Topology optimization using super-resolution image reconstruction methods

This paper proposes a new topology optimization method to obtain super-resolution images without increasing mesh refinement by using various methods. For traditional process, low-resolution (LR) images are fed into the Solid Isotropic Material with Penalization (SIMP) and Optimality Criteria (OC) methods. Here, the trained super-resolution images are added to the inner loops to reconstruct the topology and used to obtain high-resolution (HR) images from the LR images at the end of each iteration. After finishing the reconstruction process, the main topology optimization method recovers the original size images from the HR images for the next iteration. Several examples are presented to demonstrate the effectiveness of the proposed method. The final topologies provide noticeably improvement over those of typical SIMP method and create a much sharper and higher contrast images. Moreover, the proposed strategy using the super-resolution image reconstruction methods can give valuable innovation for conventional topology optimization process.