The invention relates to the technical field of
signal processing, in particular to a magnetoencephalogram decoding method based on image features. The method comprises the steps: giving a
signal stimulation to the brain, obtaining the needed magnetoencephalogram data, carrying out the preprocessing of the magnetoencephalogram data, carrying out the spatial filtering of the magnetoencephalogram data, and obtaining the magnetoencephalogram data; the spatial filtering is used for reducing the dimension of the
evoked potential, improving the
signal-to-
noise ratio of the
evoked potential and further virtualizing the magnetoencephalogram data, the virtualized image is small in data volume during
processing, the space-time dynamic process of magnetoencephalogram signals is considered in the formation of grids, that is, the spatial signals are arranged according to the
time sequence, the classification precision is improved, and the classification efficiency is improved.
Gist features are selected as image features of magnetoencephalogram decoding, due to the fact that the
Gist features are biologically inspired features, a good effect is achieved in image classification, finally, a linear
support vector machine classifier is adopted for classifying magnetoencephalogram signals, and the method solves the problem that in the prior art, precision is not high when magnetoencephalogram data are classified.