Selecting Scene Matching Method Study

Based on the analysis of present commonly used scene matching area selection algorithms, this thesis presents a new a approach in the selection of scene matching area, GIS supported scene matching area selection based on image content. The new method shows great efficiency in solving the problems faced in present algorithms, which are high reliance on image quality, low resistance to noise, low veracity and mass process time.The feasibility and validity of the new approach is firstly observed in this thesis. Special efforts are done on the analysis of the key factors of the new approach, which are the relationship between different image content and its matching suitability and the reliability of the matching probability used as the matching suitability of an image. In this thesis, the measurement of the matching suitability of different image content is given. It has also been proved that the simulation of the matching probability of an image used in this thesis is acceptable.Tow practical realizations of the new approach is given: GIS supported scene matching area selection based on the distribution of image content and GIS supported scene matching area selection based on image edge density. Evaluation of the two algorithms with the standard based on the minimization of mistaken probability presented in this thesis show that both algorithms have high veracity and efficiency in the selection of scene matching area. It has been experimentally proved that the two algorithms presented in this thesis takes on high feasibility and reliability.