Imaging for X-optimised imaging algorithms for totally unknown defects and components
A Croxford and E Lopez-Villaverde University of Bristol, UK
Ultrasonic phased arrays are used in non-destructive evaluation for detection and characterisation of defects. Over the last decade, provided solutions have been based on the post-processing of the full matrix of transmit-receive signals, involving application of various imaging techniques and scattering matrix analysis for defect characterisation. Conventional imaging methods convert the array data into an image, which facilitates the interpretation of the data. However, in this case essential information needed for characterisation of small defects is lost. The study of a defect scattering matrix, which describes the scattering behaviour of a defect, is a more suitable alternative. However, when other scatterers are located close to the target, the measured scattering descriptor is contaminated due to the overlapping of time responses. This is the case in coarse-grained structures, where grains act as randomly distributed scatterers adding a coherent structural noise to the signal reflected from the defect. This paper investigates how to isolate the target time responses from those of the background noise in order to measure reliably the defect scattering matrix.
The approach is based on a reversible imaging concept. First, an image is produced using the forward-imaging method. Then, the image is spatially filtered according to the region of interest. After that, the inverse imaging operation is performed in order to recover the desired temporal responses corresponding to the signals scattered from the defect. Finally, the target scattering information is extracted from the filtered array data. Two implementations of this procedure are developed on a graphics processing unit (GPU), using delay-and-sum and frequency-wave number approaches. A study of a grained structure is carried out in order to compare the performance of the proposed algorithms with other available techniques. It is shown that detection of small defects is enhanced in highly scattering materials in comparison with standard imaging techniques.
