Abstract: The application of remote sensing technology to monitor and investigate the bamboo resource can provide basic data support for the scientific management and efficient utilization of the bamboo resource quickly and effectively. The author used the TM images of Shunchang County Fujian Province, using the maximum likelihood method, the sub-pixel classification and the classification based on spectral characteristics 3 classification methods to extract the bamboo information. The result showed that: the sub-pixel classification was the most precise, and the overall accuracy up to 77.33% , followed by the classification based on spectral characteristics, maximum likelihood method, the accuracy respectively was 76.50% and 76.17% ; the Kappa coefficient respectively was 72.8% , 71.8% and 71.4% ; and compared the results of the 3 classification methods to the bamboo forest area of the second survey in 2007, the accuracy of the classification based on spectral characteristics was the highest 95.68%, followed by maximum likelihood method 93.41%, the sub-pixel classification 92.97%.