Research On Multi-image Mosaic Algorithm

Panoramic image mosaic has been a research hot spot of the computer vision technology and image processing. Through approachs of Image Based Rendering, it represents a high-resolution and smooth image with large view, which consists of a number of discrete images information, and which well resolves the conflicting issues field of image view and resolution. In virtual reality technology, tourism, medical image processing, remote sensing measurement, aviation and military fields, the technology has a wide range of applications. Access to high-quality panorama of traditional methods relies on special equipment, for this image stitching algorithm, it is difficult to meet the needs of the wider. In this paper, considering the characteristics of the camera to obtain images, we research image registration and image automatic sorting problem.Based on the current work, we have analyzed a wide range of image registration algorithm, and proposed a combination of Euclidean distance ratio and median filter to improve the RANSAC (Random Sample Consensus) registration algorithm, which resolves the problem of the accuracy and robustness of image registration; In addition, we summarize semi-automatic and fully automatic image stitching algorithm in the field of image mosaic, based on the improved registration algorithm, and realize a fully automatic panoramic image stitching algorithm.Experimental results show that the algorithm can splice the region of two overlapping images effectively, improve the accuracy of image matching and the stability of image homography, reduce the implementation time, and have a good robustness; And the algorithm can achieve a number of images automatic sorting and matching, which has a good stability, without human intervention. Finally, we summarize the content, and propose the next phase of research and the field of image mosaic of the prospect of future development.