Research On Multi-view Images Stitching And Arc Scanning Base On Cloud

Multi view image stitching is to solve the limited of the perspective of imaging equipment,cannot be taken 360 degree field of view image and put forward the theoretical method.Multi view image stitching has been widely used in many fields,such as aviation,remote sensing image analysis,medical image analysis,it has very important practical significance and engineering application perspective.It is also a hot issue in the field of virtual reality.In this paper used single camera arc scanning mode to obtain the images,the technical principle of multi-view image mosaic is studied deeply,and the corresponding solutions are given to the difficulties in multi-view image mosaic technology.Image acquisition,image preprocessing,image registration,image fusion,panoramic images output five key steps in multi view image stitching is analyzed in detail,each step of the experimental results are given the result figures.Finally the multi view image stitching system is designed and implemented based on VS2012 used C++ language.In the process of image acquisition,this paper used the arc scanning method,the camera is fixed on the aerial platform,get a set of multi view images with overlapping regions are obtained by rotating the camera.In image preprocessing,according to the some original images have fog,used dark channel prior method deal with the image dehazing.It effectively solves the problem that the multi view image stitching is ineffective because of the unclear images.In the process of image registration,this paper studied the multi-view image registration problem of the same scene acquired under different acquisition time and acquisition angles,made a deep research and analysis center on image registration based on feature.Firstly,used ORB feature extraction methods obtained image characteristics,then used image matching by characteristic points,finally used improved RANSAC algorithm remove the false matching points complete the image registration.Image fusion firstly solve how to confirm the overlap region of the images,and then used multiresolution fusion method to fuse the images.At last the paper discuss the luminance difference caused by the change of the shooting angle and illumination,and used the intensity adjustment method to eliminate the “seam line”,have good image fusion results.In order to obtain multi view panoramic image,some kinds of projection models are discussed,and used spherical projection to coordinate transformation.Then can get the result of multi view image stitching based on 50-80 images.Finally,by using the virtual reality software,the multi view image is transformed into a virtual reality image.