With the increasing number of channels, fabricating a multi-channel fiber Bragg grating (FBG) filter requires a considerably high maximum index modulation, which is beyond a physically realizable level. Hence, an effective optimization method based on particle swarm optimization (PSO) algorithm and direct design method to design a multi-channel FBG filter is proposed. By introducing the tailored group delay coefficients into the target reflectivity spectrum, a mathematical model aiming at minimizing the maximum index modulation of the grating is established. The PSO algorithm is employed to find the optimal group delay parameter for each channel. It leads to a more even distribution of the refractive index modulation, thereby reduces the maximum index modulation to a physically realizable level. Design examples of the 40-channel and 106-channel FBG filters both with uniform reflection spectra demonstrate that the proposed method yields a remarkable reduction in the maximum index modulation which is below 0.001.