Visual Saliency Detection for Water Surface Contaminants

Visual saliency detection is an intuitive and effective means for water environment monitoring. Conventional saliency detection algorithms usually have high computational complexity, and the detection results are usually unsatisfactory for the complex and diversified water scene images. In this paper, we propose a novel method which is simple and effective for the detection of water surface contaminants. In our method, the input image is reconstructed according to block-based compressed sensing for reducing the computational complexity. Then, a local contrast principle is used to obtain the primary saliency map. Based on this, a linear combination of color coefficients is designed to construct final saliency map. The experimental results show that the proposed method has good detection performance in terms of accuracy and running time.