Limits...
Robust Eye Center Localization through Face Alignment and Invariant Isocentric Patterns.

Pang Z, Wei C, Teng D, Chen D, Tan H - PLoS ONE (2015)

Bottom Line: In addition, we apply our method to non-frontal faces to test its robustness and accuracy, which are essential in gaze estimation but have seldom been mentioned in previous works.Through extensive experimentation, we show that the proposed method can achieve a significant improvement in accuracy and robustness over state-of-the-art techniques, with our method ranking second in terms of accuracy.According to our implementation on a PC with a Xeon 2.5Ghz CPU, the frame rate of the eye tracking process can achieve 38 Hz.

View Article: PubMed Central - PubMed

Affiliation: School of Physics and Engineering, Sun Yat-Sen University, Guangzhou, China.

ABSTRACT
The localization of eye centers is a very useful cue for numerous applications like face recognition, facial expression recognition, and the early screening of neurological pathologies. Several methods relying on available light for accurate eye-center localization have been exploited. However, despite the considerable improvements that eye-center localization systems have undergone in recent years, only few of these developments deal with the challenges posed by the profile (non-frontal face). In this paper, we first use the explicit shape regression method to obtain the rough location of the eye centers. Because this method extracts global information from the human face, it is robust against any changes in the eye region. We exploit this robustness and utilize it as a constraint. To locate the eye centers accurately, we employ isophote curvature features, the accuracy of which has been demonstrated in a previous study. By applying these features, we obtain a series of eye-center locations which are candidates for the actual position of the eye-center. Among these locations, the estimated locations which minimize the reconstruction error between the two methods mentioned above are taken as the closest approximation for the eye centers locations. Therefore, we combine explicit shape regression and isophote curvature feature analysis to achieve robustness and accuracy, respectively. In practical experiments, we use BioID and FERET datasets to test our approach to obtaining an accurate eye-center location while retaining robustness against changes in scale and pose. In addition, we apply our method to non-frontal faces to test its robustness and accuracy, which are essential in gaze estimation but have seldom been mentioned in previous works. Through extensive experimentation, we show that the proposed method can achieve a significant improvement in accuracy and robustness over state-of-the-art techniques, with our method ranking second in terms of accuracy. According to our implementation on a PC with a Xeon 2.5Ghz CPU, the frame rate of the eye tracking process can achieve 38 Hz.

No MeSH data available.


Related in: MedlinePlus

Accuracy achieved for varying kernel size.The center map is blurred before the maximum isocenters (MICs) are selected using a mean filter. The accuracy changes for varying kernel sizes.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4591357&req=5

pone.0139098.g006: Accuracy achieved for varying kernel size.The center map is blurred before the maximum isocenters (MICs) are selected using a mean filter. The accuracy changes for varying kernel sizes.

Mentions: Slight errors could cause inaccurate displacement vectors and, as a result, the distribution of the center map might not actually represent the most likely eye-center location. To achieve robustness, we blur the center map before we select the MICs. In comparison with the findings of [18] and [28], we found that a mean filter with size (6, 6) could achieve better results than a Gaussian kernel, although the optimal size should be determined by the image scale. After a number of experiments, we found this size to have good performance, if the scale was ignored. We illustrated our results using the BioID database in Fig 6.


Robust Eye Center Localization through Face Alignment and Invariant Isocentric Patterns.

Pang Z, Wei C, Teng D, Chen D, Tan H - PLoS ONE (2015)

Accuracy achieved for varying kernel size.The center map is blurred before the maximum isocenters (MICs) are selected using a mean filter. The accuracy changes for varying kernel sizes.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4591357&req=5

pone.0139098.g006: Accuracy achieved for varying kernel size.The center map is blurred before the maximum isocenters (MICs) are selected using a mean filter. The accuracy changes for varying kernel sizes.
Mentions: Slight errors could cause inaccurate displacement vectors and, as a result, the distribution of the center map might not actually represent the most likely eye-center location. To achieve robustness, we blur the center map before we select the MICs. In comparison with the findings of [18] and [28], we found that a mean filter with size (6, 6) could achieve better results than a Gaussian kernel, although the optimal size should be determined by the image scale. After a number of experiments, we found this size to have good performance, if the scale was ignored. We illustrated our results using the BioID database in Fig 6.

Bottom Line: In addition, we apply our method to non-frontal faces to test its robustness and accuracy, which are essential in gaze estimation but have seldom been mentioned in previous works.Through extensive experimentation, we show that the proposed method can achieve a significant improvement in accuracy and robustness over state-of-the-art techniques, with our method ranking second in terms of accuracy.According to our implementation on a PC with a Xeon 2.5Ghz CPU, the frame rate of the eye tracking process can achieve 38 Hz.

View Article: PubMed Central - PubMed

Affiliation: School of Physics and Engineering, Sun Yat-Sen University, Guangzhou, China.

ABSTRACT
The localization of eye centers is a very useful cue for numerous applications like face recognition, facial expression recognition, and the early screening of neurological pathologies. Several methods relying on available light for accurate eye-center localization have been exploited. However, despite the considerable improvements that eye-center localization systems have undergone in recent years, only few of these developments deal with the challenges posed by the profile (non-frontal face). In this paper, we first use the explicit shape regression method to obtain the rough location of the eye centers. Because this method extracts global information from the human face, it is robust against any changes in the eye region. We exploit this robustness and utilize it as a constraint. To locate the eye centers accurately, we employ isophote curvature features, the accuracy of which has been demonstrated in a previous study. By applying these features, we obtain a series of eye-center locations which are candidates for the actual position of the eye-center. Among these locations, the estimated locations which minimize the reconstruction error between the two methods mentioned above are taken as the closest approximation for the eye centers locations. Therefore, we combine explicit shape regression and isophote curvature feature analysis to achieve robustness and accuracy, respectively. In practical experiments, we use BioID and FERET datasets to test our approach to obtaining an accurate eye-center location while retaining robustness against changes in scale and pose. In addition, we apply our method to non-frontal faces to test its robustness and accuracy, which are essential in gaze estimation but have seldom been mentioned in previous works. Through extensive experimentation, we show that the proposed method can achieve a significant improvement in accuracy and robustness over state-of-the-art techniques, with our method ranking second in terms of accuracy. According to our implementation on a PC with a Xeon 2.5Ghz CPU, the frame rate of the eye tracking process can achieve 38 Hz.

No MeSH data available.


Related in: MedlinePlus