ABSTRACT
The use of tone mapping in High Dynamic Range (HDR) video applications needs to take into account different kind of temporal artifacts that can appear when processing frames of an HDR video independently. A useful operation for many video tone mapping operators (TMOs) is to extract statistics per frame of an HDR video. This chapter introduces the main techniques for achieving local TMOs that maintain temporal coherency globally and locally. Most of these techniques rely on spatio-temporal filtering and the use of optical flow. The operator is based on base and detail layer decomposition and it is composed of three steps. The first one is a tone mapping curve for reducing the dynamic range. The second step is an iterative and computationally fast spatial edge-aware filter inspired by both the bilateral filter and the anisotropic diffusion. The third step comprises further processing of the detail layer that is modulated relative to the visibility of noise in the tone mapped frame.
