Eyeblink artifacts often contaminates electroencephalogram (EEG) signals, which could potentially confound EEG's interpretation. A lot offline methods are available to remove this artifact, but an online solution is required to remove eyeblink artifacts in near real time for EEG signal to be beneficial in applications such as brain computer interface, (BCI). In this work, approaches that combines unsupervised eyeblink artifact detection with Empirical Mode Decomposition (EMD) and Canonical Correlation Analysis (CCA) are proposed to automatically identify eyeblink artifacts and remove them in an online setting. The proposed approaches are analysed and evaluated in terms of artifact removal accuracy and ability of the approaches to retain neural information in an EEG signal. Analysis has discovered that the approaches have achieved more than 98 accuracy in detecting and removing eyeblink artifacts in real time. The approaches have produced very low reconstruction error as well, the least is 0.148 in average. These algorithms took about 12ms in average to clean a 1s length of EEG segment, which is fast enough to process the signals in real time. © 2019 IEEE.