3D with Rolling Shutter

Cenek Albl (CMP CIIRC Prague, Czech Republic)

Abstract:

Rolling shutter is a mechanism present in vast majority of today's consumer cameras, ranging from the cheapest cellphones to professional DSLR's. When the camera or the object in the image moves the resulting image contains various kinds of distortion that can pose a challenge for the standard computer vision algorithms.

In the talk we focus on the issues caused by rolling shutter and the challenges it poses in 3D Vision tasks such as Structure from Motion and Augmented Reality. We show how to upgrade existing algorithms such as Camera Absolute Pose and Bundle Adjustment to use Rolling Shutter camera models that describe the camera motion. Both the benefits and the drawbacks of using such camera models will be shown and discussed.