We establish new numerical radius inequalities for n × n operator matrices based on mixed Schwarz type inequalities. Using polar decomposition [Formula Presented] and continuous nonnegative functions f, g: [0, ∞) → [0, ∞), we develop a Cauchy Schwarz type numerical radius inequality for composite operator blocks of the form [Formula Presented] This unified approach extends several classical inequalities for operator matrices and yields the estimate[Formula Presented] where [Formula Presented] is an n ×n operator matrix and̂[Formula Presented] is an upper triangular scalar comparison matrix, whose entries depend only on the functions f and g. As an application, choosing suitable functions f and g together with a specific choice of the block operator [Formula Presented] leads to a new inequality involving the Moore-Penrose inverse. The resulting estimates recover several known inequalities and also provide new sharper bounds for the numerical radius of operator matrices.