Polarimetry SAR images give more information about ground targets than single channel SAR data.Through polarization decomposition, multiple parameters would be received. These parameters could reflect the backscattering character of ground target from various aspects. Different polarimetric parameters could reflect different backscattering characters of targets, which would be important basis in target recognition, However, some features in some areas due to the scattering of similar type, often they will come up with different surface features shown on some parameters of the distribution of the same or similar, in order to identify surface features have brought much inconvenience. At the same time, due to the different surface features in the shape, surface roughness, dielectric constant and other differences, there is no one or two parameters can effectively differentiate among all the features in class. Furthermore, there are man-made target and much nature target in our study area. Man-made structures target is usually the case for the second scattering and volume scattering dominated, while the HH-HV and HV-VV channel correlation coefficient is relatively sensitive to secondary scattering and volume scattering. Taking all these factors, this paper applied in an algorithm jointing hierarchical classification and special multidimensional analysis based on parametric statistics for land cover type recognition from quad polarimetric SAR image. The first step is to calculate Correlation coefficient to extract man-made target, the second being dominant scattering statistics to gain major classification, the third being multidimensional parameter space analysis to further distinct the targets without differentiated from the second step. Through optical imaging, land-use maps and field reconnaissance, the proposed algorithm can be verified whether or not effectively extract the scattering characteristics of surface features, and to further distinguish surface features category.The study area locates in the north of Tianjing City, china. The experimental area is the waterfront area, covering the features in the image include residential areas, farmland, woodland, wetlands, water and other surface features target. Land cover types are more complex, suitable for use in image classification research.The results can be seen by multi parameter analysis technique can effectively extract the dominant target scattering characteristics of surface features, and scattering parameters through a leading distribution in the feature space to determine the form of object classes to meet the purpose of class distinction. Through land use maps and field survey and investigation results show the algorithm was able to in other parameters of the surface features in the confusing distinction between objective, accurate extraction results. so,this paper introduces the multi-dimensional parameter analysis thinking, breaking the traditional single parameter analysis and segmentation based on two-parameter plane target feature extraction algorithms, reduces the amount of data to run and optimize the analysis process and improve operational efficiency and results of precision.