Mcgill University

Heather L Smith

Henrietta L Galiana

For segmenting a sampled signal having at least two temporally separate interleaved dominant components for the purposes of extracting one or more of the separate components, an automated method of analyzing the sampled signal is used. By selecting a model for the signal and a processing window dimension, a variance between the signal and a model value for the signal is measured within the window over the sampled domain to obtain a noise indicator value. A corner geometry value for the sampled signal is calculated within the same window over the domain. A transition indicator value is generated based on a ratio of the corner geometry value and the model value. The segmentation points are identified based on the transition indicator value and at least one of the noise indicator value and the model value within the window over the domain.

Growing community of inventors

Montreal, Canada

Heather L Smith