To extract shoreline from video monitoring images, a method is proposed by applying Radon transform and gradient tracing algorithm. Based on the different color properties of land and water image, the gradient vector is calculated to present edges in color space efficiently. Then, with Radon transform and its basic ability of linear singularity detection, a profile analysis technique is used to further select the corresponding peak area. Meanwhile, the approximate direction and area of waterline is obtained in the Radon domain. In addition, the image region including interface of shore and sea boundaries is located by reconstructing image using inverse Radon transform. Finally, the shoreline is determined by an improved searching strategy to explore the maximum gradient path. Experimental results indicate that the proposed approach is effective in finding the shorelines and the algorithm has robustness with varies lightening conditions.