Related Experiment Video
Updated: Aug 4, 2025
10:14
3D Scanning Technology Bridging Microcircuits and Macroscale Brain Images in 3D Novel Embedding Overlapping Protocol
Published on: May 12, 2019
7.4K
Hybrid High Dynamic Range Imaging fusing Neuromorphic and Conventional Images
Summary
This study introduces NeurImg-HDR+, a hybrid imaging system combining neuromorphic and RGB cameras. It reconstructs high dynamic range (HDR) images and videos by fusing data from both sensors, overcoming limitations of individual camera types.
Area of Science:
- Computer Vision
- Image Processing
- Computational Photography
Background:
- Conventional RGB cameras struggle with high dynamic range (HDR) scenes, leading to over- or under-exposure.
- Neuromorphic cameras offer HDR capabilities but lack spatial resolution and color information.
Purpose of the Study:
- To develop a hybrid imaging system (NeurImg) that fuses data from neuromorphic and RGB cameras.
- To reconstruct high-resolution, high dynamic range images and videos by bridging sensor domain gaps.
Main Methods:
- Proposed the NeurImg-HDR+ network with specialized modules for sensor data fusion.
- Captured a novel dataset of hybrid signals from various HDR scenes.
- Compared the fusion strategy against state-of-the-art inverse tone mapping and LDR merging methods.
Main Results:
- Demonstrated the effectiveness of the NeurImg-HDR+ network in reconstructing high-quality HDR images and videos.
- Quantitative and qualitative experiments confirmed superior performance on both synthetic and real-world data.
- Successfully addressed resolution, dynamic range, and color representation discrepancies between sensors.
Conclusions:
- The proposed hybrid imaging system effectively reconstructs high-resolution, high dynamic range images and videos.
- Fusion of neuromorphic and RGB camera data offers significant advantages over existing methods.
- NeurImg-HDR+ provides a robust solution for capturing and processing challenging HDR visual scenes.
