Nowadays many cameras embed multi-imaging (MI) technology without always giving the option to the user to explicitly activate or deactivate it. MI means that they capture multiple images, combine them and give a single final image, letting sometimes this procedure being completely transparent to the user. One of the reasons why this technology has become very popular is that natural scenes may have a dynamic range that is larger than the dynamic range of a camera sensor. So as to produce an image without under- or over-exposed areas, several input images are captured and later merged into a single high dynamic range (HDR) result. There is an obvious need for evaluating this new technology. In order to do so, we will present laboratory setups conceived so as to exhibit the characteristics and artifacts that are peculiar to MI, and will propose metrics so as to progress toward an objective quantitative evaluation of those systems. On the first part of this paper we will focus on HDR and more precisely on contrast, texture and color aspects. Secondly, we will focus on artifacts that are directly related to moving objects or moving camera during a multi-exposure acquisition. We will propose an approach to measure ghosting artifacts without accessing individual source images as input, as most of the MI devices most often do not provide them. Thirdly, we will expose an open question arising from MI technology about how the different smartphone makers define the exposure time of the single reconstructed image and will describe our work around a timemeasurement solution. The last part of our study concerns the analysis of the degree of correlation between the objective results computed using the proposed laboratory setup and subjective results on real natural scenes captured using HDR ON and OFF modes of a given device.
Martin Renaudin, Anna-Cecilia Vlachomitrou, Gabriele Facciolo, Wolf Hauser, Clement Sommelet, Clement Viard, Frédéric Guichard, "Towards a quantitative evaluation of multi-imaging systems" in Proc. IS&T Int’l. Symp. on Electronic Imaging: Image Quality and System Performance XIV, 2017, pp 130 - 140, https://doi.org/10.2352/ISSN.2470-1173.2017.12.IQSP-230