| Abstract [eng] |
Electromagnetic THz radiation sources have attracted a lot of attention in the last decade due to their wide applications: from non-invasive spectroscopy [1-3] to non-ionizing sample imaging [5,6]. One of the more important areas of applications of optics is the imaging of objects, their recognition and inspection of samples. Due to the complicated detection, THz imaging is usually implemented by using one pixel methods, for example, using raster scan by changing the position of the sample in the transverse plane. We tried to find the best resolution by changing the distance between the sample and the collecting element z2 and by trying different configurations of the illuminating and light collecting elements (to reduce the degree of freedom, distance z3 was chosen equal to 11,5 mm and the distance between the illuminating element and the sample z1 was chosen corresponding to the highest intensity of the beam). Four diffractive masks were investigated: cubic (Airy), axicon, thin lens, and Fibonacci lens. The image was evaluated using 4 parameters: MSE, brightness, PSNR and contrast. An additional parameter, resolution, was also introduced. It is defined by spatial frequency (measured in line pairs per millimeter) when the image contrast drops below 20%. While investigating incoherent imaging, we found that some configurations (such as thin lens and axicon, where the first element is the illuminating element and the second is the collecting element) give broad zones of high resolutions while other configuration (such as thin lens and Fibonacci lens) give very narrow zones of very high resolution. During the investigation of coherent imaging (phase and amplitude), it was found that best resolution was achieved using Fibonacci lens - thin lens configuration (for amplitude imaging) and thin lens- cubic Airy mask (for phase imaging). This means that the optimal configuration for each imaging type (intensity, phase, amplitude) is different, i.e., there is no one optimal configuration for all cases. Additionally, it was discovered that the conventional imaging configuration using thin lenses did not provide superior resolution nor a broad zone of high contrast. In general, incoherent imaging gives better results in terms of resolution and PSNR. On the other hand, coherent imaging is more suitable for semi-transparent sample recognition (in determining it's complex index of refraction). Fabrication errors were also investigated. Simulation of one pixel coherent imaging was carried out using modeled and fabricated thin lenses. Here, the fabricated lenses were numerically analyzed by scanning their surface with a profilometer and uploading the data collected to a computer. The findings reveal that fabrication errors make it harder to predict the ideal conditions (mainly the distance from the sample and the collecting element) using simulations. Thus these errors pose a significant challenge to the effectiveness of coherent imaging and must be tackled first. |
