Image Mosaic Technology Research Based On Streaming Video

Image mosaic is an important research area of computer vision, image processing and computer graphics all the time. The purpose of image mosaic is to synthesize small-sized overlapping region images of the same scene into a large-size image, even a panoramic image of some large scene. The technology of image mosaic has been widely used in the fields of virtual reality, medical image processing, remote sensing and military affairs. In the field of photography, in order to reduce costs, the use of ordinary cameras capture images to get the wide viewing angle of the image becomes a popular research direction, as a number of important branches of image processing, image mosaic technology provide a good solution.Generally, this paper is about image mosaic technology based on video, and the author introduces two image mosaic methods, which are image mosaic algorithm based on gray-scale and image mosaic algorithm based on the displacement statistics, and its application in the container residue testing system is of great practical significance.This paper summarizes the efforts and studies of image mosaic that the researchers have achieved in this field. Then following a basic flow chart of image mosaic, which contains image preprocessing, image registration and image integration, the paper introduces the technology of image mosaic simply. In this study, the video streaming taken by video camera has been distorted, so distorted images should be preprocessed before image mosaic. According to the reason why images are distorted, this paper uses a classical mathematical model to correct distorted images. Secondly, based on the study, this paper introduces two image mosaic methods, which are image mosaic algorithm based on gray-scale and image mosaic algorithm based on the displacement statistics. On the base of image preprocessing, the image mosaic algorithm based on gray-scale uses classical image registration methods, which is model matching method, to match overlapped fields of two frames, then mosaic these images to be a complete image. According to the characteristic of imaging, the second algorithm, that is image mosaic algorithm based on the displacement statistics. intercepts undistorted image part of each frame to be a new image, then executes image mosaic based on these new images. However, the second algorithm must be sure there are overlapped areas between each intercepted images, or the algorithm will be failed. Then the paper discusses the robustness of these two algorithms for application. Experiments show these two to improve the algorithm has good robustness. Finally, the paper makes a summary of the problems still existing in image mosaic and gives prospect of image mosaic outlook.