Research On Image Mosaic Technology Based On 360-degree Panorama

The 360-degree panorama is a technology that provides users with high resolution images that exceed the human eye's perspective.This technique is advantaged with the simple data acquisition,rapid modeling and providing favorable sense of reality.For these reason,this technique is widely applied in the fields inclusive of business affairs,military,tourism,real estate and medical treatment.The core technology of 360-degree panorama is to stitch the images.Image mosaic is specifically researched in this paper.The main work and innovation points are as follows:1.The related works of image mosaic is mastered,and on this basis,three relevant technologies of image mosaic are anatomized,image preprocessing,image registration and image fusion.The theories of the classical algorithm are analyzed and the advantages and disadvantages of each algorithm are concluded.2.Feature point detection.The merits of traditional feature point detection method,ineffciency and error detection,are optimized.Firstly,the non-feature points are quickly filtered out by using the gradient relation between the pixel and its surrounding pixels.Secondly,the surplus pixels are detected by applying the Harris feature point response function,which can acquire the pending feature point set.Lastly the pending feature point set is precisely detected by the proposed rhombus algorithm.Experimental results show that the non-feature points are massively reduced,the efficiency of detection is obviously promoted and the feature points are distributed in a more uniformed pattern according to this optimized algorithm.3.Image registration.The detected feature points are preliminarily matched through applying the NCC algorithm,and error registration of preliminary registration is precisely detected through applying the RANSAC algorithm as to eliminate the error registration pairs in the preliminary registration stage.Lastly,the parameters of transformational model are acquired.4.Image fusion.The splicing of the image directly after the transformation may result in splicing or ghosting at the splice.To avoid this,the theories of Poisson formula and Poisson fusion algorithm are derived in detail.Through applying these algorithms,the trace of stitching according to the image exposure difference and the double image generated after the image stitching while occurring moving objects can be effectively eliminated.At the same time,this paper proposed a method which can improve the deformation caused by multiple splicing when the images are divided into several groups then are spliced...