The work aims to expand the access to knowledge related to multi-image design of product families and improve the efficiency of product family modeling innovation design. Based on the coupling characteristics of modeling images about product family and in combination with the concept of multilayer networks in complex networks, a product family modeling imagery coupling network model was constructed by using the multi-image relationship between modeling features. The coupling network model was analyzed with the node sensitivity and centrality methods. The attribute type of each modeling feature node was identified as extracted knowledge. Further, the types of modeling feature nodes applicable to the product family design were summarized and their type application rules were formulated in combination with the characteristics of product family design, so as to assist designers to carry out the product family modeling innovation design. Taking the modeling design of Mazda's front face as an example, a modeling imagery coupling network model of Mazda's front face was constructed, then relevant design knowledge of product family was acquired by analyzing it and successfully applied to the design of the front face. The product family modeling imagery coupling network model enriches the knowledge of product family multi-image design. It can be used to assist the product family design and improve the design efficiency.