Estimation and Removal of Motion Blur by Capturing Two Images with Different Exposures
Lim, Suk Hwan; Silverstein, Amnon
Keyword(s): Image stabilization, Deblurring, Mobile Imaging, Multi-frame Imaging, Image enhancement
Abstract: Deblurring refers to restoring digital photographs or videos that have been degraded by optical blurring such as motion blur. Motion blur typically occurs due to the long exposure time relative to the amount of motion of the object or the camera. A simple reduction in the exposure time does not produce desirable images since it results in high noise (i.e., low signal-to- noise ratio). Short exposure images on the other hand have the advantage of much less blur while the long exposure images have the advantage of much less image noise. In this paper, we describe an approach to deblur the long exposure image with additional information from the short exposure image. Our method i) captures two images of the same scene with different exposures, ii) estimates the blur kernel of the long exposure image from the images and iii) uses the estimated kernel to deblur the long exposure image while regularizing on the short exposure image. The goal is to combine the merits of the long and short exposure images and produce a high quality image with low noise and little motion blur. We show experimental results that illustrate the accuracy of the blur kernel estimation and the effectiveness of our deblurring method.
External Posting Date: October 21, 2008 [Fulltext]. Approved for External Publication
Internal Posting Date: October 21, 2008 [Fulltext]