Abstract: The wetlands in Anhui province were taken as research object, a method of classification based on spectral and spatial information was applied to extract wetland information in the study area, based on the TM image removed noise and principal component analysis. The results indicated that the accumulated contribution rate of the first four principal components was 99.5%, which could represent most of the original data information. The average classification accuracy of wetland information extraction technology integrated by spatial information was 84.6%, while the average classification accuracy of maximum likelihood method was only 69.3%. The wetland classification method of combining spatial information could effectively weaken the noises and improve a certain extent classification. This study has certain reference value on the other related research of wetland remote sensing information extraction based on spatial information.