Using remote sensing technology in Mangrove Community Classification is very significant for surveying

taking advantage of and protecting Mangrove resource. In this study

based on the spectrum characteristics of mangroves

vegetation index and texture information calculated from WorldView-2 satellite imagery

we used object-oriented classification method

SVM (Support Vector Machine)

in conjunction with field surveys to map mangrove forest at communities-level in Daweiwan District

Qi'ao Island

Zhuhai. The single-scale and multi-scale classification were also compared. The results indicated that WorldView-2 data

a very high-resolution satellite remote sensing imagery with 8 bands are very suitable for mangrove forest classification using objectoriented and SVM method. The overall accuracy and Kappa indices for mangrove forest classification at the species level in the study area were 84.2% and 0.794 for multi-scaled analysis and 69.8% and 0.616 for the single-scaled.