Land Use Information Extraction by Object-Oriented Technology Based on Remote Sensing Image

Abstract: Traditional classification method based on spectrum of pixels can not make full use of spatial information of high resolution remote sensing image and the classification accuracy is not good. A new information extraction method-object oriented information extraction-is introduced in extracting information from high resolution remote sensing image. This paper combined the research of this method and its application on information extraction of land-use based on Quick Bird image. The main contains and results as following:(1)In the paper, the object-oriented scaling was researched based on the pixel scaling. The result of scaling was also evaluated by using Mean, StdDev, and fraction dimension for indices. The result indicates that the appraisal indicator values of classes vary with different scales, and the variety tendency of different classes is not same.(2)The optimal scale is the key of multi-segmentation extraction accuracy increasing. a view of optimal split-scale model was proposed based on previous study and the test of image segmentation was taken. The results showed that this model can get optimal split scale expediently and fleetly, avoid the visual subjectivity, improve the efficiency of image segmentation.(3)The experimental area was segmented by choosing the optimal optimum scale. Class hierarchy was constructed and fuzzy rules and feature space were defined to extract information by fuzzy classification.(4)The paper compares and appraises the classification result between the method of object-oriented and pixel-oriented (such as the maximum likelihood classification) .The result indicates that it has higher precision when used object-oriented method to classify the high-resolution image, the object-oriented method is so effective to reduce the "Pepper and Salt Phenomenon", the classification result of object-oriented analysis is more easy to understand and explain…
Key words: remote sensing image; object-oriented; multi-scale; segmentation; fuzzy classification