Description:TITLE OF THE INVENTION

“PUPIL DETECTION USING AN EVENT-BASED SENSOR”

Technical Field
[0001] The invention relates generally to eye-tracking techniques using pupil de-tection methods.
Background
[0002] Eye-tracking may be implemented in various devices such as virtual or augmented reality glasses or goggles, camera viewfinders, etc. Eye-tracking is often based on pupil detection with vision sensors and specific illumination condi-tions, usually with infrared sources that do not disturb the normal vision.
[0003] A "bright pupil" detection technique generally involves illuminating the retina through the pupil with a source placed near the optical axis of the camera, or sufficiently far in front of the eye that it appears to be on the optical axis. Under such conditions, the pupil appears to be very bright relative to its surroundings and is relatively easy to detect through contrast differences. Such techniques are not suited to devices placed close to the eye (goggles, glasses, viewfinders), since the infrared source cannot be placed close enough to the optical axis with respect to its distance to the pupil.
[0004] A "dark pupil" detection technique is better suited to devices close to the eye. It involves illuminating the eye from the side, whereby the pupil absorbs al-most all light and appears much darker than its surroundings.
[0005] Detection is usually carried out with a frame-based camera and an image analysis performed on the frames produced by the camera. To save power, the eye is illuminated intermittently or flashed at the rate of the frames. The technique however remains power-hungry, because it involves performing an analysis on a significant amount of frame data. Moreover, frame-based sensors cannot detect movements faster than the frame rate, and increasing the frame rate to capture faster movements would increase the required computation power.
[0006] The use of event-based sensors in pupil detection applications overcomes the lack of speed of frame-based sensors, but the amount of data to process re-mains large, as shown below.
[0007] Figures 1A and 1B illustrate two phases of a dark pupil detection method conventionally using an event-based sensor and an intermittent illumination of the eye. Since there is no frame rate limitation, the illumination flash frequency may approach the reaction speed limit of the event-based sensor.
[0008] Figure 1A illustrates the appearance of the eye as perceived by the sensor in the two illumination phases, and figure 1B illustrates a corresponding brightness variation of the pupil and the rest of the eye. On the left, there is no illumination, and the eye and pupil are uniformly dark (black). On the right, the illumination is turned on. The pupil remains dark or is barely brighter, whereas the rest of the eye becomes significantly brighter (white).
[0009] It is thus apparent that a large number of pixels around the pupil, in any case more pixels than in the pupil area, see a significant contrast change. As a consequence, an event-based sensor will trigger a significant number of events that require a corresponding amount of processing power. In essence, although the use of an event-based sensor for dark pupil detection improves the speed of detection, the amount of processing power remains significant.
Summary
[0010] A general method is provided for detecting a pupil of an eye with an event-based vision sensor, comprising the steps of directing a first light source towards the eye; observing the eye with the event-based sensor; and intermittently submitting the sensor to a uniform background signal.
[0011] The method may further comprise the steps of setting a level of the back-ground signal such that a relative brightness change sensed for the pupil is greater than a relative brightness change sensed for the rest of the eye; and setting a detection threshold of pixels of the sensor such that the relative brightness change of the pupil triggers an event, and the relative brightness change of the rest of the eye does not trigger an event.
[0012] The method may further comprise a step of submitting the sensor to the background signal during phases where the eye is illuminated with the first light source.
[0013] The method may alternatively comprise a step of illuminating the eye in phase opposition with the background signal.
[0014] The background signal may be increased gradually until events are trig-gered by the sensor.
[0015] The background signal may be obtained by directing a second light source towards the sensor.
[0016] The background signal may alternatively be an electrical signal added in-side pixel circuits of the sensor that has the effect of a uniform illumination of the sensor.
[0017] The electrical signal may be a constant current added to photocurrents generated in the pixels.
[0018] The sensor may be fitted with a lens, and the second light source be ar-ranged such that light partially reflected by the lens penetrates a pupil of the eye.
[0019] The first light source may be a display device viewed by the eye.
[0020] A general method is also provided for detecting a pupil of an eye with a vision sensor fitted with a lens, comprising the steps of observing the eye with the sensor; directing a light source towards the lens of the sensor such that i) the sen-sor is submitted to a uniform background signal and ii) light partially reflected by the lens penetrates a pupil of the eye; and performing a bright pupil detection on the image perceived by the sensor.
Brief Description of the Drawings
[0021] Embodiments will be exposed in the following description provided for exemplary purposes only, in relation to the appended drawings, in which:
[0022] Figures 1A and 1B illustrate two phases of a dark pupil detection method conventionally using an event-based sensor and an intermittent illumination of the eye;
[0023] Figures 2A-2C illustrate two phases of a dark pupil detection method using an event-based sensor according to a first embodiment;
[0024] Figures 3A and 3B illustrate absolute and relative logarithmic contrast variation graphs used in the embodiment of figures 2A-2C;
[0025] Figures 4A and 4B illustrate two phases of a dark pupil detection method using an event-based sensor according to a second embodiment; and
[0026] Figure 5 illustrates partial pixel circuits modified according to an embodi-ment to add a uniform background signal.
Detailed Description
[0027] As previously mentioned, when performing eye tracking with an event-based sensor using a dark pupil method and intermittent illumination of the eye, the number of events triggered at each illumination flash, i.e., the amount of data to process, is significant. This is because a significant portion of the eye sees a brightness change that triggers events, whereas the region of interest, the pupil, sees hardly any change and triggers no events.
[0028] It is proposed herein to illuminate the eye under specific conditions that reverse the regions that see a significant brightness change, i.e., such that the pupil sees the most significant change, whereas the rest of the eye sees a change insuffi-cient to trigger events.
[0029] Figures 2A-2C illustrate, respectively, a first embodiment of a two-phase illumination configuration, a corresponding eye appearance variation as perceived by the sensor, and a corresponding brightness variation of the eye and pupil re-gions.
[0030] In a first phase, on the left, the eye 10 is illuminated by a source 12, pref-erably infrared, placed, for instance, between the eye and the sensor 14, and offset with respect to an optical axis between the eye and the sensor, in a dark pupil con-figuration. The sensor includes an array of event-based pixels and is usually fitted with a lens, not shown. Both the sensor and the source 12 may be located on a rim of glasses or goggles worn by the user. In other configurations, the source 12 may be located in a plane behind the sensor and may even be the display device usually included in goggles.
[0031] The illumination intensity of source 12 need not be as bright as required in a conventional situation. As shown on the left of figure 2B, the pupil remains dark, and the surrounding region is gray instead of white. Exemplary brightness levels are shown on the left of figure 2C, on a linear amplitude scale - the pupil brightness is barely above zero, whereas the rest of the eye region has a brightness at a given level, shown as lower than that of the conventional illumination phase of figure 1B, as an example, to save power.
[0032] In a second illumination phase, on the right, the sensor 14 is submitted to a uniform background signal. This background signal may be produced, as shown as an example, by a second illumination source 16 directed towards the sensor. The source 16 may be placed outside a sensor housing incorporating the lens and be directed towards the lens. The source 16 could also be placed inside the sensor housing and could thus be in the visible spectrum since it wouldn't interfere with the user's vision. Alternatively, the background signal may be an electrical signal added inside the pixel circuits that would produce the same effect as a uniform illumination of the sensor, for example a constant bias current added to the photo-current of each pixel in the sensor array, as described below in relation to figure 5.
[0033] This background signal causes all regions viewed by the sensor to see their brightness increase by a same step, as illustrated on the right of figure 2C. As shown on the right of figure 2B, for purposes of illustration, the pupil changes from black to gray, whereas the rest of the eye changes from light gray to white.
[0034] This mechanism finds its relevance in the way the event-based pixels are generally designed, i.e., brightness is measured on a logarithmic scale and changes on such a logarithmic scale correspond to ratios of change. The ratio of brightness change for the pupil happens to be much larger than the ratio of brightness change for the rest of the eye with the shown illumination method.
[0035] Figure 3A shows the brightness evolutions of figure 2C on a logarithmic amplitude scale. Here, the step seen by the pupil is significantly larger than the step seen by the rest of the eye, because the brightness of the pupil increases from nearly zero, whereas the brightness of the rest of the eye increases between two values that are relatively far from zero.
[0036] Figure 3B illustrates comparisons carried out by event-based pixels corre-sponding to the pupil region and the rest of the eye region. Each pixel is usually designed to compare the logarithmic amplitude changes to a threshold, i.e., the difference between the current logarithmic amplitude and the previous logarithmic amplitude, which reflects the ratio of change. The threshold Th may be set, as shown, such that the change for the pupil region exceeds the threshold, whereas the change for the rest of the eye region stays below the threshold. This configura-tion provides a certain latitude for the choice of the intensities of the sources 12 and 16 to lower the power consumption.
[0037] In one example, the source 12 illuminating the eye may be on continuously, but its intensity may be significantly less than required in the conventional setup of figures 1A, 1B. The intensity of intermittent source 16 may also be set at a relatively low level.
[0038] Alternatively, the two shown illumination phases may be interleaved with stand-by phases where both sources are off and the sensor is disabled.
[0039] To optimize power consumption, the intensity of the background illumina-tion source 16 may be increased gradually instead of in a single large step, until events that correspond to the pupil region are triggered.
[0040] Figures 4A and 4B illustrate two phases of a dark pupil detection method using an event-based sensor according to a second embodiment. In figure 4A, a layout similar to that of figure 2A is used for the sensor and illumination sources. However, instead of continuously illuminating the eye with source 12 and flashing source 16 to intermittently add the background illumination signal, sources 12 and 16 are flashed in phase opposition. On the left side, source 12 illuminates the eye while source 16 is off, creating an image similar to that on the left of figure 2B, revealing a dark pupil and a brighter surrounding region, shown as gray. On the right side, source 16 illuminates the sensor while source 12 is off, creating an im-age having a uniform brightness level shown as gray.
[0041] In theory, the brightness levels of the rest of the eye region in both phases could be made substantially equal by adjusting the intensities of the sources 12, 16, whereby no difference would be measurable for this region, thus increasing the discrimination tolerance for detecting the pupil. This may however be difficult in practice, since the two phases correspond to different illumination conditions (the eye on one side and the sensor on the other side). In any case, like for the embodiment of figures 2A-2C, the ratio of change for the pupil region remains substantially larger than the ratio of change for the rest of the eye region.
[0042] Figure 5 illustrates pixel circuits of an array modified according to an em-bodiment to add a uniform background signal within the sensor array. Three pixels Px1-Px3 are partially illustrated, where pixel Px3 is illustrated in more detail. Each pixel of an event-based sensor generally includes a photodiode PD having its anode connected to ground. The photocurrent generated by the diode flows from a high supply line into the cathode of the photodiode through a transimpedance amplifier TIA that usually has a logarithmic response. The output of the TIA is processed by circuitry, not shown, to produce events in any known manner.
[0043] A respective N-MOS transistor M1-M3 is connected between the cathode of each photodiode and ground. The transistors M1-M3 form outputs of current mirrors sharing a common diode-connected input N-MOS transistor M0. A current source Ibgr is connected to inject a background reference current Ibgr in the diode-connected transistor M0, which current is copied in each of transistors M1-M3 and adds to the photocurrent measured by the TIA. The current Ibgr is adjust-ed to produce the desired background signal intensity.
[0044] An N-MOS transistor Mb connects the gates of transistors M0-M3 to ground and is controlled by an active-low background enable signal /BGR. Signal /BGR may thus switch the sensor between a normal mode and a simulated back-ground illuminated mode.
[0045] As previously mentioned, one embodiment of the background signal source 16 is an illumination source directed towards a lens of the sensor, according to the configuration shown at the right of figure 2A. The lens will partially reflect the light of source 16 along a cone. If the source 16 is sufficiently close to the optical axis or the lens has a sufficient curvature, the reflected cone may intersect the pupil, whereby light penetrates the pupil along the optical axis, putting the pupil in bright pupil detection conditions.
[0046] Such a configuration further increases the brightness variation of the pupil. In other words, the linear amplitude step for the pupil in figure 2C is greater than the step for the rest of the eye. As an illustration, at the right of figures 2B and 4B, where the source 16 is turned on, the pupil will be a lighter shade of gray.
[0047] This particular configuration may be simplified, by using only the source 16, to operate a bright pupil detection technique. The source 16 is then permanent-ly turned on at a relatively low intensity, generating a uniform gray background on the sensor. The source 12 is absent or permanently off. The light partially reflected by the lens illuminates the eye and pupil, which create an image with a bright pupil superimposed on the background gray level of the sensor. The pupil region, although it may not be as bright as in the situation of an appropriately exposed eye, will still be noticeably brighter than the rest of the perceived image. The pupil will thus be relatively easy to track with any type of sensor (event-based or frame-based), offering the advantage of bright pupil sensing techniques, i.e., avoiding the false positives a dark iris may cause with dark pupil sensing techniques.
, C , Claims:CLAIMS

We Claim:

1. A method for detecting a pupil of an eye with an event-based vision sensor, comprising the steps of:
directing a first light source (12) towards the eye (10);
observing the eye with the event-based sensor (14); and
intermittently submitting the sensor to a uniform background signal (16) causing all regions viewed by the sensor to see their brightness increase by a same step.
2. The method of claim 1, comprising the steps of:
setting a level of the background signal (16) such that a relative bright-ness change sensed for the pupil is greater than a relative brightness change sensed for the rest of the eye; and
setting a detection threshold of pixels of the sensor such that the relative brightness change of the pupil triggers an event, and the relative bright-ness change of the rest of the eye does not trigger an event.
3. The method of claim 2, comprising a step of submitting the sensor to the background signal during phases where the eye is illuminated with the first light source.
4. The method of claim 2, comprising a step of illuminating the eye in phase opposition with the background signal.
5. The method of claim 1, wherein the background signal is increased gradually until events are triggered by the sensor.
6. The method of claim 1, wherein the background signal is obtained by direct-ing a second light source (16) towards the sensor (14).
7. The method of claim 1, wherein the background signal is an electrical signal added inside pixel circuits of the sensor that has the effect of a uniform illu-mination of the sensor.
8. The method of claim 7, wherein the electrical signal is a constant current add-ed to photocurrents generated in the pixels.
9. The method of claim 6, wherein the sensor is fitted with a lens, and the second light source (16) is arranged such that light partially reflected by the lens penetrates a pupil of the eye.
10. The method of claim 1, wherein the first light source is a display device viewed by the eye.
11. A method for detecting a pupil of an eye with a vision sensor fitted with a lens, comprising the steps of:
observing the eye with the sensor (14);
directing a light source (16) towards the lens of the sensor such that i) the sensor is submitted to a uniform background signal and ii) light partially reflected by the lens penetrates a pupil of the eye; and
performing a bright pupil detection on the image perceived by the sensor.