Performances of an highly parallel image processing chain for capturing high dynamic range video
Michele Benetti, Arnaud Peizerat, Antoine Dupret, Leonardo Bruno De, Michaël Tchagaspanian · 2014
In this paper we address the process that permits to obtain a High Dynamic Range (HDR) video starting from a set of different exposed Low Dynamic Range (LDR) acquisitions. A hard constraint in the acquisition process is represented by the video rate, that limits the number of LDR images that can be acquired with given hardware resources, such as data throughput or buffer memory. We present an HDR reconstruction image Processing Chain (PC) that has been conceived highly parallel: it measures the image by blocks with dedicated LDR. We present results from the PC hardware implementation and evaluated performances.