A novel human–machine interface based on recognition of multi-channel facial bioelectric signals

Abstract

This paper presents a novel human-machine interface for disabled people to interact with assistive systems for a better quality of life. It is based on multi-channel forehead bioelectric signals acquired by placing three pairs of electrodes (physical channels) on the Frontalis and Temporalis facial muscles. The acquired signals are passed through a parallel filter bank to explore three different sub-bands related to facial electromyogram, electrooculogram and electroencephalogram. The root mean square features of the bioelectric signals analyzed within non-overlapping 256 ms windows were extracted. The subtractive fuzzy c-means clustering method (SFCM) was applied to segment the feature space and generate initial fuzzy based Takagi-Sugeno rules. Then, an adaptive neuro-fuzzy inference system is exploited to tune up the premises and consequence parameters of the extracted SFCMs rules. The average classifier discriminating ratio for eight different facial gestures (smiling, frowning, pulling up left/right lips corner, eye movement to left/right/up/down) is between 93.04% and 96.99% according to different combinations and fusions of logical features. Experimental results show that the proposed interface has a high degree of accuracy and robustness for discrimination of 8 fundamental facial gestures. Some potential and further capabilities of our approach in human-machine interfaces are also discussed.

Robust Facial Expression Recognition for MuCI: A Comprehensive Neuromuscular Signal Analysis

“…Considering the processing steps, segments with different lengths were adopted. In [29], five time frames between 64 to 1024 msec were inspected on facial myoelectric signals and segment length of 256 msec was recommended, in line with the findings by Englehart et al [4] for upper limb EMG classification. In terms of feature extraction, time-domain features have been widely used for classification of myoelectric signals as they are easy to compute and can provide better discriminative information than frequency and timefrequency features [45].…”

“…They stated that accuracy is not considerably enhanced by increasing the segment length for all single features which supports the idea that a segment with 200 msec length contains enough information. Recently, it was reported that segments with 256 msec length would be the best choice when dealing with facial EMG signals [29]. Thus, in this study EMGs are segmented into nonoverlapped windows with 256 msec length prior to feature extraction.…”

“…However, in previous studies this behavior has not been widely considered since only very few methods for either feature extraction or classification have been investigated. Classification accuracy has always been the most popular measure to assess the system performance; however, computational cost as another essential metric has only been considered in [29], [36][37][38][39] and [42].…”

Facial neuromuscular signal classification by means of least square support vector machine for MuCI

“…They reached 92.6% recognition accuracy by extracting RMS features and classifying them with subtractive fuzzy c-means (SFCM) clustering method. In their recent study [18], they attempted to classify eight gestures using SFCM plus adaptive neuro-fuzzy inference system (ANFIS). By considering only EMG signals, 93.02% discrimination accuracy was attained.…”

Electrooculography in the Field of Assistive Interaction Communication

“…EOG is used in previous works so as to interact with different devices. Considering the characteristics of EOG signals, EOG-based HCI systems have become more popular in recent years [11]- [13]. For instance researchers used EOG to control a robot arm moment [17], guide a wheelchair [15], or a key-board [14].…”