High-speed cameras are prone to overexposure in complex scene shooting due to their fast shutter speed and significant influence of light. This research on the overexposure image enhancement algorithm based on multi-scale decomposition proposes a multi-scale enhancement algorithm combining the Laplacian pyramid with the CBDnet model to improve the overexposure phenomenon that occurs during the shooting process. The algorithm was implemented through self-built image databases and public datasets, and it was proved in experiments that the algorithm demonstrated better enhancement effects in multiple indicators.