band selection carried out by multi-resolution analysis based on orthogonal wavelet decomposition. Sea-sky division line and target potential area are determined through detecting horizontal and vertical edges of an image. Large number of background disturbance is removed by using edge fusion from the different directions and candidate target points are obtained. Gray of the candidate target points are detected, threshold is determined, the false alarm points are eliminated and the target segmentation is carried out. The experimental results show that this method can detect small IR target such as warship in a complex sea background.