【正文】
is based on the 1to1 mapping defined by the main camera lens. The idea of replacing the array of pinholes in front of the ?lm with lenses was ?rst proposed by Lippmann back in 1908 [3]. Just as with a single pinhole camera, lenses gather much more light and produce better image quality than small holes. Lipmann called his approach Integral photography. Di?erent versions of it have been proposed throughout the years, the most recent one being the plenoptic camera [6,14]. Our analysis of the integral camera in frequency space will be done in two steps. (1) We consider an array of pinholes as in the Ives’ camera, only shifted by a constant (for all pinholes) vector a. Each pinhole is covered by a prism with angle of deviation depending on the shift, de?ned as prism = a/f .(2) We consider the superposition of multiple shifted arrays of such pinholeprisms, and show that they all contribute to the ?nal image in a /f similar way. Lippmann’s integral photography is based on this coherent action of di?erent arrays. It can be viewed as the limiting case where the plane is made pletely of 36 pinholeprisms and all the light goes through. Each microlens is formed by the corresponding prisms, as a Fresnel lens. Wideangle lens has