Delay reduction for eye tracker of naked eye stereoscopic display device

Abstract

The present invention relates to a method for tracking facial features of a viewer of a naked eye stereoscopic display device, the method comprising: obtaining first sequential data of positions of the facial features relative to the naked eye stereoscopic display device; and obtaining second sequence data of the speed and / or acceleration of the facial feature relative to the naked eye stereoscopic display equipment. The first sequential data is obtained with a first delay and the second sequential data is obtained with a second delay that is lower than the first delay. Finally, sensor fusion is performed to combine the first sequential data and the second sequential data to generate final sequential data of the position of the facial feature relative to the naked eye stereoscopic display device. In this manner, the final sequential data is obtained with a final delay that is lower than the first delay. When a viewer of the naked-eye stereoscopic display device moves and rotates his head relative to the naked-eye stereoscopic display device, improved stereoscopic images can be presented to the viewer and / or better virtual reality experience can be obtained.

Description

Technical Field

[0001] The present invention relates to a method for tracking facial features of a viewer of a naked-eye stereoscopic display device, a tracking system for tracking facial features of a viewer of a naked-eye stereoscopic display device, and a naked-eye stereoscopic display device comprising such a tracking system. Background Art

[0002] An eye tracker is a device used to measure a person's real-time eye position and real-time eye movement. It typically uses a photodiode or a camera to capture images of the eyes. The eyes are then typically identified through facial landmark detection, followed by signal or image processing methods to estimate eye movement.

[0003] An important application of eye trackers is to use them as input devices for human-computer interaction. For example, in a naked-eye stereoscopic display system, an eye tracker provides the real-time position of the viewer's eyes so that the viewer can perceive realistic three-dimensional images even if he moves relative to the display. This is achieved by providing a left-eye image exclusively for his left eye and a right-eye image exclusively for his right eye. In addition, the eye position can be used to adjust the perspective view of the three-dimensional scene to the position of the viewer's head / eyes relative to the display (the so-called "look-around effect").

[0004] Like almost all electronic measurement devices, autostereoscopic display systems suffer from latency, which is generally understood as the time delay between user input and system response. In the autostereoscopic display system setting, this basically means that there is a delay between the viewer's head / eye movement (user input) and the display content adjusting in response to that movement (system response). If the latency exceeds a critical threshold, user performance and user experience will be affected. This is usually related to crosstalk and surround effect interference.

[0005] Eye trackers are a factor in a large portion of the latency observed in autostereoscopic displays. For example, a typical total system latency of 85 milliseconds may include 43 milliseconds of eye tracker latency. Therefore, solving the latency problem by reducing the latency caused by the eye tracker alone would be a significant advance. Summary of the invention

[0006] Therefore, it is an object of the present invention to reduce the delays induced in an eye tracker.More generally, it is an object of the present invention to improve the viewing experience of a viewer of a naked-eye stereoscopic display.

[0007] It has now been found that one or more of these objectives can be achieved by using additional measuring devices.

[0008] Therefore, the present invention relates to a method for tracking facial features of a viewer of a naked-eye stereoscopic display device, the method comprising acquiring first sequential data of the position of the facial features relative to the naked-eye stereoscopic display device, wherein the first sequential data is obtained with a first delay, wherein the method further comprises

[0009] - obtaining second sequential data of the velocity and / or acceleration of the facial feature relative to the naked eye stereoscopic display device, wherein the second sequential data is obtained with a second delay, and the second delay is lower than the first delay;

[0010] - performing sensor fusion to combine the first sequential data and the second sequential data to generate final sequential data of the position of the facial features relative to the naked-eye stereoscopic display device, wherein the final sequential data is obtained with a final delay lower than the first delay.

[0011] The present invention also relates to a tracking system for tracking facial features of a viewer of a naked-eye stereoscopic display device, the tracking system comprising

[0012] - a tracking camera system configured to obtain first sequential data of a position of a facial feature relative to a naked-eye stereoscopic display device, wherein the first sequential data is obtained with a first delay;

[0013] - a secondary tracking system configured to obtain second sequential data of velocity and / or acceleration of facial features relative to the naked eye stereoscopic display device, wherein the second sequential data is obtained with a second delay, the second delay being lower than the first delay;

[0014] - a sensor fusion module configured to combine the first sequential data and the second sequential data to generate final sequential data of the position of the facial features relative to the naked-eye stereoscopic display device.

[0015] The present invention also relates to a naked-eye stereoscopic display device comprising the tracking system as described above. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 An exemplary flow chart of a method according to the invention is shown.

[0017] Figure 2 A first assembly (8) according to the invention is schematically shown.

[0018] Figure 3 A second assembly (8) according to the invention is schematically shown. DETAILED DESCRIPTION

[0019] The elements shown in the figures are shown for simplicity and clarity and are not necessarily drawn to scale. For example, the size of some elements in the figures may be exaggerated relative to other elements to help better understand the various exemplary embodiments of the present invention. In particular, the relative sizes of the viewer and the naked eye stereoscopic display device cannot be derived from the figures.

[0020] Furthermore, the terms "first", "second", and the like in the description and the claims, if any, are generally used to distinguish between similar elements and not necessarily to describe a sequential or chronological order.

[0021] In the context of the present invention, the term "viewer" refers to a person who is able to consume (particularly watch) content presented by a naked-eye stereoscopic display device. Throughout the text, references to the viewer will use masculine terms such as "he", "they" or "his". This is only for the purpose of clarity and brevity, and it should be understood that feminine terms such as "she" and "her" are equally applicable.

[0022] In the context of the present invention, velocity includes linear velocity and rotational velocity; acceleration includes linear acceleration. In addition, it is understood that the position and orientation of an object (e.g., a device, a body part, or a facial feature) can be derived from the velocity and acceleration of the object and the initial known position and orientation of the object.

[0023] In the context of the present invention, tracking of facial features refers to tracking of the position of facial features over time. The acquired position is related to the position of a tracking system performing the tracking and ultimately to the position of the naked-eye stereoscopic display device.

[0024] In the context of this invention, latency refers to the time delay between user input and system response, which is caused by the limited speed at which any physical interaction can propagate and the speed at which information can be transmitted.

[0025] In the context of the present invention, sequential data refers to data that contains elements that are ordered in time. This involves measuring a quantity over time at certain time intervals (time series).

[0026] In conventional methods of tracking facial features (e.g., eyes), one or more cameras are typically used to obtain positional data of the facial features and track their positions. Such cameras then place the viewer's head within the camera's field of view. In the method of the present invention, such conventional tracking methods are supplemented by a second tracking method that is based on velocity and / or acceleration rather than the position of the facial features.

[0027] Therefore, in the method of the present invention, at least two types of sequential data of facial features are captured. First sequential data of the position of the facial features relative to the naked-eye stereoscopic display device is obtained, and second sequential data of the speed and / or acceleration of the facial features relative to the naked-eye stereoscopic display device is obtained. Therefore, the method of the present invention requires that its relative speed and / or acceleration be used as input for determining the second sequential data. Finally, the first sequential data and the second sequential data are combined using sensor fusion to generate final sequential data of the position of the facial features relative to the naked-eye stereoscopic display device. These final sequential data form the actual output data of a tracker (e.g., an eye tracker) in which the method of the present invention is performed; and the final sequential data form the input of the weaving of stereoscopic images, and, if applicable, generate a stereoscopic image with a correct viewing angle from the viewer's perspective ("surrounding effect").

[0028] In the method of the present invention, the facial feature is usually an eye (or two eyes). However, the facial feature can also be a facial feature selected from the group of eyes, nose, ears, head and mouth.

[0029] According to conventional techniques, the first sequence data is preferably obtained using one or more cameras. These one or more cameras are usually configured to record facial features of viewers in the observation field of the one or more cameras. This also usually involves identifying facial features in the images recorded by the one or more cameras and determining their positions in the images.

[0030] The second order data is usually obtained by using a dedicated measuring device, which is capable of determining the velocity and / or acceleration of facial features relative to the naked eye stereoscopic display device. Such a dedicated measuring device may include a wearable device worn on the viewer's head. For example, the wearable device is selected from the group of earbuds, headphones, glasses, headbands, headwear, rubber bands and hats. Such a device is usually capable of determining its specific force (i.e., the force acting on it; acceleration). It can also determine its angular velocity and / or its direction. To this end, the dedicated measuring device is equipped with one or more accelerometers, and is usually also equipped with one or more measuring instruments selected from the group of gyroscopes and magnetometers. In the art, the measuring device including these instruments is referred to as an inertial measurement unit (IMU). The speed of the dedicated measuring device can be derived from the measured acceleration, which is performed after the initial calibration, in which the speed of the naked eye stereoscopic display device relative to the dedicated measuring device (e.g., head-mounted device) is set to a specific value (usually zero).

[0031] The head-mounted IMU preferably includes an accelerometer and a gyroscope, as this allows the IMU to distinguish between lateral and rotational movement of the head (particularly rotations characterized by pitch and yaw). This is important for determining head pose. Preferably, a magnetometer is also present to be able to distinguish between gravity and specific forces.

[0032] It is understandable that the data obtained by the wearable device represents the speed and / or acceleration of the wearable device and not necessarily the facial features, even taking into account their fixed relative positioning (it should be noted that this may be very inconvenient when the wearable device is placed exactly in or on the facial features, for example when the eyes are involved). In particular, in the case of head rotation, the speed and acceleration of the wearable device may deviate from the speed and acceleration of the facial features. Therefore, preferably, it is necessary to know the relative position of the wearable device and the facial features so that the speed and / or acceleration of the facial features themselves relative to the naked-eye stereoscopic display device can be obtained.

[0033] Figure 1 An exemplary flow chart of a method according to the invention is shown, the upper part of which starts with a setup (1) in which a viewer (2) is located in front of a naked-eye stereoscopic display device (3). Associated with the naked-eye stereoscopic display device (3) are a tracking camera system (4), a secondary tracking system (5) and a sensor fusion module (6). The viewer (2) wears the tracking system (5) as earbuds. The lower part of the flow chart shows that first sequence data and second sequence data are obtained separately from the setup (1). Here, the tracking camera system (4) provides the first sequence data and the secondary tracking system (5) provides the second sequence data. The two sequence data are then combined together by sensor fusion performed by the sensor fusion module (6) to produce final sequence data as output of the method.

[0034] In the method of the present invention, sequential data are obtained at a specific frequency, which will be denoted by the term "sampling rate" hereinafter. The first sequential data are typically obtained at a rate in the range of 10-100 Hz, in particular at a rate in the range of 25-75 Hz (e.g. 30 Hz, 40 Hz, 50 Hz or 60 Hz). The second sequential data are typically obtained at a higher sampling rate than the first sequential data, although it is also possible to obtain them at a lower sampling rate than the first sequential data. Of course, preferably, the sampling rate is as high as possible. The sampling rate of the second sequential data is, for example, in the range of 20-1000 Hz, in particular in the range of 50-500 Hz. In particular, the IMU is capable of measuring at a high sampling rate, for example, a rate in the range of 100-1000 Hz.

[0035] In some cases, the speed and / or acceleration of the naked eye stereoscopic display device itself must be considered when obtaining the second order data. In this case, the naked eye stereoscopic display device is generally configured to generate third order data about its own speed and / or acceleration. This is usually the case when the naked eye stereoscopic display device has no fixed position relative to the ground during use, such as when the device is a handheld (mobile) device, such as a mobile phone, a tablet computer, a laptop computer or a game console. After all, only by obtaining the absolute acceleration and velocity of the facial features, as well as the absolute acceleration and velocity of the naked eye stereoscopic display device, can their relative velocity and acceleration be obtained. Only when the naked eye stereoscopic display device is truly stationary, its own absolute acceleration does not need to be considered. However, when the viewer is in a moving car or boat, the acceleration of the naked eye stereoscopic display device is usually considered, because boats and cars are usually affected by acceleration when in use. In this case, the absolute acceleration of the naked eye stereoscopic display device is usually measured first for initial calibration, in which the velocity of the naked eye stereoscopic display device relative to the facial features is set to a specific value (usually zero). Typically, in this initial calibration, the direction of the naked eye stereoscopic display device relative to the facial features is also set to a specific value. This may also involve a preliminary estimate, for example based on an initial camera image.

[0036] It is also possible that the dedicated measuring device is not worn on the viewer's head, but rather a remote device placed at a certain distance from the head (so that its movement is independent of the movement of the head). For example, the measuring principle of such a remote device relies on the Doppler effect (in sound or light), time of flight or triangulation. For this purpose, the remote device may include a source of electromagnetic radiation capable of illuminating the head.

[0037] When the stereoscopic display device and the remote device are not accelerating (i.e., stationary relative to the ground or moving at a constant speed relative to the ground), the remote device only needs to measure the velocity and / or acceleration of the facial features relative to the remote device. This is the case, for example, when the stereoscopic display device and the remote device are placed or installed at a location in a room where the viewer views the stereoscopic display device. The remote device may also form an integral part of the stereoscopic display device.

[0038] When the naked eye stereoscopic display device and / or the remote device accelerates, it is necessary to compensate for this acceleration. This is usually done by measuring the acceleration of the naked eye stereoscopic display device and / or the remote device. This situation is applicable, for example, when the naked eye stereoscopic display device is a handheld mobile device or the viewer is in a car or a boat.

[0039] To this end, the remote device and / or the naked-eye stereoscopic display device is equipped with one or more accelerometers and is typically also equipped with one or more measuring instruments selected from the group of gyroscopes and magnetometers. Preferably, it is equipped with an IMU. The speed of the remote device relative to the naked-eye stereoscopic display device can be derived from the measured acceleration after an initial calibration in which the speed of the naked-eye stereoscopic display device relative to the remote device is set to a specific value (usually zero). However, preferably, the remote device is integrated with such a handheld device so that only the acceleration of the handheld device needs to be determined.

[0040] When using a remote device (i.e., a device not worn on the viewer's head), it is preferred that the naked eye stereoscopic display device and the remote device are not accelerated (i.e., they are stationary relative to the ground), because considering their acceleration requires considerable processor resources, and in addition, such processing must also avoid delays as much as possible. A setting to achieve this is, for example, a room dedicated to providing the viewer with a virtual reality experience that is as comfortable as possible. In this setting, there is no need to wear a wearable device on the head, and multiple remote devices are positioned to be fixed and accurately obtain second order data. In addition, the naked eye stereoscopic display device will also be located in a fixed position in the room.

[0041] Traditionally, the tracking of facial features (usually the eyes) is based on capturing images using a camera that forms part of the tracking system. The capture itself causes a delay (usually half the shutter speed), and the transmission and processing of the acquired frames also cause delays, which increase the latency. However, the present invention provides a method that can reduce the latency of the entire tracking system to a lower value than the latency caused by tracking based on a camera alone.

[0042] In the method of the present invention, first sequential data is usually obtained by such a camera-based tracking system, and the first sequential data represents the position of facial features relative to the naked eye stereoscopic display device. This occurs with a first delay. Second sequential data is obtained in a different way. This occurs with a second delay.

[0043] The second sequential data is not position data, but speed and / or acceleration data. These data can be used to predict the position of facial features in the near future. Therefore, in the method of the present invention, the first sequential data and the second sequential data are combined to generate final sequential data about the position of facial features relative to the naked eye stereoscopic display device. These data form the actual output of the tracking system (e.g., eye tracking system) implementing the method of the present invention. This process occurs with a final delay.

[0044] In the method of the present invention, advantageously, the final delay is lower than the first delay. In other words, by combining the first order data and the second order data, the delay of the first order data is effectively reduced. And, advantageously, the delay of the tracker is also reduced. Therefore, the method can be advantageously performed in a tracker of a naked-eye stereoscopic display device, thereby producing an improved stereoscopic image, in particular when the viewer moves and rotates his head relative to the naked-eye stereoscopic display device, providing the viewer with a better virtual reality experience; and in particular when the naked-eye stereoscopic display device is handheld and therefore subjected to intense motion (in particular rotation), providing the viewer with a better virtual reality experience.

[0045] In the method of the present invention, the naked-eye stereoscopic display device may be selected from the group consisting of a television, a (desktop) monitor, a cinema display system, a mobile phone, a tablet computer, a laptop computer and a game console.

[0046] The combination of the first and second order data is done through so-called sensor fusion. This concept is generally understood as the process of combining data from multiple sensors to overcome the limitations of a single sensor. This can produce more reliable information with less uncertainty.

[0047] In the present invention, sensor fusion overcomes the high latency of a first sensor by combining it with a second sensor that has lower accuracy but lower latency. Thus, this sensor fusion allows the system to combine the position accuracy of a slow sensor with the velocity and / or acceleration data of a fast system to provide accurate data with low latency.

[0048] There are a variety of commonly used sensor fusion techniques. The skilled person knows which techniques are suitable for application to the method of the present invention; and can determine how to apply these techniques through routine testing without inventive effort. For example, sensor fusion utilizes a class of algorithms called Kalman filters.

[0049] The method of the present invention can be advantageously used in a variety of applications, as described above, which generally reduces latency. This is particularly advantageous in the case of handheld naked-eye stereoscopic display devices. This is because such devices are susceptible to violent motion, especially rotation. Compared with large fixed devices, such devices cause the viewer to have more extreme positions, speeds and accelerations relative to the device. Such motion of handheld naked-eye stereoscopic display devices can easily cause distortion of the three-dimensional view because the latency traditionally associated with such devices cannot keep up with the violent motion of the device.

[0050] A particular advantage of using a head-mounted IMU in the method of the present invention is that the orientation of the head, and therefore the orientation of facial features, can be measured faster and more accurately. It is difficult to distinguish between lateral and rotational movements of the head (especially rotations characterized by pitch and head turn) when relying solely on a camera. Therefore, it is difficult to detect changes in head posture. However, a head-mounted IMU equipped with an accelerometer and a gyroscope (preferably also equipped with a magnetometer) can accurately measure head rotation. In addition, the latency of such measurements is lower than that of measurements made by a camera. Therefore, by applying the method according to the present invention, viewers who change their head posture can be tracked more accurately and with lower latency.

[0051] The present invention also relates to a tracking system (7) for tracking facial features of a viewer of a naked-eye stereoscopic display device (3), the tracking system (7) comprising

[0052] - a tracking camera system (4) configured to obtain first sequential data of the position of facial features relative to the naked eye stereoscopic display device (3), wherein the first sequential data is obtained with a first delay;

[0053] - a secondary tracking system (5) configured to obtain second sequential data of the velocity and / or acceleration of the facial features relative to the naked eye stereoscopic display device (3), wherein the second sequential data is obtained with a second delay, the second delay being lower than the first delay;

[0054] - A sensor fusion module (6) configured to combine the first sequential data and the second sequential data to generate final sequential data of the position of the facial features relative to the naked-eye stereoscopic display device (3).

[0055] In such an arrangement, the method of the present invention as described above can be implemented. The tracking camera system (4) is configured to acquire first sequential data, and the secondary tracking system (5) is configured to acquire second sequential data. Final sequential data is obtained with a final delay lower than the first delay.

[0056] In a preferred embodiment, the secondary tracking system (5) is designed to be worn on the viewer's head, the secondary tracking system (5) comprising one or more accelerometers and typically also one or more measuring instruments selected from the group of gyroscopes and magnetometers. This allows the measurement of the velocity and / or acceleration of a person's head, including rotation of the head, and ultimately the velocity and / or acceleration of specific facial features.

[0057] In another preferred embodiment, the secondary tracking system (5) is designed not to be worn on a person's head, but to measure the velocity and / or acceleration of facial features from an external position (i.e. a position not directly related or associated with the viewer's head). To this end, the secondary tracking system (5) typically includes a remote device configured to measure the velocity and / or acceleration of facial features relative to the remote device by irradiating the facial features with electromagnetic radiation or sound waves, for example by utilizing the Doppler effect (in light or sound), time of flight or triangulation. Such a remote device can be (rigidly) connected to the naked-eye stereoscopic display device (3), or can be present in a different position that is not affected by the movement of the naked-eye stereoscopic display device (3) and therefore also not affected by the movement of the viewer.

[0058] The invention also relates to a naked-eye stereoscopic display device (3) comprising a tracking system (7) as described above.

[0059] The present invention also relates to a component (8) comprising the naked-eye stereoscopic display device (3) and the tracking system (7) as described above.

[0060] Figure 2 A first component (8) according to the present invention is schematically shown. The first component consists of a naked-eye stereoscopic display device (3) and a tracking system (7). The tracking system consists of a tracking camera system (4), a secondary tracking system (5) and a sensor fusion module (6). The secondary tracking system (5) includes a set of earbuds, which are worn by a viewer of the naked-eye stereoscopic display device (3). The earbuds capture the velocity and / or acceleration data of the viewer's head and communicate wirelessly with the secondary tracking system (5).

[0061] Figure 3 A second component (8) according to the present invention is schematically shown. The second component consists of a naked-eye stereoscopic display device (3) and a tracking system (7). The tracking system consists of a tracking camera system (4), a secondary tracking system (5) and a sensor fusion module (6). The secondary tracking system (5) includes a remote device that is not worn by a viewer of the naked-eye stereoscopic display device (3) and is therefore not affected by the viewer's movement (since the viewer is not wearing the device, it is identified as being far away from the viewer). Figure 3 In the embodiment, the secondary tracking system (5) is integrated with the naked-eye stereoscopic display device (3); the secondary tracking system is rigidly connected to the naked-eye stereoscopic display device (3), and therefore has no possibility of moving relative to the naked-eye stereoscopic display device; that is, the secondary tracking system is affected by the movement of the naked-eye stereoscopic display device (3).

[0062] In another embodiment (not shown), the secondary tracking system (5) may be physically disconnected from the naked-eye stereoscopic display device (3) so that it is not affected by the movement of the naked-eye stereoscopic display device (3).

[0063] The naked-eye stereoscopic display device (3) may be selected from the group consisting of a television, a (desktop) monitor, a cinema display system, a mobile phone, a tablet computer, a laptop computer and a game console.

[0064] Preferably, the secondary tracking system (5), in particular the remote device, is integrated in the autostereoscopic display device (3), so that when the viewer is viewing the content from the autostereoscopic display device (3), only one device is involved. This makes viewing more convenient and allows the invention to be integrated into a single handheld autostereoscopic display device (3), such as a mobile phone, tablet computer, laptop or game console.

[0065] When the naked-eye stereoscopic display device (3) is handheld, its movement and rotation during use need to be taken into account when performing the method of the present invention in the naked-eye stereoscopic display device (3). As described above in the description of the method of the present invention, the naked-eye stereoscopic display device (3) is preferably configured to generate third-order data about the speed and / or acceleration of the naked-eye stereoscopic display device (3) itself. This will be necessary when the naked-eye stereoscopic display device (3) is present in a mobile vehicle such as a car or a ship, even if the device is not handheld in this setting.

Claims

1. A method for tracking facial features of a viewer of a naked-eye stereoscopic display device, the method comprising acquiring first sequential data of a position of the facial features relative to the naked-eye stereoscopic display device, wherein the first sequential data is obtained with a first delay, The method further comprises - obtaining second sequential data of the velocity and / or acceleration of the facial feature relative to the naked eye stereoscopic display device, wherein the second sequential data is obtained with a second delay, the second delay being lower than the first delay; - performing sensor fusion to combine the first sequential data and the second sequential data to generate final sequential data of the position of the facial features relative to the naked eye stereoscopic display device, The final sequential data is obtained with a final delay that is lower than the first delay.

2. The method of claim 1, wherein the first sequential data is obtained by using one or more cameras, wherein the one or more cameras are configured to record facial features of viewers in the observation field of the one or more cameras.

3. The method according to claim 1 or 2, wherein the second sequential data is obtained by using a wearable device worn on the head of the viewer, wherein, The wearable device is configured to generate data regarding velocity and / or acceleration of the wearable device.

4. The method according to any one of claims 1 to 3, wherein - obtaining second sequential data by taking into account the velocity and / or acceleration of the naked eye stereoscopic display device; and The autostereoscopic display device is configured to generate third order data regarding a velocity and / or an acceleration of the autostereoscopic display device.

5. The method according to claim 3 or 4, wherein the wearable device is selected from the group consisting of earbuds, headphones, glasses, headbands, headwear, rubber bands and hats.

6. The method according to claim 1 or 2, wherein the second sequential data is obtained by using a remote device that is not worn on the viewer's head.

7. The method of claim 6, wherein the remote device relies on the use of the Doppler effect, triangulation or time of flight.

8. The method according to any one of claims 1 to 7, wherein the facial features are selected from the group consisting of eyes, nose, ears, head and mouth.

9. The method according to any one of claims 1 to 8, wherein the naked-eye stereoscopic display device is selected from the group consisting of a television, a (desktop) monitor, a cinema display system, a mobile phone, a tablet computer, a laptop computer and a game console.

10. The method of any one of claims 1 to 9, wherein performing sensor fusion comprises using a Kalman filter.

11. A tracking system (7) for tracking facial features of a viewer of a naked-eye stereoscopic display device (3), the tracking system (7) comprising: - a tracking camera system (4) configured to obtain first sequential data of the position of facial features relative to the naked-eye stereoscopic display device (3), wherein the first sequential data is obtained with a first delay; - a secondary tracking system (5) configured to obtain second sequential data of velocity and / or acceleration of facial features relative to the naked eye stereoscopic display device (3), wherein the second sequential data is obtained with a second delay, the second delay being lower than the first delay; - A sensor fusion module (6) configured to combine the first sequential data and the second sequential data to generate final sequential data of the position of the facial features relative to the naked-eye stereoscopic display device (3).

12. A tracking system (7) according to claim 11, wherein the secondary tracking system (5) comprises a wearable device that can be worn on the viewer's head and comprises one or more accelerometers and, optionally, one or more measuring instruments selected from the group of gyroscopes and magnetometers.

13. A tracking system (7) according to claim 11, wherein the secondary tracking system (5) comprises a remote device, which is configured to measure the velocity and / or acceleration of facial features relative to the remote device by irradiating the facial features with electromagnetic radiation or sound waves, for example by utilizing the Doppler effect, time of flight or triangulation.

14. An assembly (8) comprising a naked-eye stereoscopic display device and a tracking system according to any one of claims 11 to 13.

15. The assembly (8) of claim 14, wherein the secondary tracking system (5) comprises a remote device according to claim 13, wherein: The remote device is integrated into the naked-eye stereoscopic display device (3).

16. The assembly (8) according to claim 14 or 15, wherein the naked-eye stereoscopic display device (3) is configured to generate third sequence data about the speed and / or acceleration of the naked-eye stereoscopic display device (3).

Images