Art education for preschool children is the basis for the cultivation of young children’s perceptual ability, and the cultivation of young children’s perceptual ability plays an essential and fundamental role in young children’s understanding of the world, feeling of beauty, and discovery of beauty. This paper uses AR technology and microclassrooms in the context of new media to construct an educational model of preschool children’s art curriculum, and a portrait of preschool children’s scholars is built. It focuses on combining the hierarchical clustering algorithm with the radial function method of multivariate interpolation to analyze the classroom behaviors of preschool children, mining the correlation between the course learning behaviors and the learning effect through the Apriori algorithm, and constructing a personalized art learning path for preschool children based on the ant colony algorithm for practice. The results showed that preschool children were divided into four categories, with the highest proportion being excellent preschool children (52.5%). The mean score of preschool children’s art perception score was 8.225, and their scores were normally distributed, with the highest number of preschool children between 8.00 and 8.25. This study is significant for changing the art curriculum in preschool children’s education. Teachers should be guided in teaching activities and focus on children’s perceptual experiences and expressions.