Spectral Residual Based Underwater Animal Detection

In this paper, the spectral residual method is applied in the underwater image data for detecting the animals. The system is designed to assist the underwater monitor system survey operations, specialized to the task of animal detection. Firstly, the regularity for the frequency spectrum of the images collected in the underwater world is discovered by the statistical analysis. Then we transform the input image into the spatial frequency domain and singularities including in the frequency curve is extracted by average filtering. Finally, these singularities are inverse transformed from the frequency domain into spatial domain and the saliency area is detected. Experimental results, which have been performed on a set of real underwater images acquired in different environments, demonstrate the robustness and the accuracy of the proposed system in the task of underwater animal detection.