Abstract:Texture is an important research content in remote sensing image processing and computer vision field,and is has a very wide range of applications in the field of vegetation remote sensing. The article which is on the basis of the object-oriented and the multiresolution segmentation, discusses,respectively,nearest neighbor classifications which are based on spectral information and texture features extracted vegetation of a new development zone in Nanjing.The research results show that,compared with the classification based on spectral information and the features pixed-based classification,the overall accuracy of object-oriented nearest neighbor classification supplemented by texture feature is increased by 6.36% and 1.16%.