Abstract:High Dynamic Range (HDR) imaging via multi-exposure fusion is an important task for most modern imaging platforms. In spite of recent developments in both hardware and algorithm innovations, challenges remain over content association ambiguities caused by saturation, motion, and various artifacts introduced during multi-exposure fusion such as ghosting, noise, and blur. In this work, we propose an Attention-guided Progressive Neural Texture Fusion (APNT-Fusion) HDR restoration model which aims to address these issues within one framework. An efficient two-stream structure is proposed which separately focuses on texture feature transfer over saturated regions and multi-exposure tonal and texture feature fusion. A neural feature transfer mechanism is proposed which establishes spatial correspondence between different exposures based on multi-scale VGG features in the masked saturated HDR domain for discriminative contextual clues over the ambiguous image areas. A progressive texture blending module is designed to blend the encoded two-stream features in a multi-scale and progressive manner. In addition, we introduce several novel attention mechanisms, i.e., the motion attention module detects and suppresses the content discrepancies among the reference images; the saturation attention module facilitates differentiating the misalignment caused by saturation from those caused by motion; and the scale attention module ensures texture blending consistency between different coder/decoder scales. We carry out comprehensive qualitative and quantitative evaluations and ablation studies, which validate that these novel modules work coherently under the same framework and outperform state-of-the-art methods.

High Dynamic Range Image Hallucination

Exposure Correction and Detail Enhancement for Single LDR Image

Highly Robust Real-Time Image Enhancement

Multi-exposure image fusion based on well-exposedness assessment

Correction of Over-Exposure Using Dark Channel Prior and Image Fusion Technique

Correction of overexposure utilizing haze removal model and image fusion technique

Ghost Detection And Removal Based On Super-Pixel Grouping In Exposure Fusion

Texture and Exposure Awareness Based Refill for HDRI Reconstruction of Saturated and Occluded Areas

Real Color Image Enhanced by Illumination - Reflectance Model and Wavelet Transformation

High dynamic range imaging by a pseudo exposure fusion method based on artificial remapping

Hallucinating Face by Position-Patch

Robust Ghost-Free High-Dynamic-Range Imaging by Visual Salience Based Bilateral Motion Detection and Stack Extension Based Exposure Fusion.

Real-Color Image Denoised and Enhanced Synchronously Based on Wavelet Transform

Probabilistic exposure fusion.

Visual Salience And Stack Extension Based Ghost Removal For High-Dynamic-Range Imaging

Hallucinating facial images and features

A New Technique for Face Image Resolution Enhancement Based on Face Hallucination

High Dynamic Range Imaging of Dynamic Scenes with Saturation Compensation but without Explicit Motion Compensation

Generation of High Dynamic Range Illumination from a Single Image for the Enhancement of Undesirably Illuminated Images

High Dynamic Range Image Capturing And Displaying

Attention-Guided Progressive Neural Texture Fusion for High Dynamic Range Image Restoration