Special Section on Graphics Interface ExpoBlend: Information preserving exposure blending based on normalized log-domain entropy Neil D.B. Bruce n University of Manitoba, Department of Computer Science, 66 Chancellors Circle, E2-408 EITC, Winnipeg, MB, Canada R3T 2N2 article info Article history: Received 10 June 2013 Received in revised form 18 September 2013 Accepted 2 October 2013 Available online 17 October 2013 Keywords: Exposure blending Illumination Reflectance Computational photography Tone-mapping Entropy abstract In this paper, we present a solution to the problem of dynamic range compression from multiple exposures called ExpoBlend that operates in the absence of raw format images, relative or absolute exposure values, camera response functions, or known irradiance. This is achieved in relatively simplistic fashion by merging image content across provided exposures. The proposed algorithm is directed at making visible any contrast appearing across a dynamic range that exceeds display or printing capabilities through high dynamic range (HDR) compression while preserving the nature of the image structure and detail, lighting, and avoiding introducing discontinuities in illumination or image artifacts. In addition, ExpoBlend allows scaling subject to a single parameter that elicits a trade-off between the impact of illumination and fine detail in the merged result. The strategy applied appeals to an information maximization strategy wherein the local entropy evident in each exposure is computed subject to a logarithmic compression of intensities, and employs cross-exposure normalization of entropy that implies a fusion strategy based on relative entropy across exposures in combination with a soft-maximum operation. & 2013 Elsevier Ltd. All rights reserved. 1. Introduction Recent efforts have produced many innovations in the field of computational photography ranging from algorithmic efforts to provide augmented depictions of content captured by visual sensors, to innovative contributions to the nature of the hardware, optics and electronics involved in capture and display. While recently there has been a gradual shift in emphasis to innovations in hardware technology, a problem of significant focus in the past decade has been in finding means of capturing visual content in a manner that carries a much higher dynamic range. The most common means of achieving this has been in capturing a number of bracketed exposures, which are subsequently combined subject to a known or estimated radiometric relationship between irra- diance and intensity, capturing imagery across a higher dynamic range than a single exposure might present. While this presents a rich set of data for analysis and display, it presents the problem of finding means of displaying HDR data in a manner suitable for lower dynamic range display technology. Existing research efforts present a multitude of distinct strategies devoted to the problem of compressing the range of visual imagery comprising a high dynamic range, down to a level suitable for this purpose. The exact goals of performing this operation are somewhat varied, with some efforts concerned primarily with producing images that are of a pleasing qualitative nature, providing realism while maximally pre- serving detail, or producing cartoon-like or detail augmented output. While these efforts serve as important tools in the arsenal of capture and display technology, the overall process can be labor- ious and tends to call for hardware that allows a minimum level of control and stability to derive high quality radiometric response mappings from distinct exposures, or estimation of irradiance from captured imagery. In some instances, hardware allows for capture of reasonably high quality imagery (e.g. recent HD webcams), but does not carry the same properties of allowing absolute (or relative) exposure to be known. Proprietary hardware in conjunction with proprietary drivers often results in conditions where the broad range of approaches for estimating the radio- metric response curves are unsuitable, producing either unstable or inaccurate results. Moreover, even in the presence of reasonably well behaved hardware for image capture, a means of combining independent exposures without the need to resort to careful measurements or calibration procedures is of value in providing the ability to present captured content on a standard display in a relatively quick and simple fashion. To this end, we present an algorithm for combining content across a number of given exposures with unknown image para- meters (e.g. exposure) in a manner that appeals to best capturing Contents lists available at ScienceDirect journal homepage: www.elsevier.com/locate/cag Computers & Graphics 0097-8493/$ - see front matter & 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.cag.2013.10.001 n Tel.: þ1 204 474 7313. E-mail addresses: bruce@cs.umanitoba.ca, neilbruce0@gmail.com URL: http://www.cs.umanitoba.ca/ bruce/ Computers & Graphics 39 (2014) 12–23