Atmospheric correction is an important part of remote sensing quantification. The commonly used atmospheric correction method is the dark target method, which is suitable for application in dense vegetation areas but is less suitable in areas with low coverage of vegetation. In this study, an atmospheric correction method is proposed based on spectral matching. Herein, an invariable target in urban areas is used as the entry point. Further, we develop an atmospheric correction method for the Gao Fen-1 (GF-1) satellite panchromatic and multispectral sensor (PMS) camera. This method uses the 6S radiation transmission model to construct an atmospheric correction parameter lookup table for obtaining the inversion spectra of cement pavements from different images under different atmospheric conditions, and the average measured spectrum of the cement pavement is considered to be the reference spectrum. By the angle matching between the test spectrum and the reference spectrum, the closest spectral curve is estimated, the atmospheric correction parameters are determined, and the image is atmospherically corrected. The experimental results denote that this method works well, and the surface reflectivity obtained based on the inversion is consistent with the typical ground spectral data, which restores the surface to its actual situation and provides a novel atmospheric correction method that can be applied in areas having sparse vegetation.