3D Scene Recovery from Defocused Video Sequence

Abstract: The paper aims at clearing the out of focus blur. Lens defocus introduces uniform blurring, Restoring these kinds of images is an interesting research field .The low resolution cameras manufactured at lower end like mobile phone and web camera mostly suffers through out-of-focus blur. Current auto focus solutions used in commercial cameras are design to ensure that captured images are in focus by adjusting the lens position and aperture opening. This needs an auto focusing hardware. The assembly is bulky and costly, cannot employed at handheld application where cameras are rigid (fixed lens &aperture). Moreover it has fundamental limitations that when the scene contains multiple objects with largely varying depths , a single image cannot capture all the objects in focus simultaneously in that case camera captures multiple images with variable depth. We present a novel method for virtually focusing the defocused video captured by camera. Virtual Focusing stands for the technique of providing an image processing solution to recover focused image sequences from videos taken by an out-of-focus camera with fixed physical parameters (cell phone cameras and webcams).

Keywords: Depth from Defocus, PSF Modeling, Blur radius Estimation, image restoration.