| Change detection is to observe one or more kinds of data covering the same area in different time ranges,so as to identify and obtain the changes of objects in the area with a specified time range.With the development of society,the technology has played an important role in urban and rural planning,disaster assessment and other fields.According to the types of data sources,change detection can be mainly divided into High-resolution optical remote sensing image change detection and Synthetic Aperture Radar change detection.Among them,High-resolution optical remote sensing images are rich in spectral,texture,structure and neighborhood information,which is conducive to fine characterization of ground object change information.On the other hand,with the improvement of spatial resolution of sensors,higher requirements are also put forward for the phenomenon of "same object with different spectrum" and "same spectrum foreign body".SAR can get rid of the influence of weather and other conditions.The polarization features contained in SAR image are conducive to the multi-angle description of changing information.However,the software designed for the data processing of fully polarized SAR is relatively scarce,leading to the few applications of its polarization characteristics in the field of change detection,and the detection performance is still unclear.Therefore,although these two types of data have their own advantages in the application of change detection,they also face different challenges.The key scientific issues in change detection based on high-resolution optical remote sensing image and fully polarized SAR image are studied respectively.The specific contents are as follows:(1)Aiming at the change detection applications of High-resolution optical remote sensing images,a change detection method combining Adaptive Morphological Attribute Profile and decision making is proposed.Firstly,Morphological Attribute Profiles by change vector analysis were analyzed to extract the initial difference feature set.On this basis,a Balanced optimal objective function was designed,and the optimal difference feature set was extracted.Finally,based on the proposed Evidence index and Indicators of evidence confidence,a multifeature decision fusion framework was constructed to obtain change detection results.Experimental results show that the Overall Accuracy and F1 of the proposed method can reach96.41% and 88.67%,respectively,which are significantly better than the comparison method in visual analysis and quantitative evaluation.(2)Polarized SAR can simultaneously transmit and receive horizontally and vertically polarized electromagnetic waves.Compared with traditional SAR images,it can obtain more target information and has unique advantages in such application fields as change detection.As the first multi-polarization synthetic aperture radar satellite in China,the performance analysis of GF-3 SAR features in the application of change detection is rare.To solve this problem,this paper relies on GF-3 fully polarized SAR data to construct the backscattering coefficient and polarization decomposition feature set by layers,and compares and analyzes the recognition performance of different feature sets in change detection.Multiple sets of experimental results show that the feature set change detection based on the combination of GF-3 scattering and polarization information has the highest OA and F1,which are more than 89.30% and 72.53%,superior to the single feature detection results,which is conducive to the promotion and application of GF-3 satellite in the field of change detection. |
PDF Full Text Request