The Matching Pursuit Decomposition (MPD) algorithm is a greedy feature extraction algorithm with many applications, especially in Structural Health Monitoring (SHM). The algorithm is powerful, but too computationally expensive for application on complex problems.

This research was completed to improve the scalability of the MPD algorithm. Three modifications were implemented including: correlation thresholding, coarse-fine grids, and multiple atoms extractions. These modifications successfully decreased the algorithm's time complexity and improved its scalablility for more complex problems.

Code to be uploaded soon.