In this paper, a new method for the estimation of seabed elevation maps from side-scan sonar images is presented. The side-scan image formation process is represented by a Lambertian diffuse model, which is then inverted by a multiresolution optimization procedure inspired by expectation-maximization to account for the characteristics of the imaged seafloor region. On convergence of the model, approximations for seabed reflectivity, side-scan beam pattern, and seabed altitude are obtained. The performance of the system is evaluated against a real structure of known dimensions. Reconstruction results for images acquired by different sonar sensors are presented. Applications to augmented reality for the simulation of targets in sonar imagery are also discussed.
In this paper a new procedure for the computation of seabed altitude information from side-scan sonar data is presented. Although side-scan sensors do not provide direct measures of seabed elevation, their images are directly related to seabed topography. Using a mathematical model for the sonar ensonification process, approximations to the seabed characteristics can be inferred from the sonar image. The problem is however severely under-constrained, in the sense that not all the parameters involved in the image formation process can be directly determined from the side-scan image. To overcome this difficulty, we propose the utilization of a multi-resolution expectation-maximization framework to select the most probable parameters from the solution space. At every iteration the estimated solution is used to simulate a side-scan image of the observed scene, which is then be compared to the side-scan image actually observed; solution parameters are then refined using gradient-descent optimization. The process is repeated until convergence is achieved.
A proof of concept for a model-less target detection and classification system for sidescan imagery is presented. The system is based on a supervised approach that uses augmented reality (AR) images for training computer added detection and classification (CAD/CAC) algorithms, which are then deployed on real data. The algorithms are able to generalise and detect real targets when trained on AR ones, with performances comparable with the state-of-the-art in CAD/CAC. To illustrate the approach, the focus is on one specific algorithm, which uses Bayesian decision and the novel, purpose-designed central filter feature extractors. Depending on how the training database is partitioned, the algorithm can be used either for detection or classification. Performance figures for these two modes of operation are presented, both for synthetic and real targets. Typical results show a detection rate of more that 95% and a false alarm rate of less than 5%. The proposed supervised approach can be directly applied to train and evaluate other learning algorithms and data representations. In fact, a most important aspect is that it enables the use of a wealth of legacy pattern recognition algorithms for the sonar CAD/CAC applications of target detection and target classification.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.