Model-based defect recognition by 3-dimensional sparse ultrasonic arrays

A new technique for defect imaging is presented which is able to distinguish between several defect shapes and produce an accurate image with small expense of hardware and time. High efficiency is achieved by utilizing a priori information about the reflector models in the reconstruction algorithm. The main requirement for applying this technique is a limited number of possible defect shapes, each of which can be described by a sufficiently simple geometric model. By applying 1-, 2- or 3-dimensional arrays with few transmitters and receivers it is possible to obtain the required information by few measurements. A model-based reconstruction algorithm adapted to the specific measurement problem enables the classification of the defects and the estimation of the corresponding model parameters.