[Objective] To reduce the side effects of noises for remote sensing data and improve the accuracy of spatial information of traditional stand type classification. [Methods] Minimum noise fraction(MNF) was used to deal with the image. Then a kind of classification method with the combination of spatial and spectral information was applied to complete the stand type classification in study area based on remote sensing information. [Results] This identification method can effectively restrain the phenomenon of "hard spots". The average accuracy of all stand types information was 83.6% and 11.6% higher than the maximum likelihood method. [Conclusion] The stand type classification method of combining spatial information can effectively weaken the noises to a certain extent and improve classification accuracy. This method could provide references for other related researchs on remote sensing information extraction of stand types based on spatial information.