Abstract: Setellite-1 and Resources-1 remote sensing images are used as research data. This paper uses method of object-oriented multi-scale segmentation and spectral feature threshold to extract cut-over areas.Study area included harvested area and non-harvested area.Forest cover change has been detected by post-classification.The results showed that the imagery has good internal geometric consistency,the orthorectification accuracy between two-period imageries can be less than one pixel.The area accuracy of remote sensing change detection and the coincidence rate compared with the actual changes were 93.1% and 95.0%.This method can better detect changes of forest cover.It is suitable for annual update of forest resource data for the county.

Key words: remote sensing, forest cover, object-oriented, change detection