A novel algorithm to segment images under the echo broadening effect (EBE) in range-gated laser night vision technology is presented. The proposed algorithm uses average gray value of outline edges information to select the threshold extracting signal areas out. Spatial information of each pixel is applied to detect the noises from the candidate signal areas after the determination of the threshold. This algorithm avoids too much morphological operation and prevents the distortion of the images. The experimental results show that, using this method, it is very effective to segment low quality long distance range-gated images.