Land cover classification of CBERS-02 images based on object-oriented strategy - A case study in Yixing, Jiangsu Province

The recognition and translation of land cover via remote sensing images is the subject of this work. It is known that image information with different scales display distinct spatial structure, so image analysis using a single scale could not meet the heterogeneity and dynamic pattern and process for most remote sensed images. The authors present an object-oriented image analysis method that could create meaningful objects and build a hierarchical level close to surface character using multi-scale segmentation. Different geographical processes could then be represented in corresponding image-object levels. The object-oriented image analysis has realized multi-scale analysis of spatial patterns and process. The extraction of land cover classification based on an object-oriented strategy sets most priority on the multi-scale segmentation of images, the measurement of spectral, geometric and topological characteristics, the interaction between human and computer, and the construction of knowledge base. The authors use China¿Brazil Earth Resources Satellite (CBERS) CCD images taken in August of 2006. These have been geometrically corrected via methods of quadratic polynomial and bilinear interpolation to control the RMS within one pixel, choosing the typical urban building area and abundant land cover of Yixing City of Jiangsu Province (China) as the study area. The work carries out the classification experiment on the study areas using Ecognition software.