Research And Realization Of Video Image Mosaic Optimization Algorithm

With the deepening of the research on image matching and the application of the video image in life, image information gained by single video acquisition device has the limitation on the vision, in order to obtain wide field of vision and the big scene video images, multiple video image sequence with overlap for stitching which can form a new video image with more perspectives become the method we can consider which can be called video image mosaicing technology. It is widely used in the imaging map, virtual reality and telemetry. At present, there are more in-depth theory studies of the static image mosaicing technology which dynamic video image mosaicing can use.According to the basic process of image mosaic, this paper first introduces the research background, development present situation of image mosaic, then focuses on registration and image fusion. Secondly, the image registration method based on the characteristics is selected as the research content of this article, then the SIFT features are chosen as essential extract features of image registration. In the aspect of image fusion, it achieves the image fusion by taking the synthetic method of gradually progressive.Due to the large number of unstable edge points produced by SIFT operator, matching computation of SIFT algorithm takes a long time, therefore an improved SIFT algorithm combining with canny operator and K-L transformation is proposed in this artcle. Experimental results of two images’s registration show that the improved algorithm can effectively remove the old SIFT feature points which are useless, decrease matching time, increase the accuracy.Based on the improved SIFT matching algorithm and the fusion method, an improved SIFT feature mosaicing algorithm is proposed. This paper first completes the static image mosaicing with the improved algorithm, then finishes frames image mosaicing between two dynamic video images, on the basis of frames image mosaicing, successfully completes the video mosaicing. The experimental results show that the improved video image mosaicing algorithm has a good performance in improving the static images, video image mosaicing quality and mosaicing speed, which can also eliminate aperture and lighting problem in the process of image mosaic.