This project investigated two versions of Dynamic Mode Decomposition (DMD) for the disturbance wave feature identification in thin liquid films. The plain DMD method showed great promise in identifying low-rank structure in a time-series data used for film thickness measurements. Due to the type of image used in this analysis, large disturbance waves were not obvious, but small ripples were. The latter seemed to be identified by the plain DMD algorithm. However, plain DMD tends to perform poorly at extracting features (i.e. modes) that are translational and rotational invariant. A recursive approach called the multi-resolution DMD was implemented and tested with an image dataset that shows disturbance waves more clearly. Compared to the results from performing plain DMD on the same dataset, mrDMD shows time-specific spatial modes of the waves – a capability difficult to obtain from alternative methods such as the Discrete Fourier Transform. The time frequency of the detected wave features should be compared to manual measurements to validate the mrDMD methods.