Paper Info
Title
Human Joint Motion Recognition Using Ultrasound Pulse Echo Based On Test Feature Classifier
Abstract
The simultaneous joint torque and angle recognition for dynamic movement we propose using ultrasound echo and is test feature classifier is for application in human-machine systems such as wearable robots. Ultrasound emitted on the skin surface and reflected within the body is used to recognize features extracted from the echo. Features is change based on the muscle movement related to target joint movements enabling joint movement to be recognized. Experiments conducted to recognize elbow movement involved classifying a test data set into torque and angle classes based on feature scores rectified and integrated in 6 blocks extracted from ultrasound echoes. Correct classification was 54% for torque recognition, and 67% for angle recognition. Some data classified incorrectly was still in classes close to the correct class, with quasi correct classification 88% for torque and 94% for angle, corresponding to allowing larger classes for correct classification. Results demonstrated that recognition using our proposal was more successful than random selection at 11% for torque recognition and 13% for anble recognition. A strategy we describe for reducing the number of features to decrease calculation time cost, reduced by 50% while maintaining accuracy.
Year
DOI
Venue
2009
10.20965/jrm.2009.p0597
JOURNAL OF ROBOTICS AND MECHATRONICS
Keywords
Field
DocType
human motion recognition, ultrasound, test feature classifier (TFC)
Computer vision,Pattern recognition,Motion recognition,Computer science,Pulse (signal processing),Artificial intelligence,Classifier (linguistics),Ultrasound
Journal
Volume
Issue
ISSN
21
5
0915-3942
Citations 
PageRank 
References 
0
0.34
0
Authors
5
Name
Order
Citations
PageRank
Yoichiro Tsutsui100.68
T. Tanaka2814.81
Shun'ichi Kaneko323035.34
Yukinobu Sakata411.79
Maria Q. Feng59713.63