We proposed a filtered q-ball imaging (fQBI) method for the reconstruction of fiber orientation distribution function (ODF) together with the quantitative comparison to unfiltered QBI. The filter kernel increases the high angular frequency content that is beneficial for the angular resolution in resolving crossing fibers. Through a series of simulations using Monte-Carlo model, the angular resolution of fQBI was demonstrated better than traditional QBI but the deviation of fiber orientation estimate also becomes larger. The improvement of the angular resolution can also reduce the underestimation of separation angles as well as the bias of fiber orientation estimations. In conclusion, fQBI was demonstrated to improve the angular resolution of QBI in resolving crossing fibers. This improvement will be helpful to precisely reconstruct fiber tract and brain network in applications by QBI.