Method of digital signal feature extraction comprising multiscale analysis

A method of feature extraction from a digital signal comprising:

• providing a plurality of samples of the digital signal;
• performing a plurality of iterations of steps including:
• a) segmenting the samples of the digital signal based on a grouping size to form a set of groupings each comprising a subset of the samples, wherein endpoints of each one of the groupings are spaced apart by the grouping size and the set of groupings collectively includes all of the samples;
• and
• b) applying a first operator, which is associated with a desired feature to be extracted, to the subsets of the samples to derive a plurality of representative values the set of groupings;
• wherein the first operator is adapted to correspond to the grouping size;
• wherein the grouping size of each iteration is different;
• and
• wherein, in step a) of at least one of the iterations, one of the endpoints of at least one of the groupings is intermediate the endpoints of another one of the groupings of the samples such that at least two of the groupings of the set of groupings overlap.