Evaluation of features for automated transcription of dual-handed sign language alphabets

Author

Lilha, Himanshu ; Shivmurthy, Devashish

Author_Institution

Dept. of Comput. Sci. & Eng., PES Sch. of Eng., Bangalore, India

fYear

2011

fDate

3-5 Nov. 2011

Firstpage

1

Lastpage

5

Abstract

Sign language helps the deaf and mute to communicate effectively. The paper demonstrates the evaluation of various feature extraction techniques for the dual -handed sign language alphabets. The efficiency of features like Histogram of Orientation Gradient (HOG) is discussed followed by the demonstration of the Histogram of Edge Frequency (HOEF) which overcomes the short coming of HOG. The evaluation of HOG accuracy is found to be 71.4% whereas with HOEF it is found to be 98.1%. The paper also demonstrate the overall system for the sign language recognition.

Keywords

feature extraction; gesture recognition; automated dual-handed sign language alphabet transcription; feature evaluation; feature extraction techniques; histogram of edge frequency; histogram of orientation gradient; sign language recognition; Accuracy; Feature extraction; Handicapped aids; Histograms; Image edge detection; Information processing; Skin; HOEF; HOG; ISL; Sign language; dual-handed;

fLanguage

English

Publisher

ieee

Conference_Titel

Image Information Processing (ICIIP), 2011 International Conference on

Conference_Location

Himachal Pradesh

Print_ISBN

978-1-61284-859-4

Type

conf

DOI

10.1109/ICIIP.2011.6108943

Filename

6108943