To address the issues of low accuracy and high computational complexity in traditional channelization techniques for ultra-wideband signals, this paper proposes a novel rationally oversampled channelization method to enhance the accuracy and efficiency of signal processing. The proposed method is evaluated by implementing and comparing critically sampled and integer oversampled channelization algorithms. A detailed analysis of the impact of different oversampling factors and filter orders on performance is provided. The validity of the proposed algorithm is verified using baseband data from pulsar J0437−4715 observed by the Parkes telescope, demonstrating its effectiveness and correctness.