Evaluation of HDR Algorithms
John J. McCann, Alessandro Rizzi · 2011
Spatial image processing, such as Retinex, ACE, and spatial-frequency filters use the entire image in rendering scenes. Over the years there have been a number of different quantitative test targets used to evaluate HDR algorithms. This chapter discusses four such approaches to measure quantitatively the success of an algorithm: lightness test targets, ratio metric, quantitative evaluation of 3-D Mondrians, and locality test targets. It presents a block diagram of options for evaluating spatial algorithms. The process begins with a scene that is captured by a camera and processed with an algorithm. The four independent branches (faithful reproduction, beauty contest, time and calculation efficiency, and mimic vision) for evaluating four distinct goals are outlined in the chapter. Finally, the chapter discusses the lessons learnt from the quantitative studies of HDR in cameras. Controlled Vocabulary Terms chemical analysis; image processing; Spatial filters