To meet the demand for high-precision and high spectral resolution data in molecular spectral line observations, this study proposes and implements a parallel channelization algorithm for molecular spectral line signals based on an oversampled polyphase filter bank (OSPFB). The proposed method effectively overcomes the limitations of the critically sampled polyphase filter bank (CSPFB), particularly in terms of subband edge attenuation and spectral leakage. Using simulated baseband data, we conduct a comparative analysis of CSPFB and OSPFB under two scenarios: signals located at channel edges and signals distributed across adjacent frequency channels. The evaluation focuses on anti-aliasing capability and information recovery performance during the channelization process. Results demonstrate that, compared with CSPFB, OSPFB significantly reduces scalloping loss and spectral leakage, thereby improving the signal-to-noise ratio (SNR) for signals located at subband edges. To further enhance computational efficiency, the OSPFB algorithm is optimized using a graphics processing unit (GPU)-based parallel architecture. Validation is performed using baseband data of formaldehyde (H2CO) absorption lines observed with the Nanshan 26 m Radio Telescope of Xinjiang Astronomical Observatory. The results indicate that the GPU-accelerated OSPFB method achieves superior spectral SNR compared with the conventional fast Fourier transform (FFT)-based approach. Moreover, the proposed method offers high flexibility and scalability in selecting the number of channels and FFT size, providing an efficient channelization solution for multi-resolution molecular spectral line observations.