Skip to content Skip to sidebar Skip to footer

Diego González Aguilera research stay at Visual Attention Lab (Boston, USA)

Estancia investigadora de Diego González Aguilera Visual Attention Lab

Diego Gonzalez Aguilera was invited by the Visual Attention Lab (Dr. Marc Pomplun) in the University of Massachussetts (Boston, USA) for carrying out a research stay during 3 months (July-October, 2015). The research was focused on advancing methods and algorithms for connecting visual attention and photogrammetry. In particular, a connection between photogrammetry and the visual attention devices (Eye Tracker) was proposed. More precisely, we took advantage of the algorithms and approaches of photogrammetry to integrate them in the visual attention workflow. To this end, we exploited the General Method of Photogrammetry (GMP) making a parallelism between its main steps and those required by the eye trackers. As a result we aimed to improve the precision and reliability of the eye tracker measurements: from its calibration and determination of eye pose to the computation of visual attention object’s coordinates (gaze position). We also tried to improve the current applications performed on the laboratory and under a 2D context, extending towards the visual attention in a 3D environment.

Understand the basis of photogrammetry and its principles entails to advance in the knowledge and comprehension of the human vision and how we are able to perceive the reality through eyes in three dimensions. The acquisition of reality through the vision system involves that the light rays (coming from natural or artificial light source) reach and pass through the eyes, which are sensible to the electromagnetic radiation of specific wavelengths (i.e. visible spectrum). In particular, when the light comes into the eye passes through the cornea, the pupil and the crystalline, arriving to the retina where the electromagnetic energy is converted into nerve impulses that can be used by the brain. These impulses leave the eye through the optic nerve. In photogrammetric terms, the region most important of the eye is located around a small region in the retina well-known as fovea. At this place all the light coming from the field of view is focused what is equivalent to the field of view of the camera. However, it should be remarked that although the field of view of each eye separately is around 120º-200º and 130º for the overlap area of both eyes (i.e. equivalent to the fisheye objectives), the retina’s effective region is only between 40-60º (even less if we are focusing on small objects at very close distances). So, the rest of the field of view of the eye is only useful for perceiving large objects or movements. Likewise, for the digital cameras this field of view is dependent of two main geometric parameters (focal length and sensor size) which are perfectly defined, whereas for the human vision this is complicated since the eye is curved and the level of detail decreases as we move away from the fovea’s center.

The next figure outlines the relationship between the geometry of the human eye with the geometry of the camera lens.

Estancia investigadora de Diego González Aguilera Visual Attention Lab

Geometric eye model. (a) Outer view and (b) cross-section of the human eye with important components marked red. (c) Geometric eye model with components involved in the eye pose estimation.

The next figure outlines some of the experiments performed trying to improve eye-tracker calibration and thus to improve eye-tracker measurements.

EyeLink 1000 durante los tests experimentales desarrollados.