Implementation of control algorithm for mechanical image stabilization
(2017)Department of Automatic Control
- Abstract
- Cameras mounted on boats and in other similar environments can be hard to use if waves and wind cause unwanted motions of the camera which disturbs the desired image. However, this is a problem that can be fixed by applying mechanical image stabilization which is the goal of this thesis.
The mechanical image stabilization is achieved by controlling two stepper motors in a pan-tilt-zoom (PTZ) camera provided by Axis Communications. Pan and tilt indicates that the camera can be rotated around two axes that are perpendicular to one another.
The thesis begins with the problem of orientation estimation, i.e. finding out how the camera is oriented with respect to e.g., a fixed coordinate system. Sensor fusion is used for fusing accelerometer... (More) - Cameras mounted on boats and in other similar environments can be hard to use if waves and wind cause unwanted motions of the camera which disturbs the desired image. However, this is a problem that can be fixed by applying mechanical image stabilization which is the goal of this thesis.
The mechanical image stabilization is achieved by controlling two stepper motors in a pan-tilt-zoom (PTZ) camera provided by Axis Communications. Pan and tilt indicates that the camera can be rotated around two axes that are perpendicular to one another.
The thesis begins with the problem of orientation estimation, i.e. finding out how the camera is oriented with respect to e.g., a fixed coordinate system. Sensor fusion is used for fusing accelerometer and gyroscope data to get a better estimate. Both the Kalman and Complementary filters are investigated and compared for this purpose. However, the Kalman filter is the one that is used in the final implementation, due to its better performance.
In order to hold a desired camera orientation a compensation generator is used, in this thesis called reference generator. The name comes from the fact that it provides reference signals for the pan and tilt motors in order to compensate for external disturbances. The generator gets information from both pan and tilt encoders and the Kalman filter. The encoders provide camera position relative to the camera’s own chassi. If the compensation signals, also seen as reference values to the inner pan-tilt control, are tracked by the pan and tilt motors, disturbances are suppressed.
In the control design a model obtained from system identification is used. The design and control simulations were carried out in the MATLAB extensions Control System Designer and Simulink. The choice of controller fell on the PID.
The final part of the thesis describes the result from experiments that were carried out with the real process, i.e. the camera mounted in different setups, including a robotic arm simulating sea conditions. The result shows that the pan motor manages to track reference signals up to the required frequency of 1Hz. However, the tilt motor only manages to track 0.5Hz and is thereby below the required frequency. The result, however, proves that the concept of the thesis is possible. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/8926538
- author
- Gustavi, Magnus and Andersson, Louis
- supervisor
- organization
- year
- 2017
- type
- H3 - Professional qualifications (4 Years - )
- subject
- report number
- TFRT-6042
- ISSN
- 0280-5316
- language
- English
- id
- 8926538
- date added to LUP
- 2017-10-06 09:33:41
- date last changed
- 2017-10-06 09:33:41
@misc{8926538,
abstract = {{Cameras mounted on boats and in other similar environments can be hard to use if waves and wind cause unwanted motions of the camera which disturbs the desired image. However, this is a problem that can be fixed by applying mechanical image stabilization which is the goal of this thesis.
The mechanical image stabilization is achieved by controlling two stepper motors in a pan-tilt-zoom (PTZ) camera provided by Axis Communications. Pan and tilt indicates that the camera can be rotated around two axes that are perpendicular to one another.
The thesis begins with the problem of orientation estimation, i.e. finding out how the camera is oriented with respect to e.g., a fixed coordinate system. Sensor fusion is used for fusing accelerometer and gyroscope data to get a better estimate. Both the Kalman and Complementary filters are investigated and compared for this purpose. However, the Kalman filter is the one that is used in the final implementation, due to its better performance.
In order to hold a desired camera orientation a compensation generator is used, in this thesis called reference generator. The name comes from the fact that it provides reference signals for the pan and tilt motors in order to compensate for external disturbances. The generator gets information from both pan and tilt encoders and the Kalman filter. The encoders provide camera position relative to the camera’s own chassi. If the compensation signals, also seen as reference values to the inner pan-tilt control, are tracked by the pan and tilt motors, disturbances are suppressed.
In the control design a model obtained from system identification is used. The design and control simulations were carried out in the MATLAB extensions Control System Designer and Simulink. The choice of controller fell on the PID.
The final part of the thesis describes the result from experiments that were carried out with the real process, i.e. the camera mounted in different setups, including a robotic arm simulating sea conditions. The result shows that the pan motor manages to track reference signals up to the required frequency of 1Hz. However, the tilt motor only manages to track 0.5Hz and is thereby below the required frequency. The result, however, proves that the concept of the thesis is possible.}},
author = {{Gustavi, Magnus and Andersson, Louis}},
issn = {{0280-5316}},
language = {{eng}},
note = {{Student Paper}},
title = {{Implementation of control algorithm for mechanical image stabilization}},
year = {{2017}},
}