Motion Deblurring for Plenoptic Images

We address for the first time the issue of motion blur in light field images captured from plenoptic cameras. We propose a solution to the estimation of a sharp high resolution scene radiance given a blurry light field image, when the motion blur point spread function is unknown, i.e., the so-called blind deconvolution problem. In a plenoptic camera, the spatial sampling in each view is not only decimated but also defocused. Consequently, current blind deconvolution approaches for traditional cameras are not applicable. Due to the complexity of the imaging model, we investigate first the case of uniform (shift-invariant) blur of Lambertian objects, i.e., when objects are sufficiently far away from the camera to be approximately invariant to depth changes and their reflectance does not vary with the viewing direction. We introduce a highly parallelizable model for light field motion blur that is computationally and memory efficient. We then adapt a regularized blind deconvolution approach to our model and demonstrate its performance on both synthetic and real light field data. Our method handles practical issues in real cameras such as radial distortion correction and alignment within an energy minimization framework.