Scalar-field reconstruction algorithms using plenoptic cameras

Several volumetric reconstruction algorithms are surveyed and evaluated using plenoptic images of a burner flame. Automatic generation of static and dynamic masks is implemented to limit the scope of the reconstruction problem. Volumetric reconstructions of the flame are performed using all surveyed algorithms with a number of plenoptic cameras ranging from one to four. Simultaneous Algebraic Reconstruction Technique (SART), a reconstruction algorithm widely used in medical imaging, is identified as the most promising algorithm of those surveyed. When combined with inter-iteration minimization of L1 Total Variation (TV-L1), SART-TV reconstructions of a burner flame are shown to be smooth, circular, and uniform. Two-camera reconstructions recover the approximate shape of the flame, yielding qualitative information about the flame structure. Three-and four-camera yield the full three-dimensional flame structure.