(a) (b) (c)
FIGURE 6 Recognition accuracy of CCA(a), FBCCA(b), and TRCA(c).
Figure 6b unfurls the recognition accuracy comparison between the
SF_FBCCA methodology, as introduced by Yan et al. [5], and the
ERLLP_FBCCA approach of the present study. Drawing from the findings of
[15], the demarcated corner frequencies for the passband and
stopband for the initial group are registered at [6, 90] Hz, for the
second assembly at [14, 90] Hz, and for the third cluster at [22,
90] Hz. As is manifest in Figure 6b, when measuring over stimulus
durations spanning 1, 1.2, 1.4, 1.6, 1.8, and 2 seconds, the
ERLLP_FBCCA methodology manifests recognition accuracy enhancements of
2.07%, 1.17%, 2.23%, 1.65%, 2.03%, and 1.88%, contrasted against
the SF_FBCCA paradigm.
Figure 6c portrays the recognition accuracy contrasts between the
SF_TRCA approach, as articulated by Yan et al. [5], and the
ERLLP_TRCA methodology elucidated in this manuscript. The data scrutiny
ensues over intervals from 0.5 to 0.8 seconds, incremented by 0.1
seconds. In gauging the recognition accuracy for individual subjects, an
exhaustive leave-one-out cross-validation methodology was harnessed,
leveraging five from the sextet of blocks for indoctrination and
sequentially probing the recognition efficacy on the solitary sixth
block.
As illuminated in Figure 6c, over stimulus durations of 0.5, 0.6, 0.7,
and 0.8 seconds, the recognition accuracy of ERLLP_TRCA method
witnesses surges of 2.40%, 2.82%, 1.02%, and 0.98%, respectively, in
juxtaposition with the foundational SF-TRCA protocol.
In summation, through the judicious integration of the ERLLP technique,
which exhibits a proficient capability for SSVEP signal enhancement,
with established methodologies such as CCA, FBCCA, and TRCA, we have
witnessed a notable elevation in the recognition accuracy metrics. This
convergence elucidates the synergistic potential of combining advanced
signal enhancement with classical recognition techniques in the realm of
neuroscientific research.
| DISCUSSION AND CONCLUSION
In this study, we have presented a novel approach for enhancing the
analysis of steady-state visual evoked potentials (SSVEP) signals. Our
method, termed Enhanced-RL based on Laplacian pyramid (ERLLP), combines
the benefits of fractional-order differentiation and Laplace pyramid to
enhance the features of SSVEP signals. We have demonstrated the
effectiveness of ERLLP in improving the recognition accuracy of SSVEP
target identification, using three well-known algorithms: Canonical
Correlation Analysis (CCA), Filter Bank Canonical Correlation Analysis
(FBCCA), and Task-Related Component Analysis (TRCA).
Our results indicate that the application of ERLLP leads to a
significant enhancement of SSVEP signal analysis. By leveraging
fractional-order differentiation, we successfully enhance the fine
details of SSVEP signals, effectively mitigating the influence of trends
and noise. Importantly, ERLLP consistently improves recognition accuracy
across various stimulus durations for CCA, FBCCA, and TRCA algorithms.
The significance of the ERLLP approach extends beyond improved
recognition accuracy. The integration of image processing techniques and
hierarchical enhancement introduces a novel perspective in brain signal
analysis. Through the combination of fractional-order differentiation
and pyramid structures, we achieve effective noise reduction and trend
elimination in SSVEP signals. This novel approach not only advances
SSVEP target identification but also offers promising applications in
the broader field of brain-computer interface research. Despite the
promising outcomes, it is important to acknowledge challenges and
potential refinements associated with the ERLLP approach. Parameters
such as the fractional-order differentiation order and the number of
pyramid levels warrant further optimization and investigation for
optimal performance. Additionally, the algorithm’s robustness across
individual variations and real-world scenarios requires further
exploration.
In conclusion, the ERLLP algorithm presented in this study introduces
innovative strategies for SSVEP signal analysis. By employing image
processing and hierarchical enhancement techniques, we enhance signal
quality and recognition accuracy. Future research can explore the
adaptability and applicability of this approach in diverse brain signal
analysis tasks. Moreover, the optimization of algorithm parameters and
its extension to various real-world scenarios hold potential for further
advancements in brain-computer interface research.