Finite impulse response (FIR) digital filter is a commonly adopted signal processing unit in digital signal processing, due to its stability and easy implementation for linear phase response. To reduce the area complexity and power consumption of the filters, researchers have proposed algorithms to optimize the expression of coefficients with the reduced number of non-zero digits or power-of-two terms. This paper presents a new optimization algorithm based on elegant dynamic programming approach to minimize the number of non-zero digits in coefficient set, which yields to low logic operator cost in the filter circuit implementation. The proposed algorithm utilized the knapsack method from dynamic programming to effectively remove the redundant nonzero digits in the coefficients. Experiment results on benchmark filters show that the proposed algorithm can synthesize FIR filter coefficient with the reduced area complexity. Compared with two competing methods, the proposed algorithm can design FIR filters with an average of 14.49% to 47.73% reduced full adder cost over the competing methods.