Panoramic Image Mosaic Based On SIFT Algorithm And Graph Structure

Image mosaic technology is an important branch in the field of computer vision. It is used to finish spatial matching of images sequence with the part of overlapping with each other, which contains image sequence information, wide viewing angles and high resolution. Image mosaic technology has solved the problem of low equipment perspective and low resolution, and promoted the development of image processing technology. Now, image mosaic technology has been widely used in many fields, such as medical image analysis, computer recognition, remote sensing, and a virtual reality, and so on.At first, the present research background and significance, domestic and abroad research recently and basic process of image mosaic are introduced in this paper. Later, the level and several important methods of image registration and image fusion are summarized in this paper. It also analyzes the advantages and disadvantages of them. Based on the existing algorithms, this paper proposes a panoramic image mosaic algorithm based on SIFT and graph structure. Firstly, based on the SIFT algorithm, the feature information of each image is extracted and feature point is matched. Secondly, the global matching among disorderly image sequence is completed. Finally, image fusion for realize automatic panoramic image mosaic has processed.The main contributions of this paper can be summarized as follows:In the part of image feature matching, it uses BBF search algorithm offset Kd-Tree algorithm when it’s spatial dimension more than ten and has poor execution efficiency; In the part of feature point purification, it uses improved RANSAC algorithm to eliminate false matching points and reduce redundancies; In the part of image fusion, it uses a fade and out of in gradually function to mandatory correct gray on both sides of image seam-line based on hard correction method, to achieve the seam-line removal; In the part of more disordered image mosaic, it uses search method of undirected graph to automatic distinguish the disordered image sequence, to implement the intelligent of the image mosaic. Experimental results confirm that the feasibility of this method, it improves the precision and speed of the image mosaic, mean while, the visual of image mosaic are improved.