Approach proposed in this paper combines wavelet transform with morphology Top-Hat filter to suppress the mixed noise and the large-scale clutter caused by background in the infrared images. This process can decrease the blur of the images and enhance the targets. Then, proper structuring elements are selected to execute serial morphology operations to discard most false alarms, through which a few potential small targets can be obtained. Eventually, the real target can be segmented through searching maximal grayscale and assigning a threshold. The experimental results show that the method can effectively detect and segment the target in a low SNR infrared image with the complicated nature background.