Motion Deblurring of RAW Mosaic Image Using Coded Exposure Photography

Jaelin Lee, Byeungwoo Jeon

Abstract: In this paper, a new motion deblurring method is investigated for reconstructing a latent image from a raw mosaic image that has been blurred by motion. Since conventional motion deblurring methods for color images often suffer from significant errors due to nonlinear processing in color image processing pipeline, we devise a new deblurring technique by generating an invertible point spread function for deblurring of raw mosaic image by using coded exposure photography. The effectiveness of this method is demonstrated using a real dataset.