A camera apparatus

Abstract

A method and camera apparatus for capturing an image of a moving object 910 such that the object appears in the image without motion blur comprises a support structure and a camera unit 950 supported

Description

Technical Field The present disclosure relates to a camera apparatus comprising an image sensor and a lens system for focussing an image on the image sensor. The camera apparatus is configured to track motion of an object imaged by the image sensor. Background It is known to deploy optical image stabilisation (OIS) to compensate for camera shake, that is translational and / or rotational vibration of the camera apparatus typically caused by user hand movement, which degrades the quality of the image captured by the image sensor. OIS typically involves detection of vibration by a vibration sensor such as an inertial measurement unit (for example, a gyroscope sensor or an accelerometer), and control on the basis of the detected vibration of an actuator system that adjusts the camera apparatus to compensate for the vibration. Several techniques for adjusting the camera apparatus to compensate for camera shake are known. One technique is to keep the image sensor fixed and to move the lens system. An alternative technique is to keep the lens system still, and to move the image sensor. One possibility is to move either the lens system or the image sensor around two notional axes that are perpendicular to each other and to the optical axis. Another option might be to tilt the either the lens system or the image sensor or to tilt both the lens system and the image sensor. Optical image stabilisation is used generally to compensate for relatively small disturbances, such as translational and / or rotational vibrations, that arise from the camera apparatus being held unsteadily (e.g. hand jitter). Such disturbances may be to any one or more of pitch, yaw and roll of the camera apparatus. The aim of OIS is to remove motion blur caused by the movement of the camera by stabilizing the image on the image sensor. . In other words, the aim is to maintain the camera’s field of vision relative to a world frame which is consistent. OIS movement of the mechanical components may be determined by an algorithm which considers six degrees of freedom (x, y, z, rx, ry, rz) of movement. While OIS techniques are used to compensate for camera instability, they do not compensate for blur caused by scene motion. An event camera, unlike a conventional (rolling or global shutter) camera, is an imaging sensor that detects local changes in brightness. Event cameras do not capture images using a shutter as conventional cameras do. Instead, each pixel of an event camera operates independently and asynchronously, reporting changes in brightness as they occur, and staying silent otherwise. Each pixel stores a reference brightness level, and continuously compares it to the current brightness level. If the difference in brightness exceeds a threshold, that pixel resets its reference level and generates an event: a discrete packet that contains the pixel address and timestamp. Events may also contain the polarity (increase or decrease) of a brightness change, or an instantaneous measurement of the illumination level. Thus, event cameras output an asynchronous stream of events triggered by changes in scene illumination. For each change reported by the event camera, the data provided will define the x and y coordinates of the pixels experiencing change and the magnitude of those changes. Event cameras have microsecond temporal resolution (compared with millisecond temporal resolution for conventional shutter cameras), large dynamic range (typically double or more than that of conventional shutter cameras), and less under / overexposure and motion blur than shutter cameras. Event cameras may be used to detect rapid movements within an image, which makes them particularly suitable for sensing moving objects. Event cameras may also be known as neuromorphic cameras, silicon retinas or dynamic vision sensors. It is known to use an event camera in combination with a shutter camera to produce an image of a moving object with reduced motion blur. This known approach uses an output from the event camera to understand movement in an image produced by the shutter camera, and uses post processing techniques to reconstruct the image produced by the shutter camera so as to represent the moving object with reduced motion blur. Data from both the shutter camera and the event camera contributes to the ultimate image that arises from this approach. The approach requires a degree of inference to be made in order to arrive at the ultimate image. Summary Against this background, there is provided a camera apparatus for capturing an image of an object that is moving relative to its background such that the object appears in focus in the image, the camera apparatus comprising: a support structure; a camera unit supported on the support structure, the camera unit comprising: an image sensor having an optical axis perpendicular to the image sensor; and a primary lens for focusing an image on the image sensor; an actuator configured to move one or both of the image sensor and the primary lens relative to the support structure; and a controller configured to receive object motion tracking data in relation to the object that is moving relative to its background and to provide control instructions to the actuator to move one or both of the image sensor and the primary lens relative to the support structure to maintain a constant landing position on the image sensor of a light path from the object that is moving. In this way, the moving object is rendered by the image sensor without motion blur, whilst the background is rendered blurred. The object motion tracking data may comprise data indicative of direction and speed of the object moving on the image sensor. The object motion tracking data may comprise an asynchronous data stream such that a period of movement of the object comprises a stream of data points. In this way, data is not limited to arriving in complete or partial frames. The object motion tracking data may comprise x and y coordinates of one or more pixels subject to change and, for the or each pixel, a measure of magnitude of the change. The object motion tracking data may be provided by a device other than the image sensor. The camera apparatus may further comprise an event camera configured to detect motion of the object and to provide the object motion tracking data to the controller. The event camera may comprise a multi-pixel event camera sensor. The actuator may comprise a lens shift actuator configured to shift the primary lens. The lens shift actuator may be configured to shift the primary lens along a first notional axis that is perpendicular to the optical axis and along a second notional axis that is perpendicular to the optical axis so as to facilitate shifting of the primary lens about any arbitrary axis perpendicular to the optical axis. The actuator may comprise an image sensor shift actuator configured to shift the image sensor. The image sensor shift actuator may be configured to shift the image sensor along a first notional axis that is perpendicular to the optical axis and along a second notional axis that is perpendicular to the optical axis so as to facilitate shifting of the image sensor about any arbitrary axis perpendicular to the optical axis. The actuator may comprise a primary lens tilt actuator configured to tilt the primary lens. The primary lens tilt actuator may be configured to tilt the primary lens relative to the optical axis. The actuator may be configured to move the camera unit relative to the support structure. In an arrangement involving an event camera, the controller may be further configured to perform image registration between an image provided by the image sensor and a field of vision of the event camera. The camera apparatus may further comprise lens distortion compensation wherein the lens distortion compensation varies across the lens dependent upon a position of the moving object on the primary lens. In this way, distortions arising from when the light path from the object that is moving arrives at the edges of the primary lens may be provided with different compensation to distortions arising from when the light path from the object that is moving arrives at the centre of the primary lens. The actuator may be configured to track an object from one extreme of actuator stroke to another extreme of actuator stroke. The camera apparatus may further comprise a lens assembly, wherein the lens assembly comprises the primary lens. The actuator may be configured to move the lens assembly relative to the support structure. The actuator may comprise at least four SMA actuator wires connected between the camera unit and the support structure and configured to move some or all of the camera unit relative to the support structure. In a further aspect of the disclosure there is provided a method of imaging an object that is moving relative to its background using a camera apparatus so as to target less motion blur of the object than the background, the camera apparatus comprising: a support structure; a camera unit supported on the support structure, the camera unit comprising: an image sensor having an optical axis perpendicular to the image sensor; and a primary lens for focusing an image on the image sensor; an actuator configured to move one or both of the image sensor and the primary lens relative to the support structure; and a controller; wherein the method comprises: receiving object motion tracking data in relation to the object that is moving relative to its background; and controlling the actuator to move one or both of the image sensor and the primary lens relative to the support structure to maintain a position of the image of the object on the image sensor. In this way, the moving object is rendered by the image sensor without motion blur, whilst the background is rendered blurred. The method may further comprise obtaining the object motion tracking data using an event camera configured to provide an output indicative of a trajectory of the object that is moving. The method may further comprise performing a registration process to facilitate registration of the object motion tracking data with a field of view of the image sensor. Figures Figure 1A shows a camera apparatus imaging an object on the optical axis; Figure 1B shows the camera apparatus of Figure 1A in which the camera apparatus has moved relative to the object such that the object is not on the optical axis; Figure 1C shows the camera apparatus of Figure 1B in which a lens of the camera apparatus has moved to compensate for the movement of the camera apparatus; Figure 2 shows a camera apparatus with five different approaches to mechanical movements of camera elements which may be deployed in optical image stabilisation and auto focusing; Figure 3 shows a high-level process flow for operation of a method in accordance with the disclosure; Figure 4 shows a high-level process for registering an output from an event camera with an output from a shutter camera in order to facilitate performance of the method; Figure 5 shows an object at intervals during movement across a field of vision of a shutter camera; Figure 6 shows a portion of an image from a shutter camera which will exhibit motion blur resulting from the moving object of Figure 5 being captured in a single frame of a shutter camera; and Figure 7 shows a portion of a field of vision of an event camera that returns pixel change data representative of the moving object of Figure 5 moving in the field of vision of the event camera; Figure 8 shows a representation of a trajectory of the moving object inferred from initial data derived from the event camera in relation to the first part of the movement of the object; and Figure 9 shows a highly schematic representation of an apparatus in accordance with the disclosure. Detailed description The present disclosure makes use of camera apparatus compensation techniques that are often deployed in optical image stabilisation, but deployed differently for an altogether different objective. Rather than seeking to minimise motion blur resulting from camera shake, per optical image stabilisation, the approach of the present disclosure is instead to facilitate capture of a moving object within an image scene in a way that minimises motion blur of the moving object whilst allowing background blur. This is achieved by moving one or more mechanical parts of the camera apparatus which might ordinarily be moved in optical image stabilisation with the aim instead of tracking the moving object across the image sensor. For context, various implementations of optical image stabilisation are now discussed. OIS techniques seek to control the optical path between the target and the image sensor by moving mechanical parts of the camera itself. Thus, even if the camera shakes, the OIS ensures that light arriving at the image sensor does not change its landing position on the image sensor. One simple approach to optical image stabilisation may be to adjust the position of a lens in the optical path between the object and the image sensor. Figure 1A shows a basic camera apparatus 1 comprising a camera unit including an image sensor 4 and a primary lens 3. The image sensor 4 has an optical axis X. An object 30 to be captured by the camera apparatus is located on the optical axis X. A light path 40 shows the path of light between the object and the image sensor 4. Figure 1B shows the camera apparatus of Figure 1B but with the camera apparatus having been tilted upwards, representative of unsteady holding of the camera apparatus. Figure 1B shows the optical axis X tilted upwards such that, in consequence, the light path 40 falls on a different part of the image sensor 4. Movement of the light path 40 so as to land on a different part of the image sensor 4 during exposure of the image results in blurring in the final image. Figure 1C shows how, by shifting the primary lens 3 to a new position 3’, the light path 40 is effectively adjusted to land on the image sensor at its original location (per Figure 1A). By adjusting the position of the primary lens 3 continuously in response to movement of the camera apparatus (due to unsteady holding), this enables the landing position of the light path on the image sensor 4 to remain constant. The example of Figures 1A, 1B and 1C shows only one technique for achieving optical image stabilisation. Other techniques may involve mechanical adjustment of other components in a camera apparatus. Figure 2 shows various alternative approaches. The camera apparatus 1 of Figure 2 comprises a support structure 10 and a movable part 20. The camera apparatus 1 further comprises a lens assembly 3, an image sensor 4 and a controllers. The lens assembly 3 includes one or more lenses configured to focus an image on the image sensor 4. In this case, the lens assembly 3 defines an optical axis O. The lens assembly 3 may include a lens carrier, for example in the form of a cylindrical body, supporting the one or more lenses. The image sensor 4 captures an image and may be of any suitable type, for example a charge coupled device (CCD) or a complementary metal-oxide-semiconductor (CMOS) device. The camera assembly 1 may be a compact camera assembly in which each lens has a diameter of 20mm or less, for example of 12mm or less. In the (“sensor-shift”) variation of the camera assembly 1 shown in option A of Figure 2, the movable part 20 includes the image sensor 4. The lens assembly 3 may be fixed relative to the support structure 10, or may be movable relative to the support structure 10 along the optical axis O, as described below. In the (“lens-shift”) variation of the camera assembly 1 shown in option B of Figure 2, the image sensor 4 is fixed relative to the support structure 10 and the movable part 20 includes the lens assembly 3. The lens assembly 3 may be movable relative to the movable part 20 along the optical axis O, as described below. In both of these variations, the actuator assembly 2 is configured to move the lens assembly 3 relative to the image sensor 4 in any direction In the plane perpendicular to the primary axis P and hence the optical axis O. Such movement has the effect of moving the image on the image sensor 4 and enables optical image stabilisation (OIS) to be implemented in the camera assembly 1. In the sensor-shift variation, the movable part 20 may also be rotatable about the primary axis P so as to also enable compensation for roll. In the (“module-tilt”) variation shown in option C of Figure 2, the movable part 20 includes both the lens assembly 3 and the image sensor 4. Again, the lens assembly 3 may be movable relative to the movable part 20 along the optical axis O, as described below. The actuator assembly 2 is configured to tilt the movable part 20 about two axes perpendicular to the primary axis P and to each other, and optionally rotate the movable part 20 about the primary axis P, enabling OIS to be implemented in the camera assembly 1. Option D of Figure 2 does not show an implementation suitable for optical image stabilisation but, rather, shows an implementation which can be deployed for “autofocus” or “zoom”. In this arrangement, the movable part 20 includes the lens assembly 3, and the actuator assembly 2 moves the movable part 20 relative to the support structure 10 along the optical axis O. Such movement has the effect of adjusting the focus of the image on the image sensor 4 or providing zoom functionality. So, auto-focus (AF) or zoom functionality can be implemented in the camera assembly 1. In some examples (not shown), the camera assembly 1 may include a first actuator assembly for providing OIS as illustrated in options A, B and C of Figure 2, and a second actuator assembly for providing AF or zoom as illustrated in option D of Figure 2. One or both of the first and second actuator assemblies may correspond to actuator assemblies 2 as described herein. One or both of the first and second actuator assemblies may a shape memory alloy (SMA) actuator assembly. One or both of the first and second actuator assemblies may be a non-SMA actuator assembly, e.g. a voice-coil motor (VCM) actuator assembly. As will be appreciated, in the lens-shift and module-tilt variations, the support structure 10 of the second actuator assembly 2 is fixed to (or corresponds to) the movable part 20 of the first actuator assembly 2. In the (“AF+OIS”) variation shown in option E of Figure 2, the movable part 20 includes the lens assembly 3, and the actuator assembly 2 produces three-dimensional translational movement of the movable part 20 relative to the support structure 10, enabling both AF and OIS to be implemented using one actuator assembly 2. Other variations are also possible. For example, in the autofocus variation or the AF+OIS variation, the movable part 20 may include the image sensor 4 rather than the lens assembly 3. The camera assembly 1 may include combinations of the above-described features, e.g. (a) lens shift and sensor shift, (b) module tilt and lens shift or sensor shift and autofocus, or (c) module tilt and AF+OIS. The controllers may be implemented in an integrated circuit (IC) chip. The controller is configured to control the actuator assembly 2. In some implementations, the actuator assembly 2 comprises SMA wires and the controller 8 generates drive signals for the SMA wires. SMA material has the property that, on heating, it undergoes a solid-state phase change that causes the SMA material to contract. Thus, applying drive signals to the SMA wires, thereby heating the SMA wires by causing an electric current to flow, will cause the SMA wires to contract and thus drive relative movement of the movable part 20. The drive signals are chosen to drive relative movement of the movable part 20 in a desired manner, for example so as to achieve OIS by stabilizing the image sensed by the image sensor 4 or to achieve AF by adjusting the focus of the image sensed by the image sensor 4. The controllers supplies the generated drive signals to the SMA wires. Any of these approaches to moving one or more components of the camera apparatus 1 relative to the support structure 10 may be deployed in the context of the present disclosure for the purposes of capturing an object moving relative to its background in such a way that the object appears in the image without motion blur, whilst rendering the background as a blurred image. In the case of conventional OIS, the camera assembly 1 may also include a motion sensor (not shown), which may include a 3-axis gyroscope and a 3-axis accelerometer. The motion sensor can generate signals representative of the motion (specifically vibrations or “shake”) of the camera assembly 1, which can be processed so as to produce signals representative of the required movement of the movable part 20 to compensate for such shake. The controller 8 receives such signals and can generate the drive signals for the SMA wires to achieve OIS. Current motion sensors deployed for OIS purposes are not capable of tracking a moving object with the aim of capturing the object without motion blur. This is because conventional motion sensing for OIS is intended to track motion of the camera apparatus (e.g. due to camera shake) relative to its starting position whereas tracking a moving object requires information regarding the movement of the object relative to the camera apparatus. Another way to consider the differences is that motion sensing for OIS purposes involves knowledge of the movement of the camera relative to a constant world frame whereas motion sensing for the purposes of tracking a moving object involves knowledge of the moving position of the object (e.g. the trajectory and speed) relative to a constant world frame. Of course, a camera apparatus configured for achieving an image of a moving object without motion blur might still be subject to camera shake, which may still need to be sensed and mitigated, in addition to the need to sense movement of the object relative to the world frame. The nature of the data that is necessary to describe movements relating to camera shake is different from the nature of the data required to describe trajectory and speed of a moving object within a field of view of the camera apparatus. Defining movement of a camera apparatus arising from camera shake may comprise data regarding six degrees of freedom. The six degrees of freedom are translational movement in any of three dimensions (axes: x, y, z), and rotational movement about any of those dimensions (rx, ry, rz). By contrast, the trajectory of a moving object within a field of view of a camera apparatus may be described by data that defines a two-dimensional vector (X, Y). This two-dimensional vector may be determined through collecting successive data points describing the two-dimensional position (x, y) of the object. Since a conventional shutter camera operates in discrete frames, it may be difficult to interpret data provided by the conventional shutter camera during a single frame to influence the mechanical movement of elements of the camera apparatus to track motion within that same frame. Therefore, in one implementation, the object tracking data may be provided at least in part by an external source such as an event camera. An event camera returns values for changes in pixel readings asynchronously in response to the changes in pixel readings. The event camera comprises a multi-pixel event camera sensor. Thus, pixel locations that correspond to the location of a moving object will prompt pixel readings while pixels that correspond to a static background will not prompt pixel readings. By tracking the trajectory of the changing pixel values across the field of view of the event camera, a two-dimensional vector (X, Y) defining the trajectory of the object may be determined. Such data may be termed object motion tracking data. An objective is to use data describing the trajectory and speed of the moving object to control the mechanical movement of elements of the shutter camera in synchronisation with the movement of the moving object throughout a frame of the shutter camera so that the light path from the moving object lands at a consistent position on the image sensor of the shutter camera throughout the exposure period of the frame. In order to achieve this, it is necessary for there to be registration between the event camera data and the shutter camera data. This may be termed image registration. This may be particularly important if the field of view of the event camera is not identical to the field of view of the shutter camera. The registration process is part of what enables effective control of the mechanical movement of the elements of the shutter camera to target the moving object appearing at a consistent position on the image sensor during the exposure. Figure 3 shows a high-level process flow 300 for operation of a method in accordance with the disclosure. The process flow 300 comprises obtaining image data from an event camera at step 310. These data may provide details of the locations (x, y) of pixels experiencing a change and the magnitude of each change. At step 320 the event camera data may be processed in order to obtain a trajectory and speed information in relation to an object moving across a field of vision of the event camera. At step 330, a control process is configured to control mechanical operation of one or more elements of the shutter camera (per the possible techniques shown in Figure 2) in order to track the established trajectory and speed of the moving object during the shutter camera frame so that the moving object appears at a consistent position on the image sensor of the shutter camera during the exposure period of the frame. Steps 310, 320 and 330 may occur continuously throughout the exposure period of the shutter camera. Steps 310, 320 and 330 may begin prior to the start of the exposure period of the shutter camera. At step 340, image data is obtained from the shutter camera for the frame during which frame the position of the moving object has been retained consistently at the same position on the image sensor. In this way, the moving object appears in the shutter camera image without motion blur while the background appears in the shutter camera image with motion blur. Figure 4 shows a high-level process for registering an output from an event camera with an output from a shutter camera in order to facilitate performance of the method. It is necessary for the field of view of the shutter camera to be registered with the field of view of the event camera. It may be, for example, that the field of view of the event camera is larger than the field of view of the shutter camera. In order to address the different fields of view, it is necessary to map one or more positions in the field of view of the event camera with the corresponding one or more positions in the field of view of the shutter camera. As shown in Figure 4, the registration process may involve obtaining shutter camera image data at step 410, obtaining event camera data at step 420 and, at step 430, performing a registration function in order to achieve registration between shutter camera image data and event camera data. What follows is a simplified example of a scenario of a moving object in the field of view of the shutter camera and the steps involved to achieve an image in which the moving object appears without motion blur in the image output by the shutter camera. Figure 5 shows a field of vision 510 of a conventional shutter camera in which an object 520 is shown at 20 different locations within the field of vision 510. The 20 locations of the object 520 represent a sequence of movement of the object 520 from left to right across the field of vision 510. In this way, each of the of the 20 locations of the object 520 is separated from the previous location by a time interval. The remainder of the field of vision 510, that is the portion of the image never occupied by the object, is taken to be static. If the object 520 moves across the field of vision within the period of a single frame (that is, if the time interval between each location is no longer than 1 / 20th of the time of a single exposure of the shutter camera) then the shutter camera would render the object 520 with motion blur. The region of motion blur 530 (as shown in Figure 6) occupies all stages of the trajectory of the object 520 across the frame. Figure 7 shows an event camera field of vision 710 which corresponds with the shutter camera field of vision 510 of Figures 5 and 6. The correspondence may be exact or may be the result of a registration between the output of the event camera and the output of the shutter camera. The event camera operates to identify changes in pixel values, where present. The change region 730 of the event camera field of vision shows the pixels that return a result for pixel value change. The remainder of the event camera field of vision 710 does not return a result for pixel value change. While Figure 7 simply identifies the location of pixels that return a result for pixel change, the event camera also provides more detailed data which are not shown in Figure 7 including the magnitude of the change in pixel values by the event camera. (Such data may be visualised in the form of a heat map whereby, for example, dark colours represent small pixel value changes and increasingly bright colours represent increasingly large pixel value changes.) As is clear from a comparison of Figures 6 and 7, the area of pixel change 730 in the event camera field of vision (Figure 7) corresponds with the area that would ordinarily (in the absence of the techniques described here) experience motion blur 530 in the shutter camera image (Figure 6). From the results provided by the event camera, data regarding the trajectory (see Figure 8) and speed of the moving object is provided to the shutter camera, during the exposure of the shutter camera, to control movement mechanical adjustment feature(s) of the shutter camera in order that light from the moving object is directed to land at a consistent position on the image sensor 4 of the shutter camera during the period of exposure of the frame of the shutter camera. In some embodiments, it may be that the field of vision of the event camera is larger than the field of vision of the shutter camera. In this scenario, trajectory and speed information obtained outside the field of vision of the shutter camera can be provided to the shutter camera before the moving object appears within the field of vision of the shutter camera. Thus, the mechanical adjustment of the lens or the image sensor may be prepared such that on entry into the field of vision of the shutter camera the light arriving from the moving object is directed to the centre of the image sensor of the shutter camera, and is actively retained at the centre of the image sensor throughout the period of exposure of the shutter camera as the object moves through the field of vision of the shutter camera. Figure 9 shows a highly schematic representation of an apparatus 900 in accordance with the disclosure. The apparatus 900 comprises an event camera 950, a shutter camera 960 and a processor 970. When image capture of a moving object 910 is initiated, the event camera 950 begins to track movement of the object 910, 910’, 910” as it moves across the field of view of the shutter camera. From this movement, the processor 970 determines object trajectory and speed information and uses that information to calculate control signals to adjust the positions of one or more mechanically movable elements of the shutter camera 960 (e.g. a lens assembly of the shutter camera 960) throughout (and potentially also prior to) the exposure period of the shutter camera 960 so that light from the moving object lands on a consistent location of the image sensor of the shutter camera 960 throughout the exposure period. In this way, the moving object is rendered in the resulting shutter camera image without motion blur. It should be noted that the event camera information is used only as part of the control process for moving the mechanically movable parts of the shutter camera. The event camera data does not otherwise contribute to the final image which may be derived entirely from the image sensor of the shutter camera. While the embodiments described above involve obtaining the data regarding the trajectory and speed of the moving object using an event camera, an event camera is not essential to the approach. For example, the data regarding the moving object trajectory and speed may be obtained using a second shutter camera, potentially with a faster read out speed that the shutter camera used to obtain the final image. In a further alternative, the data regarding the trajectory and speed of the moving object may be obtained by the shutter camera used to obtain the final image, if such camera is able to provide the data sufficiently rapidly.

Claims

1. A camera apparatus for capturing an image of an object that is moving relative to its background such that the object appears in the image without motion blur, the camera apparatus comprising:a support structure;a camera unit supported on the support structure, the camera unit comprising:an image sensor having an optical axis perpendicular to the image sensor; anda primary lens for focusing an image on the image sensor;an actuator configured to move one or both of the image sensor and the primary lens relative to the support structure; anda controller configured to receive object motion tracking data in relation to the object that is moving relative to its background and to provide control instructions to the actuator to move one or both of the image sensor and the primary lens relative to the support structure to maintain a constant landing position on the image sensor of a light path from the object that is moving.

2. The camera apparatus of claim 1 wherein the object motion tracking data comprises data indicative of direction and speed of the object moving on the image sensor.

3. The camera apparatus of claim 1 or claim 2 wherein the object motion tracking data comprises an asynchronous data stream such that a period of movement of the object comprises a stream of data points.

4. The camera apparatus of any preceding claim wherein the object motion tracking data comprises x and y coordinates of one or more pixels subject to change and, for the or each pixel, a measure of magnitude of the change.

5. The camera apparatus of any preceding claim wherein the object motion tracking data is provided by a device other than the image sensor.

6. The camera apparatus of any preceding claim further comprising an event camera configured to detect motion of the object and to provide the object motion tracking data to the controller.

7. The camera apparatus of claim 6 wherein the event camera comprises a multi-pixel event camera sensor.

8. The camera apparatus of any preceding claim wherein the actuator comprises a lens shift actuator configured to shift the primary lens.

9. The camera apparatus of claim 8 wherein the lens shift actuator is configured to shift the primary lens along a first notional axis that is perpendicular to the optical axis and along a second notional axis that is perpendicular to the optical axis so as to facilitate shifting of the primary lens about any arbitrary axis perpendicular to the optical axis.

10. The camera apparatus of any preceding claim wherein the actuator comprises an image sensor shift actuator configured to shift the image sensor.

11. The camera apparatus of claim 10 wherein the image sensor shift actuator is configured to shift the image sensor along a first notional axis that is perpendicular to the optical axis and along a second notional axis that is perpendicular to the optical axis so as to facilitate shifting of the image sensor about any arbitrary axis perpendicular to the optical axis.

12. The camera apparatus of any preceding claim wherein the actuator comprises a primary lens tilt actuator configured to tilt the primary lens.

13. The camera apparatus of claim 12 wherein the primary lens tilt actuator is configured to tilt the primary lens relative to the optical axis.

14. The camera apparatus of any preceding claim wherein the actuator is configured to move the camera unit relative to the support structure.

15. The camera apparatus of claim 6 or any claim dependent upon claim 6 wherein the controller is further configured to perform image registration between an image provided by the image sensor and a field of vision of the event camera.

16. The camera apparatus of any preceding claim further comprising lens distortion compensation wherein the lens distortion compensation varies across the lens dependent upon a position of the moving object on the primary lens.

17. The camera apparatus of any preceding claim wherein the actuator is configured to track an object from one extreme of actuator stroke to another extreme of actuator stroke.

18. The camera apparatus of any preceding claim further comprising a lens assembly, wherein the lens assembly comprises the primary lens.

19. The camera apparatus of claim 18 wherein the actuator is configured to move the lens assembly relative to the support structure.

20. The camera apparatus of any preceding claim wherein the actuator comprises at least fourSMA actuator wires connected between the camera unit and the support structure and configured to move some or all of the camera unit relative to the support structure.

21. A method of imaging an object that is moving relative to its background using a camera apparatus so as to target less motion blur of the object than the background, the camera apparatus comprising:a support structure;a camera unit supported on the support structure, the camera unit comprising:an image sensor having an optical axis perpendicular to the image sensor; anda primary lens for focusing an image on the image sensor;an actuator configured to move one or both of the image sensor and the primary lens relative to the support structure; anda controller;wherein the method comprises:receiving object motion tracking data in relation to the object that is moving relative to its background; andcontrolling the actuator to move one or both of the image sensor and the primary lens relative to the support structure to maintain a position of the image of the object on the image sensor.

22. The method of claim 21 further comprising:obtaining the object motion tracking data using an event camera configured to provide an output indicative of a trajectory of the object that is moving.5 23. The method of claim 21 or claim 22 further comprising performing a registrationprocess to facilitate registration of the object motion tracking data with a field of view of the image sensor.21

Images