Abstract: An edge-detection algorithm based on visual saliency is proposed. First, for an infrared-scene image, the proposed algorithm detects the edges using a Canny edge detector. Then, three basic edge features are integrated, including the edge contrast, edge density, and edge length. Finally, the visual salient edges are extracted from the infrared image. The proposed algorithm effectively suppresses false edges from the background and texture, highlights the information of the target edges, and provides good-quality edge images for automatic target recognition system. The experimental results verify the effectiveness of the algorithm.