| Water system is one of the very important remote sensing image interpretation signs, interms of geological interpretation, we often focus on tributary, creeks and other waterlinesnetwork of the final stage water system in particularly. Its development is controlled by lithologyand geological structure, so water system type, degree of development, the combination of statecan directly reflect the type of rock, tectonic features.In this paper, the study area selects Greater Xing’an Mountains Shilichangling to Tahe area,the AVNIR-2sensor of ALOS satellite image data, the imaging time is May2009and September2009of10M resolution images. Study area is located in the north section of the main ridgevolcanic belt of Greater Xing’an Mountains, Mesozoic volcanic rocks are its main geological units,Greater Xing’an Mountains covers with much vegetation.Water system is a special texture of remote sensing image; remote sensing image has twobasic characteristics which are spectrum and texture, commonly used method of extraction withthe spectral classification means can not meet the precision and does not get dry gullies and otherwater system, so this paper extracts water system based on texture features of the high spatialresolution remote sensing image, analysing the water system texture features of remote sensingimage, through research of texture analysis method, we decide to choose the combination ofGLCM and wavelet transform method to extract water system texture from the image. The resultswe achieve are as follows:1. The paper uses variance to reflect remote sensing image texture information, but duringthe extraction process, the operation window size, moving step, statistical direction will affectaccuracy of the extraction results, in which the size of the window is decided by the resolution ofthe image, the degree of detail of the texture, at the same tine it’s crucial to the aqueous clarity ofthe results of the extraction. Regularity characterizes the degree of smoothness of functions,compact support characterizes local signal characteristics, symmetry is relation to whether thefilter characteristic of wavelet has linear phase or not, disappear matrix reflects the smoothness ofthe wavelet function, in which Mexican hat wavelet can effectively suppress noise, and it can alsoeffectively detect gray mutation region of the image, suitable for aqueous extraction. During thehat transform, the choice of scale is essential to extraction accuracy.2. In this paper, based on the size of the window can influence GLCM, the choice of scalecan influence the wavelet transform, for the first time, this article uses a comprehensive application of wavelet transform and GLCM, based on Matlab algorithm design, to reduce theimpact of the gray window size, using the advantage of wavelet that wavelet can separate thesingle noise, weakening the impact of wavelet scale, effective avoid, remove or suppress theinterference of the above factors, using a combination of both methods to extract the water systemtexture from remote sensing image which covers Greater Xing’an Mountains, the extraction resultsare satisfactory, water system level reaches five, match with results of visual interpretationcorrectly, extraction results provide an important basis for geological structure.3. lithologic facies division of the study area. Through analysis about developmentcharacteristics of water system forms, this article interprets classification and structure of rocktypes in the study area accurately, the study area has rich water system types, with different causes,radial and ring water system is most common, as study area is mostly consisted of mesozoicvolcanic, radial water system is mostly crater, volcanic cones. There are some Feng-shaped watersystem, barbed water system, like parallel-like water system, which are all relation to lithologicdifferences and lineament. By water system type, accurately identify volcanic rock so that highvegetation coverage areas can have effective geological interpretation. |
PDF Full Text Request