Single-shot high dynamic range (HDR) imaging often relies on block-wise varying exposures to capture the full range of a real-world scene. According to the paper, this approach is effective, but it can suffer from brightness order reversal (BOR) artifacts, blocking artifacts, and loss of details.
The work, titled "Neural Block-wise Auto-exposure for Single-Shot High Dynamic Range Imaging," proposes a neural block-wise auto-exposure method to tackle these issues. The abstract identifies the artifacts as the key challenges, though the available text cuts off before detailing the architecture or experimental results.
Because only a single source was provided, there is no independent comparison to other methods. The claims are limited to what the abstract states.