The SIFT Research And Implementation Based On The Image Registration

Image Registration is a fundamental problem in remote sensing, medical, computer vision, and many other fields. In the field of remote sensing, image Registration is essential as a key step in realizing image fusion, change detection, image correction, image mosaic, and other applications. Because of the remote sensing information and a wide range of applications, automatic registration has been a goal pursued by the people.Image feature extraction and image matching is a kind of automatic registration important ways. SIFT (Scale Invariant Feature Transform) initially is as a key feature to be proposed, which is not variable in the scale and rotation in the image changes, but also has a strong adaptability in illumination and image deformation. In the same time, this feature also has high identify ability, which is easy to follow-up the match. It is through the use of these characteristics, traditional image registration has got a large degree of improvement in many of the common problems mentioned above. Therefore SIFT has a wider range of applications in the field of remote sensing, realizing image fusion, change detection, image correction, image mosaic, and other fields. SIFT Image Registration is a hot field in the forefront of research.This paper simplely introduces the image registration application and direction, then focuses on a value based way in image registration (SIFT algorithm). It includes the history of SIFT algorithm, the algorithm characteristics, main ideas and implementation steps, and then finished the realiziation of SIFT point feature extraction, design mapping function to find an adaptive registration in mapping function design, optimization algorithm, reduce the algorithm complexity and engineer complexity in SIFT algorithm, then constructed a virtual model, and eventually realize the SIFT Algorithm.