In addition to the hardware methods discussed in Chapter 3, software methods have also been developed (note that there are algorithms other than the ones presented here; many are proprietary). These methods are of great interest in photography, computer graphics, and scientific applications that require extreme dynamic ranges but need to image a static (or almost static) scene. Software methods can give much better-looking results than hardware methods, but they have several disadvantages. The first and most important disadvantage is that they require several images of the same scene, which is obviously not possible if the scene contains moving objects or variable illuminations (for example, numerous lights are a good example of varying illumination because their intensity is related to the frequency of the electricity). Secondly, the computation time, even for reasonable image resolutions and on powerful computers, is long. High-definition (HD) images are typically processed in several seconds. It is not realistic to think about a fast and efficient embedded solution using the software method described in this chapter. However, for certain applications, there are specific software methods based on very few exposures that can be used in a powerful embedded system. Finally, software methods create better pictures for people to look at, but they fail to record accurate scene data. It can be demonstrated that, with the exception of very low-glare scenes, multiple exposures fail to increase the range of accurate scene measurement.
Online access to SPIE eBooks is limited to subscribing institutions.