Analysis of system trade-offs for terahertz imaging

Abstract

The rapidly developing field of terahertz (or T-ray) imaging promises to provide a non-invasive method of identifying the composition of various objects. Biomedical diagnostics, semiconductor device diagnostics, trace gas analysis and moisture analysis for agriculture are among the growing number of important T-ray applications. We present results using an electro-optical sampling method and discuss how this can be simplified by using a micro antenna array realized by MEMs technology. The challenges and advantages of both approaches are compared. In our set-up, sub-picosecond pulses of terahertz radiation are produced via optical rectification, using an LT gallium arsenide wafer. A coherent method is used, whereby both amplitude and phase of the THz beam are detected, providing an exact signature of the material under observation. From the analysis of the amplitude and phase it is possible to tomographically reconstruct a terahertz-domain image of an object. Current terahertz systems, however, remain too slow for many real-time applications and the need for high speed image processing is evident. Previous systems have relied on simple signal processing algorithms, based on Fourier transformations, for de-convolving the system response and extracting phase/amplitude information. We discuss enhanced processing based on wavelet transforms. Wavelet techniques for signal analysis provide near optimal thresholding for signal recovery and are computationally inexpensive for pulsed waveforms. The improved signal characterization and speed of wavelet analysis over Fourier methods is demonstrated on sample T-ray data.