The Super-resolution Reconstruction Method Of Slope Based On Sparse Mixed Estimator

In the field of topography,slope is a unit surface steep degree,usually the vertical height of the slope h and horizontal distance l ratio is called gradient.Slope is a very important concept,and in all aspects of the slope is playing a very important role,so it is particularly important to obtain slope.The slope has an important influence on surface hydrology,soil erosion and land use planning,and the gradient on the regional scale is usually based on the digital elevation model.At the regional scale,it is difficult to obtain the high resolution data of slope due to the way of DEM acquisition,DEM is usually obtained by super-resolution reconstruction(Down scaling).A good algorithm plays a key role in achieving an accurate and consistent DEM data.In this study,the improved POCS algorithm is used to reconstruct the slope data based on the super resolution reconstruction of the slope data and the sparse mixture estimation method.The content can be divided into the following two parts:(1)Improved POCS method for super resolution reconstruction of slope data.This algorithm firstly DEM reads data to judge,to determine the approximate scope of this set of data values according to the variance of each data,the data and the average range of differences are eliminated,to obtain better reconstruction result;to construct the reference data by using the method of bilinear interpolation,make the result more smooth finally,reference data correction;first consider the terraced slope of great change,the data of edge detection,and the detection results obtained after the correction of the reference data.(2)Super resolution reconstruction of slope data based on sparse mixture estimation.The first step is to transform the slope data into multi direction wavelet transform,and the wavelet transform is carried out in four cases of approximation,horizontal,vertical and 45 degrees.After the establishment of the slope data block dictionary and orthogonal matching pursuit,sparse block by redundant dictionary construction slope data,the slope data characteristics is summed up,the data as the geometric block orthogonal matching pursuit,through the method of geometric block can make the data integrity can be guaranteed.The combination of wavelet analysis and block dictionary mixed interpolation of the slope data,the algorithm based on L1 and L2 two kinds of algorithms are calculated at the same time,give full consideration to the regular change of slope data,combined with multi direction wavelet transform of mixed interpolation slope data in different directions,to nsure the integrity of the data,and increase the accuracy of the reconstruction effect.This study is based on the study of the horizontal terraced fields in the Loess plateau.It is based on UAV photogrammetry technology to generate different resolution DEM slope data and extract the slope.The method and process of super resolution reconstruction of DEM slope data based on sparse mixture estimation are designed and presented.And compared with nearest method,bilinear method and Cubic method.The results show that the proposed method is superior to other methods in terms of spatial distribution and error.