Development of a mechanism for reconstruction of terahertz single-frequency images of biological samples

Abstract

Algorithmic mechanisms are used to improve terahertz (THz) image quality, which is critical to a biological sample analysis. A complete mechanism for the super-resolution reconstruction and evaluation of THz biological sample images was constructed in this study. With eucalyptus leaves as an example, the THz spectral region screening technique was adopted to select the characteristic frequencies for imaging, and the THz single-frequency images were reconstructed with the single-image super-resolution image reconstruction technique. The THz super-resolution reconstructed images without ideal reference were evaluated after the introduction of three no-reference image evaluation criteria considering the diversity and complexity of organisms. The results show that the THz image reconstruction mechanism proposed in this study led to an increase in resolution and a decrease in noise. At the same time, the imaging quality of biological samples was considerably improved, and the detailed information was enriched. These provide a reference for a THz imaging analysis of leaves and other biological samples.