Image signal-to-noise ratio (SNR) will be reduced with the range increasing because of the influence of backscattering noise when blue-green laser radar is detecting and imaging underwater target. In order to solve this problem, by analyzing the noise types for the underwater target image of the laser radar, an algorithm of image pre-processing is presented, which includes the noise suppression in frequency domain and the improved Butterworth high-pass filters. The pre-processing algorithm can realize the image restoration, and improve the image SNR. The experiments of detecting and imaging the mine in indoor pool are carried out with the streak tube imaging lidar. The algorithm is used for the intensity image of the 5 m and 15 m underwater mine. The results show that the pre-processing algorithm can effectively suppress the backscattering noise of water. The contrast of image is improved. The clear object contour is obtained.