Abstract: Traditional iterative optimization algorithms cannot guarantee that the resultant mesh does not have inverted elements. In order to solve this problem, a novel quadrilateral mesh optimization algorithm guaranteeing non-inverted elements is proposed. Firstly, the mesh is optimized by the Laplacian smoothing method. Then the local region to be handled is determined. After that, the local region is layered based on topology structure and the nodes in the region are reset layer by layer. Finally, the optimization problem is converted into a constrained optimization problem with a feasible initial value to solve. Experimental results demonstrate the effectiveness of the proposed algorithm.