This article investigates anti-swing control of tower cranes subject to the double-pendulum effect. Generalized coordinates are introduced as the fundamental fully actuated variables (FAVs) to construct a baseline fully actuated system (FAS), providing a direct foundation for controller design. The anti-swing objective is realized through a two-loop structure. In the inner loop, an elaborate filter is incorporated, with the filter parameters optimized using sequential quadratic programming (SQP) to balance swing-attenuation motion and positioning motion. The filtered signals are then coupled with the FAV to form an improved FAS, whose closed-loop dynamics constitute the outer loop of the control strategy. A comprehensive index is further proposed to characterize both the convergence of the FAV and its influence on swing motion, with the control gains determined by minimizing this index. Experimental results confirm that the proposed scheme effectively suppresses swing motion while ensuring precise positioning.