We study its expression in sculpture modeling by excavating traditional cultural imagery in sculpture art. In this paper, 20 students with design-related backgrounds were selected as subjects to view sculpture images for cultural imagery evaluation, and imagery selection, response duration, and eye-movement data were collected using Yulin Grottoes statue sculptures as eye-movement experiment samples. The collected data were then used to investigate various imagery expression methods and anticipate the traditional cultural imagery appearing in the sculptures using the DE-GWO model. With a mean square error of 0.275 for the typical user sample and the results of the DE-GWO analysis showing a p-value range of 0.446 to 0.05, with no significant differences, the degree of fit and predictive power between the predicted and actual values of the test set was generally good for subjects with relevant backgrounds. The presentation of sculpture art language under the multicultural viewpoint explores the expressive form of traditional cultural images, which has positive relevance to the development of sculpture art.