Abstract
By a News Reporter-Staff News Editor at Robotics & Machine Learning DailyNews Daily News - A new study on Machine Learning - Comp utational Intelligence is now available.According to news reporting out of Beij ing, People’s Republic of China, by NewsRx editors, researchstated, “In low-lig ht image enhancement tasks, global structure and local texture details have diff erenteffects on illumination estimation. However, most existing works fail to e ffectively explore the intrinsicassociation within them.”Financial support for this research came from National Natural Science Foundatio n of China (NSFC).Our news journalists obtained a quote from the research from the China Universit y of Mining andTechnology Beijing, “To effectively balance the structure-preser ving and texture-smoothing for illuminationmaps, this paper introduces a new il lumination smoothing loss and proposes a self-supervised adaptiveillumination e stimation network (AIE-Net). The illumination smoothing loss achieves a balance betweenstructure-preserving and texture-smoothing mainly through L2 norm, trunc ated Huber, and Gaussiankernel function with color affinity. To construct AIE-N et, we introduce a local-global adaptive modulation(LGAM) module in deep featur e extraction. The module allows local and global features to be adaptivelyfused in a spatially varying manner by predicting scaling and adding factors. Finally , we separately estimatethe illumination maps for the input image and its inver ted image, and then achieve exposure correctionwith multi-exposure fusion.”
NETL
NSTL