The recognition of underwater laser target is a new disquisitive field, in which many problems are open, and extraction is the key problem. To recognize underwater target, extraction must be performed well because there is a lot of speckle noise in underwater laser images. Target segmentation of underwater laser image based on statistic counteraction (SC) principle i.e. noise with identically statistic characteristics can counteract noise in images is proposed to remove speckle noise and extract the target in images with wavelet transform and statistics. Experiments demonstrate that this method is efficient and feasible.