Acoustic array data compression via Karhunen‐Loeve transform

Acoustic data compression has been explored in several contexts and several techniques exist. The applicability of each technique depends on the type of data processing that is intended. For example, audio data is often compressed via perceptual coding methods (e.g., mp3) in which quantization noise is distributed according to psychoacoustic principles. The Karhunen‐Loeve (KL) transform presents an opportunity to compress surveillance array hydrophone‐level data while preserving relevant features in beamformed displays. A KL codec was formulated and applied to acoustic test data from a horizontal line array. The results appear promising in that features of interest are preserved with significant data compression. The presentation will include a description of the algorithm as well as examples with actual data. [Work funded by ONR.]