Abstract: To address the significant impact of Raman spectral wavenumber range, resolution, and peak intensity differences on the retrieval correct identification rate of both the traditional hit quality index （HQI） and the improved similarity hit quality index （SHQI） algorithms, the weighted segmented correlation coefficient method （WSHQI） based on Voigt function fitting was proposed for cross-instrument Raman spectral retrieval. First, raw spectra were preprocessed by wavenumber range calibration and baseline correction. Then, the Voigt function was used for parametric feature extraction and spectral reconstruction of the target spectrum to eliminate useless information and retain peak parameters. Subsequently, the full spectrum was segmented using a Hann window, and the spectral similarity within each segment was assessed locally, overcoming the limitation of HQI, which only evaluated the overall spectrum. During the segmentation process, peak position correction was introduced to eliminate peak misalignment and improve segment matching accuracy. Furthermore, the weights for each segment's HQI were optimized. By combining the precise resolution of peak distribution from the Voigt function fiting, higher weights were assigned to regions with dense peaks and high response, thereby reducing the influence of low-response regions on the matching process. It was shown that when the WSHQI algorithm was used to retrieve the Raman spectra of 103 hazardous chemicals measured by 3 instruments, the WSHQI algorithm achieved correct identification rates above 90% for the Top-1 matched spectrum, significantly surpassing the correct identification rates of both HQI and SHQI. This method provided a technical basis for customs authorities to use portable Raman spectrometers for the rapid and accurate screening of hazardous chemicals.