Non-Local Pairwise Energy Based Model For The HDR Image Compression Problem

We present a new energy based compression model for the display of high dynamic range images. The proposed tone mapping method tends to exploit the biologically-inspired dynamic retina concept which is herein mathematically expressed via an image representation based on the specification of the statistical distributions of the non-local gradient magnitude. In this framework which also operates the notion of non-local gradient recently put forward by Gilboa and Osher, the detailpreserving contrast reduction problem is therefore expressed by a energy based model with non-local pairwise pixel interactions defined on a complete graph whose cost function is locally minimized by a conjugate gradient descent procedure. The experiments reported in this paper demonstrate that the proposed compression method is efficient and provides pleasing results on various images with different scene contents and performs competitively compared to the best existing state-of-the-art tone mapping methods recently proposed in the literature.