High dynamic range (HDR) imaging is a well-established technology that enables the acquisition, storage, manipulation, delivery, and evaluation of a higher dynamic range than the one available in traditional 8-bit per color channel technology. This has brought several advantages, such as more realistic color reproduction, more details in bright and dark areas, better contrast, improved colors, etc. This has revolutionized the way we are now experiencing entertainment, as well as the way we are using images and videos in the image processing and computer vision fields.
We are experiencing a large use of HDR technology in the entertainment sector, and are also starting to see its use in industrial applications. On the other hand, we are assisting in a paradigm change in the image-processing area, where traditional techniques are surpassed by more flexible deep-learning-based approaches. In the last few years, we are also observing this specific trend in the HDR imaging field. This has brought a number of challenges that need to be addressed in order to make deep-learning-based HDR approaches more robust and resilient to unseen data and/or data which is too noisy.
More information: