Accurate and efficient plume masking is essential for remote sensing-based detection and quantification of methane and other point source emissions, as plume masks are critical not only for quantifying emission rates, but also for visualization and source localization. However, plume masking relies largely on human operation when the retrieved plume concentrations are weak relative to the background, which hinders the automatic plume detection. This study presents an automatic plume masking method based on wavelet transform image processing. Given a methane concentration enhancement image with no prior knowledge of source locations, a 2D discrete wavelet transform is applied to enhance plume signals while suppressing background noise. The binary plume masks are then generated and filtered using criteria such as concentration, plume shape, and wind direction. The method includes tunable parameters to ensure high detection accuracy under varying background and meteorological conditions. This method detected 60 % more plumes, mainly with lower fluxes, than a thresholding method from both MethaneAIR and MethaneSAT data, while finding fewer false positives, proving its potential to realize automatic plume detection across platforms at different scales and resolutions. Its high sensitivity to low-volume emissions also enables a lower detection limit and provides a more comprehensive emission rate distribution. Compared to machine learning models, this method is computationally efficient and does not require training data. Although designed for MethaneSAT purposes, this method is broadly applicable for plume detection from concentration imagery on various airborne and spaceborne remote sensing platforms and for numerous atmospheric species.