| Digital Scene Matching Area Correlation(DSMAC)is a navigation method based on terrain image information.It determines the current position of the aircraft by comparing the input real-time map with the reference map.DSMAC has high autonomy and good accuracy.It is not subject to geomagnetic interference.However,the large amount of image matching calculation results in low real-time,low reliability and difficulty in embedding.Therefore,the improvement of image matching algorithm becomes the primary task in this direction.Firstly,the scene matching system and terrain height matching system are introduced.This paper analyzes and compares the advantages and disadvantages of the two methods and the scene matching system is selected as the research object.Then this paper describes the specific implementation methods of scene matching system and introduces the development status of several methods.After part of the scene matching system and the working principle of each part are introduced in detail in this paper,This paper compare the two main algorithms of scene matching system.One is based on gray level information,the other one is based on the characteristics of information.Then,the characteristics of SIFT(Scale-invariant Feature Transform)algorithm are introduced,which lays a foundation for the following.Then this paper introduces the principle and process of SIFT algorithm.This paper analyzes the accuracy of SIFT algorithm.In this paper,the algorithm accuracy is reduced,the feature point selection process is improved,and then the algorithm speed is improved by using kd tree search strategy.The improved SIFT algorithm is simulated and analyzed,and the effectiveness of the improved method is proved.The improved algorithm was taken as the main part,and the flight path constructed by MATLAB was used to simulate the scene matching system.The system error is also calculated,and the results show that the improved algorithm can be applied to the scene matching system. |
PDF Full Text Request