Towards hand gesture based writing support system for blinds. (September 2016)

• Record Type: Journal Article
• Authors: Modanwal, Gourav; Sarawadekar, Kishor
• Is Part Of: Pattern recognition. Volume 57(2016:Sep.)
• Page Start: 50
• Page End: 60
• Publication Date: 2016-09
• Subjects: HCI -- Reduced shape signature -- Gesture recognition -- Dactylology and rule based classifier; Pattern perception -- Periodicals; Perception des structures -- Périodiques; Patroonherkenning; 006.4
• DOI: 10.1016/j.patcog.2016.03.026
• Languages: English
• ISSNs: 0031-3203

Abstract

Abstract: In this digital age, personal computers are widely used. But the blind people cannot handle it easily. Human–computer interaction (HCI) may play vital role in overcoming this obstacle. Computer keyboard contains about 104 keys, whereas six-dot Braille keyboard supports only 63 symbols. To bridge this gap, a new dactylology is proposed in this article. Additionally, a new feature extraction technique, Reduced Shape Signature (RSS), is introduced in this article which is rotation, translation and scale invariant. Further, RSS is compact as it reduces the number of feature-sets by 35%. To discriminate intra-class gestures difference angle and polygonal area are computed along with RSS. The symbols recognition rate of the proposed algorithm is 97.53%. Highlights: A new dactylology is proposed to achieve functionality similar to a standard keyboard. A new feature extraction technique called as reduced shape signature (RSS) is introduced. Use of difference angle and polygonal area is proposed to discriminate Intra-class gestures. Symbols are entered in the Microsoft Word application with the proposed experimental setup and real time data entry support is verified.