Remote Sensing Lithology Extraction Based On Fractal And Spatial Analysis Theory

With the development in recent years,Remote Sensing,as an applied and efficient means,has been integrated into many subject areas.Lithology information extraction is an important application.However,it was hard to find an example in heavy – covered areas,especially in the loess area.The reasons were attributed to two points:(1),Background information is too strong while lithology information is weak.(2),The ac-curacy of the extraction method is low and easy to be mixed false anomalies.Therefore,this article carried out an experiment on a loess-covered area in Baoji,Shaanxi.We take the Mesoproterozoic Xiong'er group and the Mesozoic Red-sand stone outcrops as main objects,exploring the lithology extraction methods of remote sensing technology in such a geological background.The main experimental process of this article is as follows: Pre-processing ? Terrain Analysis ? Image Enhancement ? Fractal Extraction ? Spatial Analysis.The working concept of this article is “Grasping the whole picture rapidly,Narrowing the target-area gradually”.First,through geological data and ground observation,grasp the overview of the study area and the spectral and spatial characteristics of the object to be extracted quickly.And then,eliminate the regions that are rarely found true outcrops to narrow the extracting target area.The final accuracy of the Xiong'er Group and Red-sand stone are 100% and ~75%.Through experiments,three conclusions has been obtained as follow :(1)Geological data and previous field surveys are important steps to ensure the successful conduct of the later experiments.And it's necessary to do the Terrain Analysis before extracting,for it can keep the effect of background to a minimum and decrease the probability of false anomalies.(2)Using the fractal model and spatial analysis method in the loess-covered area can effectively and accurately extract the target lithology anomaly.Because it takes into account not only the DN value,but also the spatial characteristics of the abnormal pixel.