![]() To obtain a reliable pupil contour, infrared (IR) illumination is required. In this scenario, only the iris boundary is visible instead of the pupil hence the pupil contour estimation is unavailable, and the eyelid contour is only able to estimate blink but no other types of eye activity. Historically, eyelid contour estimation is often conducted on eye images under normal light conditions and in the far field. Therefore, robust and accurate estimation of eyelid and pupil contours is essential for these applications. With specific algorithms, eyelid closure, pupil size, blink event, eye movement, and gaze direction can also be measured or developed. Nevertheless, the basis for capturing these eye activities or biometrical information from an eye image are the eyelid contour and pupil contour. For example, eye shape is part of the face for face verification ( Vezzetti et al., 2016, 2017), and iris texture is extracted from eye images for identity verification ( Bowyer and Burge, 2016). As opposed to these dynamic changes in eye components (eyelid opening, pupil size and location, blink length and depth) which form eye activities, static eye images are also of interest especially in biometrics. The types and applications of eye activity that have been investigated up to now include gaze direction as a pointing device for paralyzed people ( Duchowski, 2007) pupil size, blink, and eye movement (fixation and saccade) for cognitive load measurement ( Chen et al., 2011), emotion recognition ( Lu et al., 2015), visual behavior change ( Chen et al., 2013), human activity recognition ( Bulling et al., 2011), and mental illness diagnosis ( Vidal et al., 2012) eyelid closure for emotion recognition ( Orozco et al., 2009), and fatigue detection ( Yang et al., 2012 Daniluk et al., 2014). Our work is the first study proposing a unified approach for eye activity estimation from near-field IR eye images and achieved the state-of-the-art eyelid estimation and blink detection performance.Įye activity has been of great interest since observations of human attention and intention began ( Duchowski, 2007). This unified approach greatly facilitates eye activity analysis for research and practice when different types of eye activity are required rather than employ different techniques for each type. Cross-corpus evaluation results show that the proposed method improves on the state-of-the-art eyelid detection algorithm. Blink detection can be as high as 90% in recall performance, without direct use of pupil detection. Evaluation on three different realistic datasets demonstrates that the proposed three-state deformable shape model achieves state-of-the-art performance for the open eye with iris and pupil state, where the normalized error was lower than 0.04. The most likely eye state is determined based on the learned local appearance. ![]() ![]() Unlike the facial landmark estimation problem, by comparison, different shape models are applied to all eye states-closed eye, open eye with iris visible, and open eye with iris and pupil visible-to deal with the self-occluding interactions among the eye components. This paper presents a unified approach to simultaneously estimate the landmarks for the eyelids, the iris and the pupil, and detect blink from near-field IR eye images based on a statistically learned deformable shape model and local appearance. Current approaches often estimate a single eye activity, such as blink or pupil center, from far-field and non-infrared (IR) eye images, and often depend on the knowledge of other eye components. The eyelid contour, pupil contour, and blink event are important features of eye activity, and their estimation is a crucial research area for emerging wearable camera-based eyewear in a wide range of applications e.g., mental state estimation. ![]() 1School of Electrical Engineering and Telecommunications, University of New South Wales, Sydney, NSW, Australia.
0 Comments
Leave a Reply. |
Details
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |