Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2015 May 30:14:48.
doi: 10.1186/s12938-015-0046-0.

Envelope filter sequence to delete blinks and overshoots

Manuel Merino  1 , Isabel María Gómez  2 , Alberto J Molina  3
Affiliations

Envelope filter sequence to delete blinks and overshoots

Manuel Merino et al. Biomed Eng Online. .

Abstract

Background: Eye movements have been used in control interfaces and as indicators of somnolence, workload and concentration. Different techniques can be used to detect them: we focus on the electrooculogram (EOG) in which two kinds of interference occur: blinks and overshoots. While they both draw bell-shaped waveforms, blinks are caused by the eyelid, whereas overshoots occur due to target localization error and are placed on saccade. They need to be extracted from the EOG to increase processing effectiveness.

Methods: This paper describes off- and online processing implementations based on lower envelope for removing bell-shaped noise; they are compared with a 300-ms-median filter. Techniques were analyzed using two kinds of EOG data: those modeled from our own design, and real signals. Using a model signal allowed to compare filtered outputs with ideal data, so that it was possible to quantify processing precision to remove noise caused by blinks, overshoots, and general interferences. We analyzed the ability to delete blinks and overshoots, and waveform preservation.

Results: Our technique had a high capacity for reducing interference amplitudes (>97%), even exceeding median filter (MF) results. However, the MF obtained better waveform preservation, with a smaller dependence on fixation width.

Conclusions: The proposed technique is better at deleting blinks and overshoots than the MF in model and real EOG signals.

PubMed Disclaimer

Figures

Figure 1
Figure 1
EOG electrode layout. Electrodes H and V record horizontal and vertical eye movements.
Figure 2
Figure 2
EOG signal. Saccades, blinks, overshoots and pursuit movements.
Figure 3
Figure 3
Envelope filter. a Steps 1–3 of envelope filter, b the result of envelope filter loop.
Figure 4
Figure 4
Offline EFS diagram. Steps 1–6.
Figure 5
Figure 5
Online EFS diagram. [1:N]: Input buffer length; [1:M]: input/output data length (M < N); [M + 1:N]: overlapping width; [L:M]: fragment size of filtered data (L ≤ M).
Figure 6
Figure 6
Results of Tests 1 and 2. a Best amplitudes of FS-WN; b best lengths of mean filter.
Figure 7
Figure 7
Results of Test 3. Matching IB length of {0.6, 0.7, 0.8, 0.9, 1.0} s with OB length of {0.05, 0.10, 0.15} s with overlapping of {0.1, 0.2, 0.3, 0.4} s. a Root mean square error; b cross correlation.
Figure 8
Figure 8
Blink removal. a Root mean square error; b percentage decrease of blink amplitudes; c cross correlation; d percentage of blinks whose amplitudes after filtered processing exceeded 25% of original amplitudes.
Figure 9
Figure 9
Preservation of GBM waveform. Pulse width changes from interval [0.3, 0.4] s up to [1.4, 1.5] s. a Root mean square error; b cross correlation.
Figure 10
Figure 10
Preservation of stair waveform (reading activity). Step width changes from interval [0.3, 0.4] s up to [1.4, 1.5] s. a Root mean square error; b cross correlation.
Figure 11
Figure 11
Preservation of random waveform. Step widths change from interval [0.3, 0.4] s up to [1.4, 1.5] s. a Root mean square error; b cross correlation.
Figure 12
Figure 12
Fixation slope effects. Variation of SNR of input data, MF and EFS outputs. Slope changes from interval [0, 5]% up to [35, 40]%.
Figure 13
Figure 13
Result of real EOG processing. Results of MF, offline and online EFS. a Summation of blink area reduction; b overshoot amplitude variation; c saccade slope preservation for reading, natural and go-back eye movements.
Figure 14
Figure 14
Test of real EOG. Visual analysis of results of MF, offline and online EFS.
Figure 15
Figure 15
Model EOG signal from EOG-SG. a Individual elements; b joining of all components.
Figure 16
Figure 16
Saccade model. Sigmoid function with a = 0 and b = 1.
Figure 17
Figure 17
Different fixation slopes. Model EOG signal with four different fixation slopes.
See this image and copyright information in PMC

References

    1. Carretié Arangüe L, Iglesia Dorado J, editors. Psicofisiología. Fundamentos metodológicos. Pirámide; 2000. ISBN: 84-368-0877-0.
    1. Galley N. Saccadic eye movement velocity as an indicator of (de)activation. A review and some speculation. Int J Psychophysiol. 1989;3:229–244.
    1. App E, Debus G. Saccadic velocity and activation: development of a diagnostic tool for assessing energy regulation. Ergonomics. 1998;41(5):689–697. doi: 10.1080/001401398186856. - DOI - PubMed
    1. Fabbri M, Pizza F, Mogosso E, Ursino M, Contardi S, Cirignotta F, et al. Automatic slow eye movement (SEM) detection of sleep onset in patients with obstructive sleep apnea syndrome (OSAS): comparison between multiple sleep latency test (MSLT) and maintenance of wakefulness test (MWT) Sleep Med. 2010;11:253–257. doi: 10.1016/j.sleep.2009.05.020. - DOI - PubMed
    1. Iwanaga K, Saito S, Shimomura Y, Harada H, Katsuura T. The effect of mental loads on muscle tension, blood pressure and blink rate. J Physiol Anthropol Appl Hum Sci. 2000;19(3):135–141. doi: 10.2114/jpa.19.135. - DOI - PubMed

LinkOut - more resources