The invention discloses a
broadband noise suppression method based on low-rank
tensor approximation, and relates to the technical field of
signal processing, and the method comprises the steps: obtaining an original
vertical array receiving
signal, and carrying out the preprocessing of the
signal, and obtaining a
time domain signal; constructing a three-dimensional observation
tensor related to depth, time and frequency according to the
time domain signal; and restraining a low-rank signal
tensor in the three-dimensional observation tensor, a
pulse noise tensor in the three-dimensional observation tensor and a
Gaussian noise tensor in the three-dimensional observation tensor, thereby obtaining a low-rank target tensor. According to the method, the traditional matrix
processing is expanded to the high-dimensional tensor space, and the
noise reduction problem of the original
vertical array receiving signal is expressed as the problem of
Gaussian noise and pulse interference separation from the low-rank
target signal under low-rank tensor approximation, so that the
noise suppression is performed on the original
vertical array receiving signal; and then a pure low-rank target tensor is separated out.