Under the theoretical framework of Kansei Engineering, relationship between visual design elements of automobile digital instruments and perceptual images of users were studied. Cluster analysis was used to select representative automobile digital instrument samples and appropriate perceptual image vocabulary, and semantic difference method was used to evaluate the perceptual image of sample images. Principal component analysis was used to reduce dimension, with a few variables represented overall data information. The visual design elements of automobile digital instrument were extracted，and the perceptual intention space was constructed. The quantification theory I was applied to build a prediction model of the relationship between visual design elements of digital instrument and user’s perceptual image characteristics. The relationship between the emotional appeal and the automobile digital instrument were quantified. The application of this model in the visual design process of automobile digital instrument can effectively improve the scientificity and objectivity of the visual design of automobile digital instrument, providing a more reliable theoretical basis for the visual interface design of automobile digital instrument.