Level set-based topology optimization considering aesthetic preferences based on texture energy

Abstract

For improving consumer satisfaction, a design process for individual production based on additive manufacturing technology is required to consider appearance as well as functionality. Although topology optimization is a powerful technology to design highly functional structure, it has difficulty considering aesthetic features. This paper presents a new topology optimization method considering aesthetic preferences with a manufacturing constraint by incorporating the image processing used for style transfer and object recognition. To consider aesthetic features, we introduce texture energy which evaluates the similarity between the input preference image and structure represented by the level set method. To identify the unmanufacturable regions disconnected from the main structure, the connected component labeling process based on the object recognition method is applied to the binary image of the level set function. A topology optimization problem of maximizing stiffness is formulated considering aesthetic preferences and imposing the structural connectivity constraint, where the objective function is defined as a combination of minimizing mean compliance and texture energy. A reaction diffusion equation is used to update the level set function, where the Lagrange multiplier of structural connectivity constraint is calculated to eliminate unmanufacturable disconnect regions. Numerical examples are provided to confirm the validity and utility of the proposed method.

Keywords

• optimum design
• topology optimization
• level set method
• style transfer
• texture energy
• structural connectivity constraint