Imaging multilayered objects with complex geometry

摘要

Imaging of solid layered objects using ultrasound presents normally a difficult task. Delay-and-sum (DAS) method, a standard and well-known method used for improving resolution in immersion setup, cannot properly reconstruct an image of layered objects. More advanced algorithms have been proposed for that purpose, like multi layered delay-and-sum (MLDAS) and ray tracing or phase shift migration (PSM). However, these algorithms work properly only when layers are parallel to each other and their surface is perpendicular to the transducer's axis. When an object has complex geometry the problem is even more complicated and only one of those algorithms (ray tracing) is capable of reconstructing the ultrasound image properly. In this paper two algorithms capable of imaging layered objects with complex geometry are presented and their performance is compared: first, is the modified MLDAS that takes under consideration object geometry, and second, the generalized version of PSM (GPSM). The Matlab implementations of those algorithms are evaluated on the experimental data from a copper block immersed in water. The results are compared and advantages and disadvantages of the presented methods are discussed. Half power resolution and signal-to-noise ratio are measured and presented as a quality indicator. Additionally, some modifications to the GPSM algorithm are presented that can enhance the reconstructed image quality.