Image Stabilization Based on Inertial Sensor Data

OData support
Kovács Viktor
Department of Automation and Applied Informatics

MEMS based sensors (accelerometers, gyroscopes) went through a realtively fast evolvement process in the last few years. This quick process made it possible to use them in multiple fields of applications where precise motion information is required.

When taking a photo in a less than ideal environment (low light conditions, holding the equipment in shaky hands) with long exposure times, the result is usually unsatisfactory. The photos become degraded as the movement of the camera during the exposure adds motion blur to the image. However if the movement information of the camera is also captured during the exposure it can be possible to enhance the picture by various deconvolution methods.

It is worth to note that the modern cell phones usually have built-in MEMS sensors so the above mentioned process is applicable on sensor equipped smart phones as well.

In my thesis I examine the existing solutions for the offline image stabilization problem. And in addition to that I look into the inertial sensors and their applicability on the image stabilization problem.


Please sign in to download the files of this thesis.