GB2475388A

GB2475388A - Optical image stabilisation system for camera

Info

Publication number: GB2475388A
Application number: GB1018537A
Authority: GB
Inventors: Richard John Topliss
Original assignee: Cambridge Mechatronics Ltd
Current assignee: Cambridge Mechatronics Ltd
Priority date: 2009-11-10
Filing date: 2010-11-03
Publication date: 2011-05-18
Also published as: GB201018537D0; GB0919640D0

Abstract

The present invention describes an improvement to the control accuracy of an optical image stabilisation (OIS) system that uses a multiplicity of shape memory alloy (SMA) actuators. Specifically, the control gain of each actuator has a variance from device to device that is larger than the required system accuracy. Therefore the present invention is to use feedback from the image sensor captured images to calibrate and optimise the control gain to minimise the camera shake.

Description

Optical image stabilisation This brief relates to the invention of an optical image stabiliser (OIS) module that makes use of shape memory alloy (SMA) to provide the actuator driving force.

This has been the subject of several previous disclosures by the same applicant. The present invention represents a further development of the technology.

It is recognised that the function of the OIS is to suppress base excitations to the camera system. Current market requirements (over a certain narrow bandwidth approximately between 5-10Hz) are for the OTS to reduce base excitations by 20dB (approximately an order of magnitude).

On this basis, assuming for example a base excitation rotating the camera by 0.1° at 6Hz, the OIS approximately needs to reduce this to 0.0 10. This is a system tolerance.

In order to do this, the OIS system needs to be highly accurate, so that when for example there is a command to rotate the camera angle by 0.1°, the delivered rotation is accurate to 10% worst case (i.e. 0.09° to 0.11°). This assumes that there are no errors in the rotation sensing parts of the OTS system (typically gyroscopes), which in practice there will be. On this basis, an OIS actuator system tolerance requirement of 5% is perhaps more realistic.

It is known, based on Cambridge Mechatronics Limited's development of SMA auto-focus actuator modules that the stroke tolerance band for a given command from module to module is greater than 10%. Factors that affect this tolerance include the SMA wire properties and how they vary from wire to wire, combined with tolerances on the actuator bias springs, and assembly tolerances.

The basic control philosophy is to use the electrical resistance of each SMA wire as the feedback parameter to control position, as over a proportion of the actuator movement range (as the SMA wire stretches or contracts) the actuator change in position is approximately proportional to the SMA wire change in resistance. There are several non-linearities, but in addition, the affect of the tolerance variations is that there is a variation in the resistance versus actuator position gain.

The present invention is to provide a compensation for this variation to deliver the 20dB oscillation suppression demanded by the market.

The core invention is to use information from the camera image sensor to calibrate the resistance versus position gain.

There are two potential non-limiting embodiments of this core invention described, although there may be others.

The first is to calibrate the SMA control gain as part of a start up (or periodic) calibration cycle. It could even form part of a final manufacturing set up procedure of the system before the product containing the OIS and camera system leaves the factory. If this latter option is preferred, the results of the calibration would need to be stored in non-volatile memory in the system. If the option is taken that the system is calibrated locally every time it is turned on or reset, the calibration data does not need to be stored beyond the given camera use.

The basic algorithm is that on turn on, the image sensor is looking at a scene.

Depending on the solution chosen, this may be a test target such as a chequerboard pattern, appropriately illuminated, if carried out in the manufacturing facility.

Alternatively it may be a random scene if being calibrated locally for a given use. In either case, the system in the first instance assumes there is no external base excitation. An oscillatory signal is then applied to the OIS system to generate an oscillation and rotate the camera. This rotation is detected by the image sensor, combined with a suitable detection algorithm. The amount a given feature has moved on the image can be used to assess the real amplitude of the oscillation. This can be related to the amplitude in the applied change in resistance, to calibrate the real position versus resistance gain.

A further improvement to this basic algorithm is that it may not be necessary to assume no external base excitation is applied to the system. This is because the gyroscopes or other rotation sensing parts of the OIS system are functional and can be used to account for the image shake observed on the image, so that the gain calibration can still be performed by deconvolving' the oscillation induced from the actuator with that produced by the base excitation.

An alternative embodiment uses no separate system calibration stage. There are known digital (software) anti-shake algorithms, which assess the movement of features on the image collected by the image sensor and then move' the perceived image to try and keep the objects stationary from frame to frame. These algorithms are generally poorer in terms of image quality than full optical image stabilisation, since information is lost from the image, as at best the outer frame' of the image is deleted. However, this digital correction can essentially be used to optimise the performance of the optical image stabilisation system. For the present invention, the output from the continuous digital shake detection' is used to optimise the OIS gain.

The output from the digital shake algorithm can be collapsed to a camera shake metric (or maybe two such metrics in orthogonal directions). The resistance versus position gain of the SMA actuator can then be adjusted to minimise the camera shake metrics.

In this way the gain of the system is altered to continuously optimise the shake cancelation, and so over time, no accurate assumption need be made about the resistance gain of the SMA actuators.

In an OIS system comprising 4 SMA actuators, ideally the optimisation algorithm may consider the gains of each of the four SMA actuators separately. For example, over a certain time period, the gains of three of the SMA actuators may be held constant, and only the fourth is adjusted to minimise the camera shake metric, before moving onto the next actuator.

Other non linear system parameters and models, such as SMA hysteresis and creep may also be optimised over time by the same method of only adjusting one parameter at a time and finding a local minimum to the camera shake metric.

Claims

Claims 1. An optical image stabiliser (OIS) for miniature cameras that uses a multiplicity of shape memory alloy (SMA) actuators to move all or part of a camera assembly to compensate for externally applied camera shake with the aim of minimising the image shake, where the control signal for each SMA actuator to deliver target movements is calibrated and adjusted based on feedback on the image shake measured on a series of images captured by the image sensor.
2. The OIS of claim 1, wherein an oscillatory drive signal is applied to the one or more actuators of the system to generate an oscillation of the image projected onto the image sensor, and where the amplitude of the image oscillation is assessed by analysing the movement of features in a series of images, these measurments then being used to calibrate the drive signal applied to the actuators, so that the rotation of the image caused by driving the actuator with a given drive signal is known for future image stabilisation operations within acceptable bounds.
3. The OIS of claim 1, wherein the image shake is continually assessed by analysing a series of captured images from the image sensor to form one or more image shake metrics, and further wherein the gain of the applied signals to the SMA actuators are adjusted over time to minimise the one or more image shake metrics.
4. The OIS of any proceeding claim, wherein the drive signal to the SMA actuators whose gain is optimised or calibrated to minimise image shake is based on the resistance of the SMA actuators.