No icon

Camera Shutter-Independent Registration and Rectification

Camera Shutter-Independent Registration and Rectification

Abstract:

Inevitable camera motion during exposure does not augur well for free-hand photography. Distortions introduced in images can be of different types and mainly depend on the structure of the scene, the nature of camera motion, and the shutter mechanism of the camera. In this paper, we address the problem of registering images taken from global shutter and rolling shutter cameras and reveal the constraints on camera motion that admit registration, change detection, and rectification. Our analysis encompasses degradations arising from camera motion during exposure and differences in shutter mechanisms. We also investigate conditions under which camera motions causing distortions in reference and target image can be decoupled to yield the underlying latent image through RS rectification. We validate our approach using several synthetic and real examples.

Existing System:

We specifically address the differences in the distortions present in captured images from RS and GS cameras. We tackle two important problems in this scenario. The first is to perform occlusion-robust registration of a noblur reference image onto a motion blurred image. More specifically, we address the case of registering a rollingshutter- no-blur (RSNB) image onto a rolling-shutter-motionblurred (RSMB) image, both from same or different cameras. Since rolling shutter is a more general form of imaging model, our method is equally applicable when the first image is global-shutter-no-blur (GSNB). Also, the second image can be RSNB or global-shutter-motion-blurred (GSMB), both of which are simplified forms of RSMB. It is common to encounter RSNB-RSNB/RSMB image pairs while capturing images using an RS camera under single and multi-exposure settings. Since the images can be from an RS or GS camera, this also opens up the possibility of registration across cameras with different shutter mechanisms. Such a cross-camera registration is valuable for internet imaging, surveillance from multiple cameras etc.

 

Proposed System:

We determine the conditions under which the scope of the image registration problem can be expanded to accommodate more general scenarios vis-a-vis existing works. Specifically, we expand the scope of by allowing the reference image to be RS distorted, under fairly general conditions of camera motion i.e. translations, out-of-plane rotations about X and small out-of-plane rotations about Y .

Since our work allows for the reference image to be RS affected, this opens up the possibility of shutter independent registration across cameras (a first in the literature).

We propose a method to perform latent image estimation from a single RSNB-GSMB image pair, which is also a first attempt of its kind.

Comment As:

Comment (0)