The Study On Optimal Segmentaion Scale Of High-resolution Images

In recent years a growing number of high-resolution remote sensing images are applied. Traditional methods cannot meet the demands of high spatial resolution remote sensing image classification. The basic processing units of object-oriented method are the image units with fine integrity and oneness through multi-scale segmentation. The extraction of prolific geographic spatial information in high-resolution remote sensing images needs to be proceeded at different scales. The choice of proper segmentation scale of different land cover types is a key problem in object-oriented classification.The following research produces are obtained in this paper:1. The smallest computable unit is proposed.1) The choice of smallest spatial unit at different spatial scales(the best resolution of remote sensing images)2) The choice of smallest spatial unit at different temporal scales.3) The choice of smallest computable unit (smallest patch) in object-oriented method.2. Pixel ratio method is proposed to solve the choice of best segmentation scale. The main purpose is to balance the "drowned" and "broken" phenomenon.3. Object-oriented method is used to proceed feature extraction,...