3D Reflectivity Reconstruction by Means of Spatially Distributed Kalman Filters

In seismic, radar, and sonar imaging the exact determination of the reflectivity distribution is usually intractable so that approximations have to be applied. A method called synthetic aperture focusing technique (SAFT) is typically used for such applications as it provides a fast and simple method to reconstruct (3D) images. Nevertheless, this approach has several drawbacks such as causing image artifacts as well as offering no possibility to model system-specific uncertainties. In this paper, a statistical approach is derived, which models the region of interest as a probability density function (PDF) representing spatial reflectivity occurrences. To process the nonlinear measurements, the exact PDF is approximated by well-placed Extended Kalman Filters allowing for efficient and robust data processing. The performance of the proposed method is demonstrated for a 3D ultrasound computer tomograph and comparisons are carried out with the SAFT image reconstruction.