A Variational Method for Flicker Removal in High Speed Video Sequences

Deflickering of color image sequences constitutes a fundamental task in many video applications. Recently, it has received increased attention with the popularity of consumer-level high speed video acquisition devices, for instance in newer smartphone generations or sports/action cameras. These are capable of acquiring videos at more than 100 frames per second. In this context, flicker corresponds to undesirable intensity and chroma variations due to the interaction between the acquisition frequency on the one hand, and the alternating current powering artificial light sources on the other. Non-uniform periodic flicker is not easy to correct since its estimation requires object tracking, and since most tracking techniques assume consistent illumination. In this paper, we adopt a new optimization approach to solve this problem. We propose causal and non-causal robust correction schemes using a block matching technique paired with color variation estimation. A post processing step is introduced to remove block artifacts in an efficient manner. A multi-resolution approach is also developed to reduce computation time without affecting the quality of our results. We demonstrate the efficacy of our method both on simulated and real data.