Study Of Digital Image Mosaicing Technique

As an important branch of image processing field, digital image mosaicing technology is used to mosaic a series of images with same areas into a high resolution image with no obvious seams. At present, image mosaicing technology is widely used in the field of geo-information system, remote sensing image processing, virtual reality technology, computer vision, medical image analysis, etc. Image mosaicing technology has already become a popular research subject in the international academic world.image mosaicing systems mainly consists of five parts, image preprocessing, image registration, searching transformation model, image compensation, and image fusion. Among them, the image registration and image fusion are two key technologies. As the basis of image fusion, the innovation of the image registration determines the development of image mosaicing technology greatly.This thesis discussed several common used image registration algorithms at present and focused on searching a precise and accurate registration method. Firstly, several common image registration algorithms were studied systematically, and then their registration performance was verified. Secondly, based on the classic SIFT method, an improved one was proposed based on obvious feature extraction method (OFE). Finally, this paper analyzed the SURF feature registration algorithm and improved it with the OFE. The simulation results show that under the condition of accuracy and applicability, the registration speed of SIFT-Based method with OFE is two times faster than the original one. More over, the study in this paper proved that the SURF-based registration algorithm is robust and anti-jamming and well noise immunity, besides that, its registration speed is two times faster than SIFT, so it meets the speed and precise simultaneously.