Auto-focusing method for remote gaze tracking camera

Abstract

Gaze tracking determines what a user is looking at; the key challenge is to obtain well-focused eye images. This is not easy because the human eye is very small, whereas the required resolution of the image should be large enough for accurate detection of the pupil center. In addition, capturing a user's eye image by a remote gaze tracking system within a large working volume at a long Z distance requires a panning/tilting mechanism with a zoom lens, which makes it more difficult to acquire focused eye images. To solve this problem, a new auto-focusing method for remote gaze tracking is proposed. The proposed approach is novel in the following four ways: First, it is the first research on an auto-focusing method for a remote gaze tracking system. Second by using userdependent calibration at initial stage, the weakness of the previous methods that use facial width in captured image to estimate Z distance between a user and camera, wherein each person has the individual variation of facial width, is solved. Third, the parameters of the modeled formula for estimating the Z distance are adaptively updated using the least squares regression method. Therefore, the focus becomes more accurate over time. Fourth, the relationship between the parameters and the face width is fitted locally according to the Z distance instead of by global fitting, which can enhance the accuracy of Z distance estimation. The results of an experiment with 10,000 images of 10 persons showed that the mean absolute error between the ground-truth Z distance measured by a Polhemus Patriot device and that estimated by the proposed method was 4.84 cm. A total of 95.61% of the images obtained by the proposed method were focused and could be used for gaze detection.

Improved iris localization by using wide and narrow field of view cameras for iris recognition

“…To overcome this limitation, Lee et al used the least squares regression method to enhance the accuracy of the Z distance estimation by updating the model parameters for estimating the Z distance. 23 However, their method requires user-dependent calibration at the initial stage to obtain the user-specific parameters.…”

“…The detailed explanations are as follows. Figure 6 shows a conventional camera optical model, 23 where Z represents the distance between the camera lens and the object. V is the Z distance between the image plane and the camera lens.…”

“…W and w are the object sizes in the scene and image plane, respectively, and f is a camera focal point. According to a conventional camera optical model, 23 we can obtain the relationship among Z, and V, W, w as shown in Eq. (3):…”

A fast autofocus sharpness function of microvision system based on the Robert function and Gauss fitting

“…A measure of image sharpness, or sharpness function, is extracted from captured images at different lens positions. The in‐focus position then is obtained by locating the peak value of the sharpness function (Lee Kang & Cha, ; Xiao, Han, Zhu, Wang, & Xu, ). Different sharpness functions have been introduced for passive autofocus.…”

New method for face gaze detection in smart television

“…[1][2][3] Other research projects have proposed methods based on eye gaze-detection technology using near-infrared (NIR) light cameras and illuminators. [4][5][6][7] However, the devices based on additional NIR cameras and illuminators are difficult to use with smart TVs owing to their large size and high cost, and using conventional (visible light) web cameras can resolve these drawbacks for gaze-detection technology. Previously, Nguyen et al 8 used a conventional (visible light) web camera for gaze detection with a smart TV.…”