Select dataset 1 and dataset 2 to carry out experiments, and use SVM, RVM, and MSRR algorithm for classification.
The recognition results by SRC and MSRR are in Figures 13 and 14, respectively, where the numbers 1-7 represent anger, disgust, fear, happiness, neutral, sadness, and surprise, respectively.
The difference between the smallest residual value and the second smallest residual value of MSRR is larger than that of SRC, which means that difference between classes in MSRR is relatively significant.
Take the expression anger as test object, and get the final result by SRC and MSRR, which is shown in Figures 15 and 16, respectively.
In Figure 15, the results for SRC and MSRR are on the left and right, respectively.
MSRR algorithm suffers less influence of unknown samples compared with SVM and RVM algorithm.
And the real and imaginary parts of permeability of CSRR and MSRR are shown in Fig.
It is observed that the MSRR is effective in broadening the bandwidth and realizing dual-band resonance of THz metamaterial.
The different electrical responses of CSRR and MSRR to the electromagnetic wave of normal incidence can be analyzed by the surface current distribution, which is shown in Fig.
Considering the similarity of surface current distribution and resonance frequency, one may presume that the resonance of MSRR at 515 GHz is mainly originated from SRR1.
In order to verify the design and simulation results, the metal planar array pattern of CSRR and MSRR were fabricated using photolithography and magnetron sputtering.