<p>Biological structures are renowned for their multifunctionality, with honeycomb structures being a prime example, known for their lightweight and directional superior energy absorption capabilities. To integrate biological features into structures and achieve adjustable feature scales, an image–physics&#xa0;driven topology optimization method (IPTO) is proposed. In this method, “image” represents the biological feature image, while “physics” refers to the optimization process based on the solid isotropic material with penalization (SIMP) framework. The visual geometry group 19-layer network (VGG19) is employed to extract biological features, such as honeycombs and fish scales, and integrate them into the SIMP framework to minimize compliance, thereby embedding biological features into materials. To adjust the size of biological features in the generated structure, a block periodic filling strategy is adopted. Features extracted by the VGG19 are adaptively reduced with network depth <i>l</i>, and empty regions are filled to achieve a stepwise reduction of feature size with increasing depth. To adjust the density of biological features in the generated structure, a biological feature density control weight <i>β</i> is introduced, allowing effective adjustments to the density of the biological features. In addition, refer to images of earthworm skin, cuttlebone, teeth, a combination of cuttlebone and teeth, and leaf venation to further validate the effectiveness of the suggested method. Finally, static pressure tests are conducted to assess the energy absorption advantages of biological features, taking the optimized honeycomb structure as an example. Compared to the traditional SIMP structure, the optimized honeycomb structure has a 17.4% improvement in energy absorption, with only a 2.3% increase in compliance. These results demonstrate the potential of IPTO for optimization of multifunctional structures.</p>