Explore Object-oriented Classification for Land Use from High Resolution Satellite Imagery

Land use is an indispensable prerequisite for credible causes and consequences investigation of global environment changes. With the increasing availability of very high resolution remote sensing imagery, more accurate and effective analysis of land use is becoming possible. However, the traditional method of imagery interpretation is focused on pixel-based analysis, which has fundamental limitations in addressing particular land use characteristics due to finite spatial extent. Taking advantage of recent advances in imagery interpretation methods, a supervised procedure based on object-oriented image analysis, is proposed in this study to reduce manual labor and objectify the choice of significant object features and classification thresholds. A sequence of image segmentation, feature selection, object classification and error balancing was discussed in details. In order to verify the validity of object-oriented Classification for high resolution satellite imagery, a scene of 2.4-meter multispectral image of Quickbird is respectively classified by the pixel-based analysis from Erdas and the object-oriented method from eCognition. It can be concluded by comparison that the object-oriented classification is better fit to exact the land use information from high resolution remote sensing imagery than the pixel-based method.