Abstract: Based on the second survey of wetland resources in Zhejiang Province and the domestic high-resolution images (GF-1, GF-2, ZY-3), the dynamic monitoring of wetland resources in 2016~2018 was carried out by using the object-oriented classification method in eCognition. The research showed that the classification accuracy based on eCognition, GF-1, GF-2, and ZY-3 images is significantly superior to that based on ArcGIS and SPOT-5 images. Using eCognition, GF-1, GF-2 and ZY-3 images combined with previous vector data had obvious advantages in wetland resource classification such as higher accuracy, more abundant resources, more intense timeliness, and lower economic costs. According to The Ramsar Convention, the wetland areas of Zhejiang province have remained stable, showing a trend of declining from 2012-2018. Protection and restoration of wetland resources need to be strengthened in Zhejiang Province.