DocumentCode :
707402
Title :
Comparative analysis of edge detection operators for better glint detection
Author :
Sharma, Anjana ; Abrol, Pawanesh
Author_Institution :
Dept. of Comput. Sci. & IT, Univ. of Jammu, Jammu, India
fYear :
2015
fDate :
11-13 March 2015
Firstpage :
973
Lastpage :
977
Abstract :
Eye gaze is either the gaze direction or the point of gaze of an eye relative to the head to determine a person´s line of sight or point of fixation. Gaze based input is processed to initiate different commands remotely in computing systems to provide ease of use, low cost, speedy and safer interaction. For detecting gaze basically the position of a user´s pupil and glint is determined and then mapped into the region of interest of the user on an interface device. Different edge detectors with wide range of parameters can be used for detecting the pupil and glint coordinates. In this research paper, the output of different edge detectors has been compared for glint formation on the basis of various eye images. These images have been tested with a wide range of thresholds and alpha factor to find out the best edges of the glint. The result of this study shows that different control parameters are required for better results. Further this research can be used to study the effect of various attributes on the accuracy of eye gaze techniques. Various image processing functions of MATLAB have been used for the proposed research.
Keywords :
edge detection; feature extraction; gaze tracking; user interfaces; MATLAB; edge detection operator; eye gaze direction; glint coordinate detection; image processing function; interface device; pupil coordinate detection; Detectors; Estimation; Feature extraction; Image edge detection; Iris; Silicon; Standards; Edge detector; Ellipse; Eye gaze; Glint; Pupil;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Computing for Sustainable Global Development (INDIACom), 2015 2nd International Conference on
Conference_Location :
New Delhi
Print_ISBN :
978-9-3805-4415-1
Type :
conf
Filename :
7100392
Link To Document :
بازگشت