Hand Gesture Identification

Classification of Finger Movements with Surface Electromyography

Recently, there are numerous applications which are related to hand gesture identification. Therefore it is a big field of research to make this interference more precise than before. The hand gesture identification is used mostly for the use of old people for the BCI interaction, also it can be used for the new technology. There are many modalities like, vision based systems, mechanical sensors, etc. We use EMG signal to find the hand gesture because of its ease of recording and non-invasiveness. EMG measures electrical currents that are generated in a muscle during its contraction and represent neuromuscular activity. The pattern recognition of the signal which helps the interfacing is the most difficult part because of its variations. The process of pattern recognition can be broken down into three main phases: feature extraction, feature reduction, and classification. Comparing the EMG signal with the other bio signals EMG is a much noisier signal and it contains complicated noises which are caused by environment, motion artifacts, and electromagnetic induction. Therefore, it is necessary to have a phase named pre-processing (e.g. amplification, filtering) in our process in order to reduce the noise of our EMG signal. Feature extraction is to transform signals to a set of feature signals. Feature reduction is the process of reducing the dimensionalities of feature vectors which may simplify the classifier’s task. Classification map feature vectors into certain classes using training data. Various classification data have been employed including linear discriminant classifiers. There will be a continual expansion of the available tools in this field of research in the near future.

Data Acquisition:

Two channels : 

Channel 0 on the buttom, 

channel 1 on the top of the wrist. 

Each 60 seconds, 

40KHz sample rate

00: fist (finger flexion) every 5 sec (and on 59s),hand in rest in other times

01: double fist (finger flexion) every 5 sec, hand in rapid motion in other times

02: extension fingers every 5 sec, hand in rest in other times

03: double extension every 5sec, hand in rapid motion in other times

04: swipe left (wrist flexion) every 5 sec (and on 59s),hand in rest in other times

05: double swipe left (wrist flexion) every 5 sec, hand in rapid motion in other times

06: turn right wrist every 5 sec, hand in rest in other times

07: double turn right every 5sec, hand in rapid motion in other times

08: turn left wrist every 5 sec, hand in rest in other times

09: double turn left every 5sec, hand in rapid motion in other times

10: keep fisting for 10sec, keep extending fingers for 10s, keep left swipe for 10s,

keep right swipe (wrist extension) 10s, keep turn right 10s, keep turn left 10s

Collaborator:

NR Sign Inc. EEG & EMG Solutions