| The successful launch of GF-4 satellite that is the first high resolution optical imaging satellite in geosynchronous orbit has a great significance to the development of high orbit and resolution remote sensing satellite technology.As the corresponding relationship that acquire two or more images of the same object from different period of time,sensors,weather and illumination conditions,remote sensing image registration is a process of correcting the image through calculating the geometric transform parameters between each other.Remote sensing image registration technology is facing a variety of new problems and challenges in the process of gradually mature.How to realize the automatic registration between GF-4 Multi-source images is a key technology in the application and popularization of the new satellite data,and it is a hot issue in the current remote sensing image processing technology.Aiming at the problem of GF-4 high precision automatic homologous image registration,this paper proposed a multi-feature hybrid automatic registration algorithm with adaptability and robustness,which is based on the advantages of Harris and SIFT algorithm.Then,for the registration of GF-4 visible and mid-wave infrared heterogeneous images,only using the algorithm of feature points is difficult to achieve the same point on the extraction and matching in the process.So that this paper put forward a kind of visible and infrared image automatic registration technologies based on contour and point features.The experimental results show that using the algorithm of this technical process can well realize the GF-4 image spectrum between and among different camera image automatic registration,and homologous image registration precision can reach within 0.2 pixel,heterogeneous image registration accuracy at about 3 pixels.Therefore,this algorithm of technical processes for GF-4 image automatic registration are feasible and robust,which can provide some reference for later use of GF-4 satellite data. |
PDF Full Text Request