A virtual inputting device comprises: a signal collection unit including a bioelectrical sensor for collecting bioelectrical signals and an acceleration sensor for collecting acceleration signals; a signal preprocessing unit for performing preprocessing for the bioelectrical signals and the acceleration signals collected by the signal collection unit; a signal segmentation unit for performing segmentation processing for the preprocessed bioelectrical signals and acceleration signals so as to obtain a plurality of gesture segments; a feature extracting unit for extracting feature values from the bioelectrical signals and the acceleration signals for respective gesture segments; a feature combination unit for combining feature values extracted from the feature extracting unit to form a combined feature vector; a gesture recognition unit for performing gesture recognition based on the combined feature vector; and a character mapping unit for obtaining characters corresponding to the recognized gesture according to a predetermined mapping relationship between characters and gestures.