Rehabilitation training system for advanced brain dysfunction

The invention relates to the field of computer vision, in particular to a rehabilitation training system for advanced brain dysfunction. Obtaining a key point center matrix and a facial center feature value vector according to the key point data sequence in the facial expression image of the patient; obtaining a feature discrete difference coefficient according to the key point center matrix; correcting the face center feature value vector in combination with the feature discrete difference coefficient to obtain a face inherent feature vector; according to the left eyebrow curve, the right eyebrow curve and the gray scale information, constructing an eyebrow guiding coefficient; combining eyebrow, eye and mouth guide coefficients with the face inherent expression vector to obtain a face expression guide matrix; and inputting the facial expression image of the patient and the facial expression guide matrix into a neural network to obtain a facial expression category of the patient, and completing rehabilitatio