An innovative calibration based integral photography rendering algorithm for medical application and its evaluation

Autostereoscopic has long been proposed to fulfill medical display in image-guided surgery and clinical education to provide more intuitive position information of clinical interest zone thus improving surgery safety and accuracy. As one category of flexible autostereoscopic 3D display, computer generated integral photography (CGIP) has been studied in medical application by many researches for its convenience and cost-efficiency. However, IP still suffers from inaccurate light field reconstruction, which limits its practicality in surgery. In this paper, we propose and apply a flexible fish-eye model based micro lens array (MLA) distortion calibration method and pre-distorted retracing rendering algorithm to render elemental image array (EIA) of CGIP. Furthermore, we also evaluate light field of the proposed algorithm in depth cue, and signal noise ratio of IP images by phantom experiment.