Diffusion Coded Photography for Extended Depth of Field

摘要

In recent years, several cameras have been introduced which extendrndepth of field (DOF) by producing a depth-invariant point spreadrnfunction (PSF). These cameras extend DOF by deblurring a capturedrnimage with a single spatially-invariant PSF. For these cameras,rnthe quality of recovered images depends both on the magnitudernof the PSF spectrum (MTF) of the camera, and the similarityrnbetween PSFs at different depths. While researchers have comparedrnthe MTFs of different extended DOF cameras, relatively littlernattention has been paid to evaluating their depth invariances. Inrnthis paper, we compare the depth invariance of several cameras, andrnintroduce a new camera that improves in this regard over existingrndesigns, while still maintaining a good MTF.rnOur technique utilizes a novel optical element placed in the pupilrnplane of an imaging system. Whereas previous approaches use opticalrnelements characterized by their amplitude or phase profile, ourrnapproach utilizes one whose behavior is characterized by its scatteringrnproperties. Such an element is commonly referred to as anrnoptical diffuser, and thus we refer to our new approach as diffusionrncoding. We show that diffusion coding can be analyzed in a simplernand intuitive way by modeling the effect of a diffuser as a kernel inrnlight field space. We provide detailed analysis of diffusion codedrncameras and show results from an implementation using a customrndesigned diffuser.