It Takes Two: A Duet of Periodicity and Directionality for Burst Flicker Removal
Lishen Qu, Shihao Zhou, Jie Liang, Hui Zeng, Lei Zhang, Jufeng Yang
2026-04-01
Summary
This paper addresses the problem of 'flicker' in photos taken with short exposure times, which creates noticeable and distracting visual artifacts. The researchers developed a new system called Flickerformer to remove this flicker and improve image quality.
What's the problem?
When you take a quick photo, sometimes the lighting isn't perfectly consistent across the entire image, or the camera doesn't capture all parts of the image at exactly the same moment. This causes 'flicker,' which looks like uneven brightness or ghosting. Existing image editing tools aren't great at fixing flicker because it's a unique problem – it has a repeating pattern and a specific direction, unlike simple noise or darkness. Current methods often either don't remove the flicker completely or create new, unwanted artifacts like ghosting.
What's the solution?
The researchers realized flicker has two key characteristics: it repeats in a pattern (periodicity) and appears to move in a certain direction (directionality). They built Flickerformer, a system based on a powerful type of artificial intelligence called a 'transformer.' Flickerformer has three main parts. First, it compares frames to find repeating patterns. Second, it looks for similar structures *within* each frame. Finally, it uses a technique called 'wavelet analysis' to pinpoint the direction of the flicker and fix the dark areas precisely, without causing ghosting. Essentially, it intelligently analyzes the image to understand and remove the flicker's unique characteristics.
Why it matters?
This work is important because it provides a much better way to fix flicker in photos, especially those taken in challenging lighting conditions. By specifically addressing the patterns and direction of flicker, Flickerformer produces clearer, higher-quality images than previous methods, and avoids the common problem of introducing new artifacts during the correction process. This is useful for anyone who takes photos, especially in situations where consistent lighting is difficult to achieve.
Abstract
Flicker artifacts, arising from unstable illumination and row-wise exposure inconsistencies, pose a significant challenge in short-exposure photography, severely degrading image quality. Unlike typical artifacts, e.g., noise and low-light, flicker is a structured degradation with specific spatial-temporal patterns, which are not accounted for in current generic restoration frameworks, leading to suboptimal flicker suppression and ghosting artifacts. In this work, we reveal that flicker artifacts exhibit two intrinsic characteristics, periodicity and directionality, and propose Flickerformer, a transformer-based architecture that effectively removes flicker without introducing ghosting. Specifically, Flickerformer comprises three key components: a phase-based fusion module (PFM), an autocorrelation feed-forward network (AFFN), and a wavelet-based directional attention module (WDAM). Based on the periodicity, PFM performs inter-frame phase correlation to adaptively aggregate burst features, while AFFN exploits intra-frame structural regularities through autocorrelation, jointly enhancing the network's ability to perceive spatially recurring patterns. Moreover, motivated by the directionality of flicker artifacts, WDAM leverages high-frequency variations in the wavelet domain to guide the restoration of low-frequency dark regions, yielding precise localization of flicker artifacts. Extensive experiments demonstrate that Flickerformer outperforms state-of-the-art approaches in both quantitative metrics and visual quality. The source code is available at https://github.com/qulishen/Flickerformer.