Eye Detection using Composite Cross-Correlation form Face Images in Varied Illumination

Mohd Nordin, Abdul Rahman and Mohd Rizon, Mohamed Juhari and Kutiba, Nanaa (2014) Eye Detection using Composite Cross-Correlation form Face Images in Varied Illumination. In: International Conference on Artificial Life and Robotics, 11-13 January 2014, OITA, JAPAN.

Text FH03-FIK-14-00397.pdf Restricted to Registered users only Download (3MB)

Abstract

Facial feature detection is essential process in the field of face recognition. Template match based approach is widely used to implement the issue of facial detection. However, many obstructions adversely affect in template based implementation such as varied illumination, size, face pose and the status of eye in either open case or close case. the objective of this paper is to detect left eye and overcome the obstruction of low illumination in given image using composite cross-correlation. Two single templates form the proposed composite template. First single template is created from an eye image in ordinary illumination while the second single template is created from an eye image in low illumination. The experimental results is concluded by applying our templates on the PICS Database whose varied illumination images and show that applying composite templates gives a better result than applying the both single templates individually.

Actions (login required)

View Item