Curvilinear mask has received much attention in recent years due to its ability to obtain better image quality in advanced nodes. A common method for optimizing curvilinear mask in optical proximity correction (OPC) flow is moving control points on the edge directly (MCED-based OPC), but it requires storing mass data. This paper uses distance-versus-angle signature (DVAS), a one-dimensional function, to represent a two-dimensional boundary of mask. To the best of our knowledge, DVAS and its second derivative are applied for the first time to reduce the design data storage space and guide the direction of optimization in EUV lithography systems. Compared with storing the change distances of the control point in both horizontal and vertical directions, directly storing the amplitude change of the DVAS can effectively reduce the storage space consumption. Simulation results demonstrate that unlike the MCED-based OPC method, the DVAS-based OPC method using second derivative achieves a lower pattern error and requires less time while reducing the storage space.