Expansion of ultrasonic signals in Karhunen-Loeve basis for data compression

摘要

Data compression of discrete scattered ultrasonic signals from polycrystalline AISI 304 stainless steel and aluminium was achieved by Karhunen-Loeve Transform (KLT). When these signal were reconstructed from the data sets in their compressed domain, they showed very small error. We could reduce the dimensionality of these signals and we suggest that this is the intrinsic dimensionality of the dynamics of the interaction of the acoustic wave with the microstructures of these materials. We have also tried to prove that the dimensions that were discarded for compressing these signals using KLT essentially arise due to instrumental noise.