3.2 | SSVEP Feature Enhancement Method based on ERLLP
In this comprehensive study, we leverage the multi-channel signal, designated as, which encompasses selected SSVEP signal analysis channels namely O1, O2, Oz, PO3, PO4, POZ, PO5, and PO6. When conceptualized as the input image and subsequently processed through the intricacies of the Laplace pyramid in conjunction with the Enhanced-RL operator, as delineated in Figure 2, the resultant is a meticulously sharpened and filtered signal, termed . Notably, when the cutting-edge ERLLP algorithm is integrated into spectral analysis, especially in realms as nuanced as medical diagnostics and neuroscience research, this refined signalis positioned as the focal point of analysis, serving as a pivotal analyzed signal.
Drawing a comparison to prevailing methodologies outlined in literature [5], where the Prewitt, Sobel, and Laplace operators are predominantly employed, our research pivots towards utilizing the Enhanced-RL operator. This approach not only serves as an innovative filtering template, tailored for gaze target recognition in the SSVEP BCI paradigm, but also seamlessly facilitates hierarchical filtering. Intriguingly, pyramid images cultivated at diverse hierarchical levels exhibit variances in resolution and intricate detail information. The granularity of this detail information becomes increasingly conspicuous as one navigates towards the base of the pyramid. Given this gradation, higher pyramid tiers inherently dictate the selection of an augmented fractional order of differentiation. As the research progresses, there’s a synthesis of this advanced sharpened filtering technique with renowned methods such as CCA, FBCCA, and TRCA. This amalgamation births the novel methodologies christened as ERLLP_CCA, ERLLP_FBCCA, and ERLLP_TRCA, visually elucidated in Figure 3.
To dissect the analytical approach further, initially, the discrimination coefficients of the pristine test signal , as calculated through CCA, FBCCA, and TRCA methodologies, are labeled as. Subsequent to this, the ERLLP algorithm is harnessed to refine X. Post this enhancement, the discrimination coefficients of the sharpened and meticulously filtered signal, as derived from CCA, FBCCA, and TRCA techniques, are christened as. Culminating the analysis, the ERLLP_CCA, ERLLP_FBCCA, and ERLLP_TRCA methodologies harness an integration of the coefficients fromand, presenting them as the ultimate discrimination coefficients, aptly captured in the ensuing mathematical equation:
| Experimental Results and Discussion
4.1 | Signal Processing
Figure 4 provides a visual elucidation of the transformative impact of ERLLP on SSVEP signals, offering a comparative analysis of waveforms before and after the application of ERLLP as a sharpening mechanism for the SSVEP signals.
In Panel 4-A, the depicted waveform represents the innate characteristics of the unaltered SSVEP signal. This original waveform prominently showcases significant fluctuations, where the signal intermittently deviates from its baseline, leading to an inconsistent distribution over time. This pronounced deviation, termed as the ’trend term’, carries substantial weight, as it profoundly impacts the overall efficacy and interpretability of the signal.
Transitioning to Panel 4-B, a striking contrast is observed. Presented here is the waveform of the SSVEP signal post its refinement through the ERLLP algorithm. The resultant waveform demonstrates a remarkable stability, underscoring the prowess of ERLLP in adeptly mitigating the previously observed trend term inherent to the SSVEP signal. This not only enhances the clarity and consistency of the waveform but also underscores the algorithm’s efficacy. Additionally, the refined waveform subtly alludes to another advantageous trait of ERLLP: its capability to attenuate extraneous noise in the SSVEP signal, further bolstering the signal’s integrity and reliability for subsequent analytical endeavors.