A Comparative Study of Canonical Correlation Analysis and Power Spectral Density Analysis for SSVEP Detection

Abstract

Steady-state visual evoked potentials (SSVEPs) are widely employed for target detection in brain-computer interfaces (BCIs). Canonical correlation analysis (CCA), which extends ordinary correlation to two sets of variables, is a new method for SSVEP detection. In this paper, the performance of CCA is compared with that of traditional power spectral density analysis (PSDA) in terms of power spectral amplitude, recognition accuracy, information transfer rate and operating speed. The results show that the CCA method outperforms the PSDA in all these technical indexes.