Weighting methods, such as Hamming window, can suppress the peak sidelobe level of matching filter output of Linear Frequency Modulated (LFM) signal, but they also cause obviously mainlobe widening which results in range resolution worsening. Considering requirements of both sidelobe suppression and range resolution, a novel sidelobe suppression method without mainlobe widening is proposed. The amplitude outputs of matching filter and weighting window are firstly normalized, and then compared point by point, finally the minimum value of each point is chosen as the output data. This method fuses both merits of matching filter and weighting processing. The -3 dB mainlobe widening coefficient of the proposed method is only 1 compared with matching filter, and sidelobe suppressing performance is equivalent with the employed weighting window. Simulation results and lake experiment demonstrate the validity of the proposed method.