Reservoir Detection Based On High Scattering Remote Sensing Image Method Study

China's high score special start to open the domestic high-scoring satellite in the municipal planning,traffic construction,resource monitoring,disaster prevention and so on the application,high score image of the space has a strong show,texture information is rich,but its data The traditional low-resolution remote sensing image processing method is no longer applicable.For example,the pixel-based change detection does not take into account the spatial relationship between the pixel sets,which leads to the unsatisfactory detection effect and so on.Therefore,it is very important to study the computational speed,and to fully exploit the rich texture information in the image,and the excellent image processing technology is very important.The wavelet fusion method can well preserve the rich spectral information,and the gray scale morphology can well describe the fusion region edge information and denoising.Object-oriented image processing has also shown a significant advantage over pixel-level processing in recent years,and the application of machine learning methods in image classification has become increasingly widespread.In this paper,based on the above experimental analysis,the main research content has the following points:(1)The fusion of mathematical morphology and wavelet transform Firstly,the methods of image fusion are introduced systematically,and the advantages and disadvantages and applicability are discussed.It is shown that the wavelet transform fusion method of frequency domain is better than the transformation of space domain,and the spectral retention is better after fusion But also based on the shortcomings of the fusion of wavelet transform,that is,the fusion rules of low-frequency information do not take into account the edge of the situation,the introduction of mathematical morphology,introduced the gray morphological structure elements and basic operations and morphological gradient definition;The principle and procedure of the wavelet fusion method of morphological semi-gradient operator are compared with those of GF-2 image.It is concluded that the method has good fusion effect in terms of spatial detail and spectral preserving.(2)Image classification method based on support vector data description Firstly,the related concepts and theoretical derivation of machine learning and support vector machine are expounded.It is shown how the support vector machine method solves the optimal classification plane in the case of linear separability and linear indivisibility.Then,according to the support vector machine method,it can not be applied to large data The SVDD and multi-class SVDD algorithm are introduced.Finally,according to the reservoir change detection which needs to be done in this paper,a single classification SVDD is used to select the training samples to carry out the water and non-water bodies.Classification.(3)Object-oriented high score image reservoir change detection method Firstly,the methods of existing reservoir change detection are introduced,including the method of water body extraction and threshold-based change detection.Then,the object-oriented image processing technology is introduced,and the image object and property,image segmentation theory and method are summarized.Finally,The reservoirs of Xianning City in Hubei Province were tested based on the high score lm and 2m multi-spectrum,and the comparison of the water area was carried out,and the accuracy of the method was verified by comparing with the general data.