Additive manufacturing is well-suited for topology optimization given its capability to produce complex geometries. Yet optimization methods frequently result in designs with features, like thin bars, that are difficult to print. Consequently, a post-optimization phase is usually necessary to correct these defects and ensure the part is manufacturable. This presentation introduces a novel approach to enforce minimum member size in topology optimization. Unlike conventional methods that often depend on additional finite element analyses or complicated filters, this strategy addresses length scale control through a geometric perspective. We propose a methodology based on local perimeter control to strictly regulate feature sizes. This technique utilizes standard filtering processes, which facilitates implementation and improves computational efficiency. The core of the method involves a secondary smoothing operation applied to a perimeter representation derived from density variables. By maintaining the smoothed local perimeter below a theoretical threshold, the optimization algorithm ensures that both solid and void phases respect the prescribed filtering size. Furthermore, this local formulation offers the advantage of curvature control, effectively rounding sharp edges.