Algorithm for shake detection in videos for advanced driver assistance systems

Résumé

The aim of this research is to develop a computer vision algorithm for the detection of blurred images caused by the shaking of the camera in an Advanced Driver Assistance System (ADAS). Cameras on board of cars, as part of ADAS, are used to detect various scenarios such as impeding forward collision, pedestrian detection, traffic sign detection, traffic light detection, etc. In case of bumpy roads the user needs to be informed that the system cannot provide accurate warning as the video cannot be used. Also, in case of computer vision based ADAS features running on tablets or smart phones, the user needs to be informed that they will no longer be able to use the ADAS. The acquired images are processed with the proposed algorithm in order to assess if shaking was detected in online video.