What Are the Different Types of Ocular Movements?

Eye movement testing is the process of monitoring eye movements and gaze directions of a user when looking at a specific target, and performing related analysis through gaze tracking technology. Need to use eye tracker and related software during the process. In the early days, people mainly used photographs, film photography, and other methods to record eye movements. Now, advanced tools such as eye trackers can be used to obtain more accurate records.

Eye Movement Experiment

The principle of the eye movement experiment is mainly: the current popular gaze tracking technology is mainly based on "non-invasive" technology of eye video analysis (VOG, Video oculographic). Point, and the sensor is not in direct contact with the user. Some instruments using this technology allow users to move freely, such as Tobii's X6

The widely used method in the gaze tracking technology is called the pupil center cornea reflection technique, and the feature that remains unchanged during the eye movement process is the Purchin spot on the outer surface of the cornea Purkinje image)-A bright spot on the cornea of the eye, produced by the corneal reflection of light entering the pupil on the outer surface of the cornea.

Because the position of the camera is fixed, the position of the screen (light source) is fixed, and the position of the center of the eyeball is unchanged (assuming the eyeball is spherical and the head is not moving), the absolute position of the Purchin spot does not change with the rotation of the eyeball (in fact, the head Small amplitude motion can also be calculated by corneal reflection). But its position relative to the pupil and eyeball is constantly changing-for example, when you look at the camera, the Purchin spot is between your pupils; and when you look up, the Purchin spot is in your Under the pupil. In this way, as long as the pupils and Purkin's spots on the eye image are located in real time, and the corneal reflection vector is calculated, the geometric model can be used to estimate the direction of the user's line of sight. Based on the relationship between the characteristics of the user's eyes and the content displayed on the computer screen during the previous calibration process (that is, allowing the user to look at a specific point on the computer screen), the instrument can determine what the user is looking at on the screen. .

Positioning the center of the pupil is a key step in eye tracking technology, but one problem is that compared to the extremely obvious dividing line between the iris and the eye white, the dividing line between the pupil and the iris is not so clear, especially It is our dark eyes and yellow skin. Therefore, in order to improve the accuracy of this step, the researchers also designed a "bright and dark pupil difference scheme", that is, alternately emitting near-infrared light to the human eye with light sources with different orientations, and then in every two adjacent images, Obtain the user's bright pupil (bright pupil) and dim pupil (dark pupil), respectively, and superimpose the difference to "pull out" the pupil more clearly, and then calculate the parameters such as the center of mass and shape of the pupil.

Whether the "bright pupil" or "dark pupil" is captured depends on whether the camera is in line with the light source. If the camera and the light source are on the same line, the pupils captured by the camera are illuminated by light, that is, "bright pupil". This is similar to the principle of "red eye" appearing in the photo when the camera's flash is directly facing the subject when taking a picture. It's irrelevant. Secondly, the pupils will become smaller under strong light, otherwise they won't be blinded). If the two are not in line, the normal "dark pupil" will be captured. Therefore, there are two sets of near-infrared light sources on the eye tracker that support bright and dark pupil tracking.

The reason why near-infrared is used is because human eyes cannot detect it, so as not to dazzle and affect users. These beams are weak. As long as the researcher arranges the user to sit at the distance indicated on the eye tracker's instructions (for example, 60cm or more from the eye tracker), the user will not be exposed to radioactivity even if he / she is in front of the eye tracker for 8 hours.

The gaze tracking technology corresponding to the aforementioned "non-intrusive" technology requires that the user directly contact the sensors on the test equipment. For example, in the early eye movement test, something like a hard contact lens was stuffed into the tester's eyes, and the magnetic field that changes with the movement of the eye was monitored to know where you were looking or around the tester's eyes. Stick the electrodes and monitor the potential changes. These methods sound a bit scary and troublesome to operate, but the data obtained is relatively accurate.

So how accurate can ordinary commercial eye movement tests be? This depends on the specific parameters of the eye tracker for testing. There are two dimensions of space and time: the related parameters in space are accuracy, drift and screen size, and the parameter in time is the sampling rate (delay). For example: Tobii X120's accuracy is 0.5 degrees, and the drift with time is within 0.3 degrees. If the user calculates the distance from the screen 60cm, the offset is about 0.13mm; its sampling rate is 120Hz, and the delay is 17ms. Because the pupil can be counted every two frames. However, some researchers have found that the position deviation in the actual test is much larger than the value calculated here, which may be related to the user moving the head or the calibration problem. If you use the tobii series as a reading test, you may not be able to accurately locate which line of the user is looking at the interface. Therefore, in the analysis, you should avoid too much confidence in the gaze points given in the results.

At the same time, when testing, you should also try to follow the experimental specifications. Current commercial eye trackers can generally compensate for head movements, but even if the eye tracker allows users to move freely, there is a specified range of head movements. For example, the Tobii X60 and T60 models have a head movement range of 44 × 22 × 30cm. (Length, width, and height), while X120 and T120 have high frequencies and a smaller head movement range of 30 × 22 × 30cm (length, width, and height). The user should ensure that the head movement range is within this range. During the calibration, the user should be allowed to move the head within the specified range, and take the head movement into consideration during the calibration phase.