Water is an indispensable type of feature in remote sensing classification.Due to the influence of cloud,cloud shadow,mountain shadow,water pollution,water body width and water depth,conventional extraction methods can not satisfy the water extraction of complex areas Accuracy.In view of the above factors,the water body can not be accurately extracted and the threshold selection is not universal,the Landsat8-OLI image is used as the data source.On the basis of the existing research theory,a new water extraction model is proposed to realize the water extraction of automation.Firstly,the LBV transformation model of each scene image is deduced by using the bands 4,5,6 and 7 in the Landsat8-OLI image study area for the thick cloud region on the remote sensing image.V can be removed by the thick cloud,through the three components of the inverted'V'type can be extracted from the water.In order to remove the influence of paddy fields and shadows,the combination of LBV transform and TC component is combined to form a new water extraction model,which can get rid of the subjectivity of the threshold.In order to solve the problem of fracture in the extraction of small water bodies,the morphological processing is used to make the fault situation greatly improved.Finally,the algorithm of automatic extraction of water body is realized by IDL,and the result of the algorithm is verified by the accuracy of the algorithm.The results show that the model can successfully extract the water body,And the Kappa coefficient reaches 0.8795,which is higher than that of multi-band spectrum,water body index,supervised and unsupervised classification,and so on. |