Contourlet-Based Acoustic Seabed Ground Discrimination System

In this paper, the problem of automatic segmentation and classification of seafloor using acoustic ground discrimination systems is discussed and a new split and merge algorithm based on contourlet transform is presented. The contourlet transform is a new two-dimensional extension of the wavelet transform using multiscale and directional filter banks. It appears to be a suitable tool for this task, because it allows analysis of images at various levels of resolution as well as directions, which effectively capture smooth contours that are the dominant feature in seabed images. In the proposed approach, the input image of seabed is split into M times M blocks. Each block is classified separately using four levels of contourlet transform with 28 directional bandpass sub-bands. First-order and second-order statistics of the high frequency details from all sub-bands of the contourlet coefficients are extracted as features and a weighted distance classifier is used for block classification. Moreover, the effectiveness of other contourlet domain features such as sub-band energy in discriminating different seabed textures is presented. Finally, the classified blocks on the boundaries of the segmented image, are refined and merged to produce the final segmented image. The proposed method provides a fast tool with enough accuracy that can be implemented in a parallel structure for real-time processing. In addition, the simulation results are compared with the results of wavelet-based methods as well as other well-known techniques to show the effectiveness of our proposed algorithm.