A New Method for Acquisition of High-Resolution Seabed Topography by Matching Seabed Classification Images

The multibeam echo sounders (MBES) can acquire accurate positional but low-resolution seabed terrain and images, whereas side scan sonars (SSS) can only acquire inaccurate positional but high-resolution seabed images. In this study, a new method for superimposing corrected-positional SSS images on multibeam bathymetric terrain is proposed to obtain high-resolution and accurate-positional seabed topography using traditional MBES and SSS. Three steps, including the normalization by the z-score, sediment classification by the k-means++ algorithm, and denoising processing using morphological operations, are processed for both MBES and SSS images to obtain the corresponding sediment images. Next, a segmented matching method is given based on the common sediment distributions and features of MBES and SSS sediment images. The two kinds of sediment images are matched segmentally using the speeded up robust features algorithm and random sample consensus algorithm. Then, the positions of SSS images are corrected segmentally using thin plate splines based on matching points. Finally, the corrected SSS image is superimposed on MBES bathymetric terrain, based on positional relationship. The proposed method was verified through experiments, and high image resolution and high position accuracy seabed topography were obtained. Moreover, the performances of the method are discussed, and some conclusions are drawn according to the experiments and discussions.