In this paper, we propose a covariance matrix matching technique using a uniform circular array (UCA) to localize a near-field source in indoor multi-path environment. Firstly, we exploit off-line measured covariance matrices for suitable spaced reference points and storing these matrices as a fingerprint; Secondly, a matrix matching algorithm is designed to obtain the position of source. Compared with received signal strength (RSS) fingerprint, covariance matrix can offer more channel information of indoor environment, hence, our proposed algorithm outperforms RSS based algorithm in accuracy of localization. Finally, some simulation results show the efficacy of our proposed technique.