Object-based high contrast travel time tomography

摘要: We consider travel time tomography problems involving detection of high contrast, discrete velocity structures. This results in a nonlinear inverse problem, for which traditional grid-based models and iterative linearized least-squares reconstruction algorithms are not suitable. is because paths change significantly near the contrast structure, making it more difficult to inversely calculate path infer along path. propose model-based approach describe structure using pre-defined elementary objects. Compared model, our has complexity that increases as function number objects, rather than increasing with cells (usually very large). A new algorithm developed provides estimates probability appears at any point region interest. Simulation show method can efficiently sample model parameter space, we map parameters into structures spatial domain generate "probability map", represent appearance different regions. only gives highest optimal solution, but also includes other possible well.