A fast blink detector using Canonical Correlation Analysis

Blinking is an involuntary action performed by people to keep their eyes moist but this becomes a problem when capturing an image. This problem becomes more pronounced when a flash is being used for image capture. Some of the algorithms that are present in the literature today are quite complex. Also, the detection accuracy of the present algorithms in literature is heavily dependent on temporal information. Temporal information is derived from the continuous video stream, but in the case where a flash is used for capture of an Image, temporal information may not be available. In this paper we propose a simple classification algorithm that is based on CCA (Canonical Correlation Analysis) to perform fast Blink detection for computationally constrained devices.