The paper aims to study the feasibility of factor analysis to construct music "image schema", and provide a general design method for improving audio-visual communication. Bossa Nova's music visual design is taken as an example. With design interviews, the auditory sense of music is transformed into semantic description and image description. Schema factors are extracted, and then combined with factor analysis to construct the visual image schema of this musical style. Finally, different schemes are designed to verify the rationality of the image schema. As a potential structural cognition, the image schema can be accurately constructed by factor analysis. The application of the image schema constructed by factor analysis to the practice of audio-visual design can improve the accuracy and inclusiveness of sensory cross-mode conversion in the design process, thereby enhancing the matching degree between the design scheme and the perception of the group.