Objective In terahertz digital holographic imaging, factors such as the transmittance and noise of shelters will affect the imaging effect, and when combined with practical applications of terahertz imaging technology such as security detection and medical imaging, shelters have become an important issue that needs to be studied. Double-exposure digital holography can achieve better imaging quality than single-exposure for relatively stationary targets. According to our investigation, there are no reports on the influence of shelters in terahertz double-exposure digital holography. Paper is a common material that does not transmit visible light but has high terahertz wave transmittance, and it is easy to change the noise and transmittance conditions in the experiment by changing the material; thus, studying the influence of paper shelters in the terahertz imaging process has a good reference value for the use of other materials.
Methods In this study, the imaging results and objective evaluation indexes of material targets with lateral resolutions of 0.2 mm and 0.3 mm under different shelters are obtained through simulation and experiments by the D-APRA (double-exposure amplitude constrained phase retrieval algorithm) using a terahertz double-exposure digital holographic imaging system with a wavelength of 118.83 μm, a detector pixel size of 0.1 mm, and a pixel number of 124×124. Through comparative analysis, we study the effect of B5 paper and thermal paper on the reconstruction of amplitude images.
Results and Discussions First, the transmittance and noise distribution of a single sheet of B5 paper and thermal paper are measured to determine the shelter’s simulation parameters (Fig.4). The transmittances are 0.4 and 0.7, respectively, with 0.002 and 0.006 Gaussian noise variances. Second, the detector’s noise parameters are deduced from the experimental data. The mean value of the detector’s Gaussian noise is about 0.15, and the variance is about 0.0009 (Fig. 5). The imaging results of two targets with resolutions of 0.2 mm and 0.3 mm under different conditions of shelters are obtained through simulations (Fig. 6 and Fig. 7), and the objective evaluation indicators of different conditions are compared and analyzed (Table 1). Finally, the imaging results of different situations are obtained through experiments (Fig. 8 and Fig. 9), and some objective evaluation indicators (Table 2) obtained by calculation are close to the simulation results. The overall change trend of the experimental results is similar to the simulation results.
Conclusions This research investigates the effect of B5 paper and thermal paper shelters on the amplitude image reconstruction of 2.52 THz double-exposure digital holography using experiments and simulation for the two targets with lateral resolutions of 0.2 mm and 0.3 mm. First, preliminary experiments are used to evaluate the transmittance and noise distribution of a single sheet of B5 paper and thermal paper; the transmittance is about 0.4 and 0.7, and the Gaussian noise variance is about 0.002 and 0.006, respectively. Second, the result shows that the mean value of the detector’s Gaussian noise is about 0.15, and the variance is about 0.0009. The simulation imaging results of two targets with different shelters are compared and analyzed. The reconstructed amplitude image of a target with 0.2 mm lateral resolution sheltered by thermal paper and the images of a target with 0.3 mm lateral resolution sheltered by B5 paper and thermal paper are obtained through experiments. The overall trend of experimental results is similar to the results of the simulation. The contrast ratio of the 0.3 mm lateral resolution target with the B5 paper shelter is lower than the result of simulation by more than 0.2, mainly caused by the low transmittance of B5 paper. With a lateral resolution of 0.3 mm, the contrast ratio of the peripheral area sheltered by the thermal paper target is 0.46, which is similar to the simulation result and only 0.07 lower. Experiments yield a reconstruction image of an unsheltered 0.2 mm lateral resolution target, with a contrast ratio of 0.59 in the central area, which is only 0.02 higher than the simulation result. Finally, the mean square error of the simulation and experimental results in various scenarios are determined, which is more accurate in determining the deviation degree of the reconstructed images from the original images. It works in concert with the contrast ratio result.
Low and unstable laser energy, low detector sensitivity, and environmental noise are the main reasons for the discrepancy between experimental and simulation results. It is necessary to further study the conditions and specific values that need to be added in the simulation and improve the simulation model to reduce the difference between the simulation and experimental results. Darkrooms and shields can also be used in the imaging system to reduce the impact of detector noise, and the reconstruction algorithm can be improved in digital image processing to solve problems such as laser fluctuations and detector noise.
