Research On Large Field-of-view Image Stitching Method Based On Cloud

Image stitching is a kind of technique that can synthesize a number of small field-of-view images with overlapped regions from the same scene into a large field-of-view image.With the development of computer technology and machine vision technology,image stitching technology has been widely used in the field of panoramic surveillance,medical image synthesis,virtual reality and military fields,people's demand for large field-of-view image is also more urgent.However,most of the existing image stitching methods only focus on the expansion of horizontal field-of-view,and there is little research on the horizontal and vertical field-of-view,therefore,it is very meaningful to study the large field-of-view panoramic image stitching.Based on the needs of project and research,a new method of large field-of-view image stitching based on Cloud is proposed in this paper,which main research and implementation the large field-of-view image stitching algorithm and software.Based on the characteristics of the image acquisition and the real-time requirements of project,a research program has been proposed in this paper,the research program using ORB algorithm extract image feature,hamming distance match this feature,improved RANSAC algorithm calculate the image transformation model,spherical projection uniform coordinate transformation,apply brightness adjustment technology and multi-band image blending algorithm to complete the process of large field-of-view image stitching.This paper first introduces the basic process of image stitching,and deeply studies the basic principle of image stitching.Then,through the research and analysis of image acquisition principle,the ORB algorithm is used to detect the key-points.Aiming at the problem that the classical RANSAC algorithm solves the instability of the transformation matrix with less number of iterations,an improved RANSAC algorithm is proposed,without increasing the number of iterations,the improved RANSAC algorithm can optimize the transformation matrix,and improve the accuracy of image registration.Through the research of several panoramic image projection models,the spherical projection is transformed into the coordinate transformation.In view of the fact that there are some differences in the brightness of the real scene image,a brightness adjustment method based on linear transformation is introduced.Based on the research of image blending technology,the multi-band image blending algorithm is used to blend the transformed image.Finally,through the VS2012 and Opencv platform using C++ language,the whole process of large field-of-view image stitching based on Cloud is realized.Through a series of experimental results,it is proved that the effectively and adaptability of the large field-of-view image stitching method based on the Cloud.In addition,by displaying the virtual roaming result for the image obtained by proposed method,the correctness of the method is proved again.