Research And Implementation Of Image Stitching Algorithm Under Limited Focal Length

The popularity of panoramic cameras is low, which due to the high price of the panoramic camera and the professional training of the operation processing. However, in our real life, people are eager to get super visual images, and look forward to the image which contains more information about the scene. For example, people eager to have an image which can display the scene that include all the scenes that the human eye can see when turn a circle. In this context, in order to get the "big picture" of the super human eye-view, and give people a kind of immersive feeling, we can use image stitching technology.The using of image stitching algorithm is the most economical, and efficient way. Through the way of image mosaic to get the panoramic image is that using the image mosaic algorithm to splice the overlap and adjacent images which are collected by the digital camera. Image mosaic technology has the advantages of high efficiency, convenience and economy. Since it has been put forward, it has been widely used. The first is proposed by the military, and in the military field has been tested. After that, with the good test results, people began to promote the medical imaging, agricultural research, video surveillance system. With the rapid development of computer technology and graphics and image processing technology, image mosaic technology has become a hot topic which has attracted much attention of researchers.In this paper, the image stitching algorithm is studied. Aiming at the problem of feature points detected in image registration, a registration algorithm based on image block is proposed. This algorithm firstly divides the image into the same number of sub blocks; Secondly, by comparing the correlation between the sub blocks to determine the overlapping area between the two images; Finally, we use the feature points which are extracted in the overlapping area to replace feature points extracted in the whole image and realize image mosaic. Experimental results show that the algorithm has a strong advantage. The main research contents are as follows:(1)Through study a large number of domestic and foreign information about image stitching algorithm, we record and familiar with the image stitching algorithm connotation and basic process. A variety of image mosaic algorithm for learning and research. The various steps of image mosaic technology are analyzed in depth.(2)Image registration and image fusion algorithms are studied. Feature based and region based image registration are mainly studied. Analysis of image fusion methods and image fusion evaluation. According to the rotation invariance and scale invariance of feature point, the SURF algorithm is used to detect image feature points.(3)Research on image mosaic algorithm based on SURF feature. In view of the problems in the process of feature point detection, an image mosaic algorithm based on image block with limited focal length is proposed in this paper.