Saha, S. and Datta, S. and Konar, A. and Banerjee, B. and Nagar, Atulya K. (2016) A Novel Gesture Recognition System Based on Fuzzy Logic for Healthcare Applications. In: 2016 IEEE International Conference on Fuzzy Systems, 24-29 July 2016, Vancouver, Canada.
12. FUZZ-16546-A Novel Gesture Recognition System Based on Fuzzy Logic for Healthcare Applications.pdf - Accepted Version
Download (1MB) | Preview
This work demonstrates an interesting approach to gesture recognition for elderly people for the purpose of health monitoring at home. The system proposes to detect disorder symptoms on the basis of gesture analysis and generate alarms, thereby finding significance in elderly healthcare. Here the gestures are tracked using Microsoft’s Kinect sensor. From each frame captured by the Kinect sensor, four centroids representing four parts of the body are calculated and from these four centroids a novel feature set is extracted in terms of Euclidean distances and angles. We have noticed that for different persons’ body types the extracted features might vary. Thus to accommodate these non-uniformities, we have used the concept of interval type-2 fuzzy logic based classification. The unknown gesture is recognized based on matching with all the known gestures from the dataset. The proposed methodology provides a high accuracy rate of 92.14%.
|Item Type:||Conference or Workshop Item (Paper)|
|Additional Information and Comments:||(c) 2016 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other users, including reprinting/ republishing this material for advertising or promotional purposes, creating new collective works for resale or redistribution to servers or lists, or reuse of any copyrighted components of this work in other works."|
|Keywords:||Healthcare; interval type 2 fuzzy set; Kinect sensor|
|Faculty / Department:||Faculty of Science > Mathematics and Computer Science|
|Depositing User:||Atulya Nagar|
|Date Deposited:||09 Jun 2016 11:44|
|Last Modified:||03 Feb 2017 15:12|
Actions (login required)